Decision No. 154-AT-A-2022

Summary

[1] Lui Greco filed an application with the Canadian Transportation Agency (Agency) against Air Canada on October 23, 2017, because he was unable to book a ticket on Air Canada’s website while using a screen reader.

[2] Mr. Greco initially sought a reversion of Air Canada’s website to its pre-January 2017 version until the accessibility issues were resolved. Given Air Canada’s repeated assertion that such a reversion is impossible, Mr. Greco now requests that Air Canada be required to submit a quarterly report to him and to the Agency for an indefinite period of time about the accessibility status of its website and that he be allowed to make quarterly written submissions on the accessibility of Air Canada’s website for persons who use screen readers.

[3] On April 5, 2019, the Agency issued Decision CONF-8-2019 (Obstacle Decision), in which it found that Mr. Greco is a person with a disability due to a vision impairment and that he faced an obstacle to his mobility in that his access to Air Canada’s website was unequal to that of persons without a disability.

[4] In this decision, the Agency will address the following issues:

1. Can Air Canada remove the obstacle without experiencing undue hardship?
2. If so, what remedies, if any, should the Agency order?

[5] For the reasons set out below, the Agency finds that the obstacle faced by Mr. Greco was undue given that Air Canada has since been able to remove the obstacle by rendering its website accessible to Mr. Greco. As demonstrated during the oral hearing, persons who use screen readers are now able to independently complete a booking in a similar amount of time as persons who do not use screen readers.

[6] The Agency notes that much has changed since Mr. Greco filed his application. Air Canada has incorporated accessibility into its website development lifecycle and has adopted measures to effectively prevent and address accessibility issues. Furthermore, the Accessible Transportation for Persons with Disabilities Regulations^{Note 1}(ATPDR) have since come into force, obligating Air Canada to ensure that its website meets the requirements for a Level AA conformance with the Web Content Accessibility Guidelines (WCAG) 2.1.^{Note 2} In light of these changes and the work that Air Canada has done to improve its website and the processes around its website, the Agency finds that an order for corrective measures is not necessary at this time.

[7] The Agency acknowledges that both parties have reserved the right to make submissions on costs. As such, the Agency will consider requests for costs if they are submitted in accordance with the Agency’s direction.

Background

[8] Mr. Greco is blind and uses a screen reader. He explains that he had often been required to book tickets using Air Canada’s website as part of his employment and was able to do so independently until September 2016. He claims, however, that Air Canada’s website became unusable to him after it underwent significant revisions in January 2017. He contacted Air Canada about his difficulties with the website in April 2017. In October 2017, when the website was still inaccessible to him, Mr. Greco filed an application with the Agency, citing several issues as examples of problems that he experienced with the website. The Agency opened pleadings on the application in January 2018.

[9] After issuing decisions on various procedural matters, the Agency issued the Obstacle Decision in April 2019. In that decision, the Agency found that Mr. Greco is a person with a disability and that he encountered an obstacle to his mobility in that his access to Air Canada’s website was unequal to that of persons without a disability. The Agency found that Air Canada’s website continued to have issues when used with a screen reader, including some new issues that resulted from fixes to issues identified in Mr. Greco’s 2017 application. The Agency then opened written pleadings on whether and how Air Canada could remove the obstacle without experiencing undue hardship. The Agency also notified the parties that it would convene an oral hearing to supplement the record on how the obstacle can be removed or whether measures to remove it would cause Air Canada undue hardship.

[10] Between May 2019 and February 2020, the Agency received written pleadings on undue hardship and issued decisions on Air Canada’s two requests for confidentiality. The Agency also received submissions on Air Canada’s request to dismiss Mr. Greco’s application in which Air Canada argued that the Agency no longer had jurisdiction over the application because Air Canada had addressed the issues identified in the application. The Agency stayed the proceedings on March 18, 2020, because of the COVID-19 pandemic in Order 2020-A-32. On July 16, 2020, the Agency issued Decision LET-AT-A-52-2020, in which it determined that it had jurisdiction to continue the proceeding and indicated its intention to proceed with an oral hearing.

[11] On November 2, 2020, the Agency provided its direction on the scope of the oral hearing in Decision LET-AT-A-69-2020 (Scope Decision). The Agency noted in the Scope Decision that the issues listed in the Obstacle Decision “were examples of how the obstacle to accessibility were manifested and, as such, were evidence of … the ‘long-standing and evolving nature of the issues raised in this application.’”

[12] Between January and August 2021, both parties filed expert reports concerning the accessibility of Air Canada’s website to persons who use screen readers. Air Canada’s expert report and a responding report were written by Derek Featherstone, Chief Experience Officer at Level Access, an accessibility consulting and software agency.

[13] Mr. Greco’s expert reports were written by the following individuals:

• Lisa Liskovoi, Senior Inclusive Designer and Digital Accessibility Specialist with the Inclusive Designer Research Centre at Ontario College of Art and Design (OCAD) University; and
• Abhishek Gupta, Accessibility Test Lead at Professional Quality Assurance; and
• Jonathan Duncan, an expert in information systems analysis, design and development.

[14] Following several procedural requests and adjournments, an oral hearing was held September 1-3 and October 20-21, 2021. In addition to the expert witnesses listed above, the Agency also heard testimony from Mr. Greco and the following Air Canada witnesses:

• Conroy Allen, Director of Quality Assurance at Air Canada;
• Carlos Faxas, Director of Digital Product at Air Canada;
• Joe DiNero, a screen-reader teacher at Helen Keller Services for the Blind in New York and Digital Accessibility Team Lead and User Tester for UsableNet, a digital accessibility consultancy engaged by Air Canada;
• Michele Lucchini, Vice-President for Delivery and Accessibility Operations at UsableNet; and
• Mark Nasr, Senior Vice-President of Marketing Products and E-commerce and the President of the Aeroplan Frequent Flyer Division at Air Canada.

[15] Pleadings closed on April 25, 2022, following receipt of the parties’ written closing statements.

Preliminary matter

[16] In his closing statement, Mr. Greco submits that, in accordance with section 15 of the Canadian Charter of Rights and Freedoms^{Note 3}(Charter), he has a constitutional right to equal access. He argues that the Agency is a competent court for the purposes of section 24 of the Charter such that the Agency can retain jurisdiction on this case indefinitely for the purpose of an order requiring Air Canada to provide him and the Agency with quarterly reports on the accessibility of its booking website.

[17] However, in light of the fact that these arguments were raised for the first time in his reply and are moot given the Agency’s decision to not impose corrective measures on Air Canada, the Agency will not consider these Charter-related arguments.

Issue 1: Can Air Canada remove the obstacle without experiencing undue hardship?

The law

[18] The application was filed pursuant to subsection 172(1) of the CTA, which at the time of Mr. Greco’s application, stated:

The Agency may, on application, inquire into a matter in relation to which a regulation could be made under subsection 170(1), regardless of whether such a regulation has been made, in order to determine whether there is an undue obstacle to the mobility of persons with disabilities.

[19] At the time that the application was filed, the Agency used a three-part approach to determine whether there was an undue obstacle to the mobility of persons with disabilities. The Agency addressed Parts 1 and 2, disability and obstacle, in the Obstacle Decision. In this decision, the Agency will address Part 3:

If it is determined that the applicant is a person with a disability and that they encountered an obstacle, the Agency provides the respondent with an opportunity to either:

• • explain how it proposes to remove the obstacle through a general modification to the rule, policy, practice, or physical structure or, if a general modification is not feasible, an accommodation measure;

or

• • demonstrate that it cannot remove the obstacle without experiencing undue hardship.

Air Canada’s submissions

[20] Air Canada submits that it has removed the obstacle faced by Mr. Greco by rendering its website accessible to persons who use screen readers. In its view, if the Agency finds that the obstacle remains, it should also find that Air Canada cannot remove the obstacle any further without experiencing undue hardship because Air Canada is doing all that it can reasonably be expected to do to not only remove issues if and as they arise, but also to identify and prevent them.

Accessibility of the website

[21] Air Canada acknowledges that the Agency found in the Obstacle Decision that Mr. Greco faced an obstacle in that his access to the website was unequal to that of persons who do not use screen readers. However, Air Canada argues that its website has evolved since the Obstacle Decision was issued, including by becoming more accessible, and that at the time of the oral hearing, the obstacle was no longer present.

[22] Air Canada maintains that the Agency should be considering whether the obstacle still exists; in other words, the Agency should not be considering whether the obstacle can be removed, but whether it has, in fact, been removed. Air Canada argues that the best evidence in this regard is the live demonstrations presented by Mr. Greco and Mr. DiNero during the oral hearing which established that the issues identified in earlier Agency decisions—such as issues with the payment page, certain sections on the flight selection page, and the timeout of the reservation process after 10 minutes—no longer exist. It further argues that the live demonstrations showed that there is no obstacle because bookings can be completed by persons who use screen readers.

[23] Air Canada concedes that there were five issues that arose during Mr. DiNero’s demonstration of a booking using a screen reader during the oral hearing:

1. when Mr. DiNero left the date of birth blank, the focus came back to that field without reading the field aloud;
2. the radio buttons were not labelled correctly and only read “yes or no” for the Aeroplan selection;
3. flight times were read as metres rather than minutes;
4. when conducting a seat selection, the headings did not read properly; and
5. the seat selection page was not set up properly in a table.

[24] However, Air Canada submits that these issues were minor and did not prevent Mr. DiNero (who is entirely screen-reader dependent when using technology) from completing a booking because there were workarounds or the information required by the user could be understood from the context.

[25] In addition, Air Canada filed two documents to demonstrate its continued commitment to resolving accessibility issues. The first, provided pursuant to an undertaking by Air Canada, is a report titled “Booking Flow Severe Issues – April 2021 Audit” (Issues Resolved Report 1). This report shows a list of issues that had been identified in UsableNet’s 2021 first-quarter audit and that had been resolved between April 7, 2021, and October 6, 2021. The second is a report that contains a list of accessibility issues related to the booking flow that had been resolved between the September and October oral hearing dates (Issues Resolved Report 2).

[26] Air Canada asserts that the term “barrier” (now used in the English version of the CTA in place of the term “obstacle”) reframes accessibility as the user’s ability to participate in society. Air Canada argues that the standard for its website should be “equal” access rather than the same access; that is, that users who are blind or have a visual impairment should be able to use the website without significant issues and have access to the same services but not necessarily have the exact same experience as a sighted person. In other words, the accessibility of a website does not depend on whether there are issues, but rather, on the nature of such issues and their severity. Air Canada argues that demonstrations during the hearing established that a booking can be completed on its website by persons with a visual impairment who use screen readers, so there is equal access.

[27] Air Canada does not dispute that WCAG is an important guideline to help assess and achieve website accessibility. However, it argues that the Agency’s assessment of the accessibility of Air Canada’s website requires a determination of its overall usability by persons who use screen readers rather than simple compliance with WCAG.

[28] Finally, Air Canada argues that the absence of an accessibility statement on its website does not, in and of itself, make its website inaccessible. Its expert witness, Mr. Featherstone, expressed his opinion that accessibility statements are not a requirement of WCAG and points out that Mr. Greco’s witness, Ms. Liskovoi, also stated that, in her view, the lack of an accessibility statement does not make a website inaccessible.

Undue hardship

[29] Air Canada states that, if the Agency determines its website is still not accessible, the Agency must then determine whether Air Canada can remove the obstacle without experiencing undue hardship. Air Canada submits that the question of undue hardship should be framed as “whether Air Canada is doing all that it can reasonably be expected to do, currently and moving forward, to ensure that obstacles to the accessibility of its website are removed if and as they arise”. Air Canada argues that there is nothing further it can do to ensure that the obstacle no longer exists because it is already doing everything it possibly can, as expanded upon below.

[30] Air Canada points out that the Scope Decision speaks to what evidence Air Canada might submit on whether the obstacle can be removed without undue hardship, including “evidence on the procedures and safeguards, including testing, that Air Canada has in place to address the accessibility of its website…”. Air Canada submits that it has a “robust, best-in-class system in place” to not only remove issues if and as they arise, but to identify and prevent them. It provided, in its written submissions and through its witnesses at the hearing, detailed information about the testing it performs and the safeguards it has in place to identify and address accessibility issues. Air Canada also argues that the goal of removing and preventing issues should be measured by “reference to reasonable diligence” and argues that it has demonstrated it is exerting this reasonable diligence with regard to the accessibility of its website.

[31] Air Canada submits that digital accessibility is a “journey”, not an end goal, that involves an ongoing commitment of time and resources that will continue to evolve along with its product, a position echoed by several of its witnesses. Air Canada submits that it is deeply engaged in this journey, as evidenced by the measures discussed in the following paragraphs.

Contract with UsableNet

[32] As part of its efforts to identify and address accessibility issues on its external-facing website, Air Canada engaged the services of UsableNet, a digital accessibility consultancy, in 2018, to provide accessibility-related expertise and feedback. UsableNet has provided and continues to provide several services to Air Canada to assist with renovation and governance, such as remediation, quarterly auditing, testing, training, and an accessibility blueprint, which recommended the creation of a Centre of Excellence.

Centre of Excellence

[33] Following UsableNet’s recommendation, Air Canada created a Centre of Excellence in 2021 which, it suggests, demonstrates a commitment to accessible website design. The Centre of Excellence is the “control centre” for accessibility within the organization that dictates and monitors the way teams within Air Canada are working toward accessibility. According to Air Canada, the Centre of Excellence’s goal is not only to look at fixing the existing problems, but also to stay abreast of best practices as accessible website design and development and regulations evolve. Its activities include drafting a policy statement, identifying specific areas that need to be addressed, and creating a roadmap for future activities. Although still in its formative stages at the time of the oral hearing, the Centre of Excellence began meeting weekly in October 2021. The Centre of Excellence is comprised of representatives from each Air Canada department involved with Air Canada’s digital properties, including Development, Quality Assurance, Project Management, Marketing, Legal, and Communications, in addition to UsableNet consultants, some of whom have a visual impairment. All members of the Centre of Excellence have received screen-reader training.

Training

[34] Air Canada asserts that all employees involved in the website development lifecycle receive training developed by UsableNet, including designers, developers, quality assurance employees and testers. For instance, designers learn how to make sure they are designing accessible web content, developers learn how to ensure the accessibility of their own coding, and the quality assurance team receives training on how to identify and prioritize accessibility issues using various accommodation tools such as screen readers.

[35] The training is broken down into a number of sessions, totalling approximately 12-18 hours of instruction, which includes a three-hour in-depth session on WCAG and then separate sessions on techniques and syntax. Training is also provided through workshops in which UsableNet team members work alongside with Air Canada team members to resolve specific issues. For example, one workshop focused on a drop-down menu on the home page, but the goal was to provide Air Canada employees with the knowledge to fix all the drop-down menus across their website.

[36] Air Canada expects to continue to create training, including a regular certification process, to make sure its employees remain up to date, particularly as accessibility evolves.

Website development lifecycle

[37] The website development lifecycle is the process by which Air Canada develops its website. The website development lifecycle consists of nine-week development periods called product increments, which are broken down into three-week development periods called sprints. Sprint teams are comprised of the product owner, the development team and the quality assurance team, and includes people trained in accessibility. Air Canada submits that members of its product development teams are knowledgeable about accessibility, having received accessibility training and encountering or discussing accessibility in their jobs on a weekly basis or, for many, on a daily basis. Air Canada also asserts that, since engaging UsableNet, it has integrated accessibility into every stage of the website development lifecycle, including design review, development, quality assurance, integrated testing, production, user testing, and issue handling.

Compliance with WCAG

[38] Mr. Lucchini testified that UsableNet currently tests Air Canada’s website against WCAG 2.1 when it completes audits. However, several witnesses, including Mr. Greco’s witness, Ms. Liskovoi, testified that 100% compliance with WCAG is not possible for Air Canada’s website due to its complexity. Air Canada argues that being held to a standard requiring perfect compliance with the binary standard set out in WCAG would cause it undue hardship.

[39] Air Canada argues that its website is becoming increasingly complex in part to compete with other airlines and to adapt to various government requirements. It asserts that agile website development must be responsive to change and thus, issues will likely always emerge in a complex website. Air Canada’s witnesses therefore acknowledge that its website is not perfectly accessible, but submit that it has reached a level of “substantial accessibility”, meaning that users can make a booking even if there are minor WCAG infractions.

Expert report by Derek Featherstone, Level Access

[40] Air Canada submits that the only fair and rational comparison of Air Canada’s website is to that of its competitors. Mr. Featherstone pointed out that issues arise on all websites, including those belonging to competitors with much larger budgets for their digital properties. With this in mind, Air Canada hired Level Access in late 2020, to do a comparative analysis of Air Canada and the booking flows of three other airlines, including two American competitors and one Canadian competitor. Air Canada submitted the resulting report created by Level Access (December 2020 report) as evidence demonstrating that its website performed well in this comparative analysis, especially taking into consideration that the American carriers have been subject to regulatory requirements on accessibility since December 2015, compared to December 2020 for Canadian airlines.

Parameters of the study

[41] The comparative analysis performed by Level Access assessed the functionality of Air Canada’s website in December 2020. Functional use testing employs users with disabilities to identify issues that are not addressed by WCAG.

[42] Mr. Featherstone’s position is that comparative analysis is a useful way to assess the accessibility of a certain type of technology—in this case, a booking website—while controlling for the complexity of that type of technology. Level Access did not do automated or manual testing of Air Canada’s website, nor did it provide to Air Canada any technical reports or analysis behind the issues.

[43] Level Access used an experienced accessibility analyst, a facilitator (a user experience practitioner who facilitates studies) and a user tester who was blind. The testing was performed using Windows 10, JAWS 2020 screen reader, and a Chrome web browser.

Components of an airline’s booking website

[44] Mr. Featherstone sets out that most airline websites have the same general paths of tasks and subtasks, which include:

1. locating the booking component within the home page of the website;
2. specifying departure and arrival airports;
3. selecting a trip type (one-way, round-trip, multi-city);
4. selecting a fare type (economy, business, etc.);
5. specifying the number of passengers;
6. specifying dates for travel;
7. reviewing and selecting flights;
8. verifying the trip summary;
9. entering passenger details;
10. selecting travel options;
11. seat selection;
12. providing payment information; and
13. reviewing a confirmation of the purchase.

[45] Level Access rated each of the first 12 components on the websites of all four airlines on a 1 to 4 scale while the user tester booked a flight from Toronto to San Francisco, a complex route chosen for the study because it brings up many flight options, includes layovers and is more likely to bring up issues for the user tester. A rating of 1 indicated that a user tester could easily complete the task. A rating of 2 meant that the user tester required some clarification after expressing confusion. A rating of 3 meant the facilitator had to intervene to complete the task and a rating of 4 meant that the user tester could not complete the task without the facilitator taking control.

Results

[46] Level Access rated Air Canada with a score of 1 for locating the booking component, selecting trip type, specifying departure and arrival flights, trip summary and confirmation after payment.

[47] Level Access rated Air Canada with a score of 2 for selecting the fare type, specifying the number of passengers, selecting additional travel options and seat selection.

[48] Level Access rated Air Canada with a score of 3 on the most complex tasks and interfaces: selecting the date range, reviewing selected flights, entering passenger details and entering payment information. The facilitator needed to coach the user tester to complete these tasks. Mr. Featherstone elaborates with examples from Air Canada’s website as set out below.

• While using the date picker, the user tester expected to be able to type the date in, and this was not possible. Additionally, the focus would often go outside the date picker and it was difficult to get back into the date picker while using the screen reader.
• On the flight selection screen, there was a lot of information to process, and the facilitator stepped in to help the user tester choose a flight.
• The user tester was often not sure what selection they were making while entering gender or birth date in the passenger information component.
• Finally, while entering payment information, the user tester was uncertain what option they had selected based on the fields that were being read out. For instance, they thought they had chosen to pay with a credit card, but then the fields expected to pop up for paying with a credit card did not appear.

[49] Level Access did not find that there were any tasks which were completely blocked for persons who use screen readers on Air Canada’s website, so Air Canada did not get any scores of 4, unlike its major Canadian competitor.

[50] Air Canada obtained an average score of 1.92, which it points out was just slightly higher than its American competitors at averages of 1.73 and 1.75 each, and significantly lower than its Canadian competitor which scored an average score of 2.42.

Mr. Greco’s response to Air Canada’s evidence

[51] Mr. Greco argues that Mr. Lucchini’s testimony that all the issues identified in UsableNet’s April 2021 audit of Air Canada’s website had been resolved is anecdotal and uncorroborated. Although Air Canada referred during the hearing to a UsableNet audit performed in April 2021, it did not submit that audit to the Agency, nor did it provide any audit or evaluation evidence to the Agency beyond Level Access’s December 2020 comparative user-testing report.

[52] In his closing statement, Mr. Greco states that, even if Air Canada had established that its website was both technically compliant and usable at the time of the oral hearing, Air Canada also demonstrated that its website is constantly changing. Therefore, he argues, future changes could undo the accessibility of the website.

Mr. Greco’s submissions

[53] Mr. Greco disputes Air Canada’s argument that he bears any burden for demonstrating that the obstacle persists. In his view, the Obstacle Decision and Decision LET-AT-A-50-2021 require Air Canada to demonstrate that its website is fully accessible. Mr. Greco submits that Air Canada did not discharge this burden and that making the website accessible would not cause Air Canada undue hardship.

Accessibility of the website

[54] Mr. Greco acknowledges that there have been significant improvements to Air Canada’s website since his initial application; however, he argues that the inequality of access cited in the Obstacle Decision remains. In support of his submission, he submitted three video files of him using the website, presented three experts who submitted reports and testified at the oral hearing, and provided a demonstration at the oral hearing.

Video submissions of Mr. Greco using Air Canada’s website

September 17, 2020, 5:45 pm

[55] In the first video submitted by Mr. Greco, he demonstrated the use of Air Canada’s website together with the most recent version of JAWS on a Windows 10 computer. Mr. Greco did not appear on screen, but his voice was heard throughout this video providing commentary. Mr. Greco’s screen reader was not heard on the video. Mr. Greco submitted that he encountered the issues outlined below while trying to book a round-trip from Vancouver, British Columbia, to Halifax, Nova Scotia, for two adults and one child:

• he was given incorrect instructions to make selections for the number of passengers;
• he was given incorrect information about the number of passengers selected; and
• the controls were not consistent in that certain selections could be made using arrows, but others required the tab key.

September 17, 2020, 6:05 pm

[56] This video showed Mr. Greco navigating to Air Canada’s website and selecting a return trip from Halifax to Vancouver for one adult departing November 14, 2020, and returning November 29, 2020. Mr. Greco did not provide audio commentary for this video submission and he did not identify the issues he may have encountered. His screen reader was also not heard in the video.

January 10, 2021

[57] In this video, Mr. Greco allowed the audio from his screen reader to play aloud during the video. However, he did not provide any audio commentary identifying the issues he may have encountered.

Expert report by Lisa Liskovoi, Inclusive Design Research Center

Parameters of the study

[58] Ms. Liskovoi evaluated whether all tasks could be completed with a keyboard while using the booking flow of Air Canada’s website and whether all interactions were accurately and clearly communicated to the user tester. Ms. Liskovoi specifically looked at the search, flight selection and entry of passenger details portions of Air Canada’s booking website. Ms. Liskovoi completed the study on February 11, 2021, using Windows 10 Pro with a NVDA 2020.4 screen reader and a Chrome 88 web browser on a Lenovo ThinkPad, as well as macOS Mojave 10.14.6 with VoiceOver screen reader and a Safari web browser on an Apple MacBook Pro.

Results

[59] Ms. Liskovoi discovered four main categories of accessibility issues that arose during her testing.

Incorrect instructions provided

[60] Ms. Liskovoi sets out that in several instances, instructions provided to the user tester did not accurately communicate what a user tester needed to do to successfully finish a task. For instance, while using VoiceOver on a Mac, the instructions stated that the arrow keys should be used to make a selection, but the input did not respond to arrow keys. While using NVDA on a PC computer, Ms. Liskovoi submits that the user tester was given conflicting instructions: one instruction was to use the arrow keys whereas another was to use the tab key to make a selection.

Elements behave unpredictably

[61] Ms. Liskovoi submits that Air Canada used non-standard controls that did not function in a predictable way. For instance, Ms. Liskovoi suggests that Air Canada’s date picker could have been more inclusive by allowing a user tester to type the date.

Incorrect feedback provided about values within inputs

[62] Ms. Liskovoi submits that the user tester was not given accurate feedback when entering input into a form. For instance, while using VoiceOver on a Mac, if a user tester made a selection for the Origin field and moves to the Destination field, the screen reader told the user tester that the Origin value entered was also in the Destination field, even though the Destination field was actually blank. A similar issue arose while using NVDA on a PC.

Information read in the incorrect order

[63] Ms. Liskovoi submits that the reading order of content is crucial because when an element is out of order, it can impact the meaning of the content. For example, when comparing Basic and Standard fares, the information presented visually in columns was read line by line, across categories, instead of column by column.

[64] Ms. Liskovoi also notes that Air Canada’s website does not have an accessibility statement. She argues that websites should communicate to users through an accessibility statement how their products are not accessible and the steps the organization is taking to achieve accessibility.

Expert report by Abhishek Gupta, Professional Quality Assurance

Parameters of the study

[65] Mr. Gupta submits that he evaluated the Air Canada booking website using NVDA with a desktop computer, VoiceOver on an iOS mobile device, and TalkBack on an Android mobile device. Findings were compared against WCAG 2.0 AA standards to identify issues that would prevent persons who use screen readers from completing a booking on Air Canada’s website. Mr. Gupta’s team evaluated Air Canada’s website over a month-long period from December 17, 2020, to January 17, 2021, with three testers and a team lead.

Results

[66] Mr. Gupta’s report was accompanied by a spreadsheet that detailed the issues found on Air Canada’s website and their associated WCAG success criteria. In total, Mr. Gupta identified 181 issues. In his report, Mr. Gupta grouped the issues with Air Canada’s booking website that he submits would affect persons who use screen readers into the categories listed below.

Difficulty accessing calendar pop-ups using the keyboard

[67] Primarily while using VoiceOver, the mouse needed to be used to focus inside the calendar widget to select the travel dates. Mr. Gupta offered his opinion that persons who use screen readers should be able to access the calendar widget while using a keyboard. The screen reader also read aloud irrelevant information and incorrect dates.

VoiceOver screen reader read all results at once

[68] Mr. Gupta argues that each field should be read in a logical order, and that there should be a pause after each field has been spoken rather than being read all at once.

Pop-up windows were read in the wrong sequence

[69] Mr. Gupta points out that when a modal window opened comparing Economy Standard and Economy Comfort, the information was divided into two side-by-side boxes. The information should have been read from top-to-bottom within each column; instead, the information was read horizontally, as if the information were presented in rows.

Dollar values were not accessible by a screen reader on multiple webpages

[70] Mr. Gupta submits that the dollar values were not read aloud on the Your Trip Summary, Travel Options, Flight Confirmation, and Review and Pay pages. He argues that persons who use screen readers would not be able to tell how much a flight costs.

Error messages were not read aloud on the Passenger Information webpage

[71] Mr. Gupta submits that on the Passenger Information webpage, if a user tester did not populate a mandatory field, the screen reader did not read the error message, so the user tester did not know which field was left blank.

[72] Finally, Mr. Gupta argues that Air Canada’s website should contain an accessibility statement that outlines its efforts to be compliant with WCAG 2.0 Level AA Success Criteria, lists open issues and identifies compatible screen readers.

Expert report by Jonathan Duncan, Professional Quality Assurance

[73] Mr. Duncan used Mr. Gupta and Ms. Liskovoi’s reports, combined with his 25 years of experience in software development, to provide his professional opinion as to how and why the issues arose, as well as a potential way to remedy the issues. Mr. Duncan qualifies his report by pointing out that he could not do a fulsome root cause analysis because he did not review the initial design and technical details of Air Canada’s website, including the architecture or code.

[74] Mr. Duncan suggests that the following scenarios may have contributed to the issues found on Air Canada’s website and provides some possible solutions to remedy the issues.

Design

[75] Mr. Duncan suggests that designers often make choices to leave some customers out to attract more customers in general. Groups that may require specialized attention are often forgotten during the initial design phase as the business focuses on their primary goal of bringing customers to their websites.

Technology

[76] Mr. Duncan suggests that companies often make technology choices without understanding the requirements for the software. For instance, Mr. Duncan hypothesizes, based on looking at Air Canada’s website, that the technology underlying the user interface of Air Canada’s website is Angular. He points out that Angular has particular pitfalls with regards to accessibility, making it a poor technology choice if accessibility is one of the requirements. Fixing accessibility issues on a website using Angular would require some redesign and correcting issues in a “more surgical fashion”. Alternatively, Air Canada could completely redesign its site and change technologies, but that would be much less cost-effective.

Mr. Greco’s demonstration during the oral hearing

[77] During this demonstration, Mr. Greco used JAWS, what he describes as a “normal work laptop” and Chrome as the web browser. Mr. Greco indicated that he would book a multi-city trip from Saint John, New Brunswick; to Winnipeg, Manitoba; to Vancouver, British Columbia; and then back to Saint John. He submitted that he would intentionally create errors to demonstrate what a screen reader is told to do when a user makes a mistake.

[78] He was able to select two passengers over the age of 16 years and select the origin city as Saint John. He then intentionally entered the wrong city code for Winnipeg but was not immediately told by his screen reader that an error occurred. He was able to fix the error based on the information provided when he went into “review mode” on his screen reader, but he submitted that he only knew to go to “review mode” to receive the error message because he knew that he had entered incorrect information. At this point during the demonstration, he stated that “functionality [was] 90 percent”.

[79] Mr. Greco then proceeded to the date picker to select the date of his departure. He did so by entering the dates with a keyboard in the order of day, month and year, which resulted in him entering that he would be leaving Saint John on November 10, 2021, and returning October 11, 2021. He was told by his screen reader that this was not possible, but asserted that there should have been a message that identified the order in which he put the date as the error on that page.

[80] He then discovered that he entered a return date with digits where he should have entered the next departure city. The error read to him indicated that the origin city could not be found with the digits he had entered in the field.

[81] Mr. Greco began to fix the origin city of the next leg of the journey by entering V-A-N for Vancouver and then pressing Enter, but he encountered an error. He submitted that he thought that he had selected Saint John to Winnipeg to Vancouver. He realized that he needed to enter the second flight as Winnipeg to Vancouver, with the second leg leaving on October 14.

[82] Mr. Greco then proceeded to the flight selection process. Using “list navigation”, he chose a flight from Saint John to Winnipeg that has a layover in Montréal, Quebec, for his guide dog.

[83] He submitted that he wanted to skip the seat selection, but was forced into a modal window to preview seats. He was unable to find a way to exit the seat selection modal window and ended his demonstration.

Functional accessibility versus technical compliance

[84] Mr. Greco states that Air Canada’s expert, Mr. Featherstone, testified that website accessibility has two components: (1) functional accessibility, which refers to ease of use or usability and (2) technical compliance with WCAG. Mr. Greco concedes that Air Canada demonstrated the functional accessibility of its website at the oral hearing but argues that is not enough. He submits that, if Air Canada’s website was both functionally accessible andtechnically compliant, Air Canada would have removed the obstacle to the point of undue hardship, but Air Canada has not yet demonstrated technical compliance.

[85] Mr. Greco submits that website accessibility can be demonstrated by conformance with WCAG. As evidence, he submits testimony from his expert, Ms. Liskovoi, who stated that, although a website that is compliant with WCAG may still present barriers to users, a website that is WCAG-non-compliant is much more likely to pose barriers to users. Mr. Greco concedes that while Ms. Liskovoi stated 100% compliance with WCAG is unlikely, he disagrees with Air Canada’s assertion that it cannot be held accountable to WCAG guidelines because he argues WCAG guidelines are the best measure of website accessibility.

[86] Mr. Greco submits in his closing statement that Air Canada did not file any data about technical compliance despite the fact that it possesses this data. In support of this assertion, Mr. Greco points out that Mr. Allen testified that UsableNet has audited compliance of Air Canada’s website with WCAG 2.0AA and 2.1AA. Mr. Greco asserts that without technical compliance data, the Agency has no means to assess how accessible Air Canada’s website actually is. He states that this evidence would have permitted the Agency to weigh usability versus technical compliance, but as it is, the Agency is without the evidence necessary to determine whether Air Canada’s website is accessible.

Undue hardship

[87] Mr. Greco submits that the bar of undue hardship is high. Carriers can only be relieved of the responsibility of accommodating passengers with a disability if removing the obstacle would be “impossible, impracticable or unreasonable”.

[88] Mr. Greco argues that Air Canada has not provided any evidence to suggest that it would encounter undue hardship if it were to make its website compliant with WCAG.

[89] Mr. Greco further states that Air Canada has not posted an accessibility statement in the last four years, despite it being an accepted best practice. He argues that Air Canada does not provide any evidence to suggest that having an accessibility statement would cause it undue hardship. Mr. Greco recognizes that an accessibility statement is not required by the WCAG; however, he argues that the expert evidence before the Agency is that the current state of accessibility of a website should be communicated to users in some way and that accessibility statements are an established practice for doing so.

[90] Finally, Mr. Greco calls into question Air Canada’s commitment to accessibility by pointing out that its Centre of Excellence has no members who have a vision impairment other than those employed by its contracted consultant, UsableNet.

Air Canada’s response to Mr. Greco’s evidence

Evidence submitted prior to the oral hearing

[91] Air Canada argues that the evidence submitted by Mr. Greco prior to the oral hearing is “stale-dated” because the website has evolved since the evidence was submitted. It also argues that Mr. Greco’s submissions prior to the oral hearing were untested because Air Canada could not ask him questions about his techniques and problem-solving as it was able to do at the oral hearing. Even so, Air Canada submits that the problems that Mr. Greco cited in his application have been fixed along with accessibility errors identified through UsableNet’s audit reports, such as those set out in its answer of August 2019.

Mr. Greco’s demonstration at the oral hearing

[92] Air Canada asserts that Mr. Greco’s demonstration of a booking with a screen reader during the oral hearing was not a good example of how accessible Air Canada’s booking website is to a typical person who uses a screen reader because he:

• failed to listen to spoken instructions or information, such as when entering the number of passengers, how to enter the date, how to exit from the seat selection overlay, or when layover times are listed;
• made faulty assumptions unrelated to accessibility, such as not having to enter a start city for the second leg of a multi-city journey, or that the layover time was not clearly stated when it, in fact, was;
• made non-standard selections such as a multi-city booking, which Air Canada submits accounts for only 2.4% of its bookings;
• looked for or intentionally created errors, an exercise that goes beyond accessing the website to book a ticket and more closely resembles quality assurance testing;
• did not methodically, exhaustively, or consistently use standard troubleshooting techniques to resolve the issues he encountered, such as failing to “go far enough” to hear a predictably placed error message about an incorrectly entered origin city or not continuing to use the same button navigation he began with throughout the entire page in order to find what he was looking for; and
• was reluctant to acknowledge when he made honest mistakes—as all users, including some of Air Canada’s witnesses working in the field, do—such as when he accidentally pressed the Windows key on his keyboard and the Windows Start menu appeared on the screen, which made him leave Air Canada’s website.

[93] Air Canada submits that the accessibility issues that Mr. Greco encountered were minor and did not block his ability to book a flight, as confirmed by Mr. DiNero and later acknowledged by Mr. Greco. Air Canada adds that not all issues Mr. Greco encountered were related to the accessibility of the website, but rather, related to faulty assumptions or mistakes he made. Finally, Air Canada points out that Mr. Greco’s performance on the website improved over time.

Responding report by Derek Featherstone, Level Access

[94] Mr. Featherstone prepared a responding report that reviews the reports authored by Abhishek Gupta, Jonathan Duncan, and Lisa Liskovoi, and filed by Mr. Greco in support of his case. As set out in Mr. Featherstone’s responding report, he examined the testing and reporting methodologies used by all three of Mr. Greco’s experts and submits that there were many inconsistencies and reported issues that were not actually accessibility issues with Air Canada’s website.

Expert report by Abhishek Gupta, Professional Quality Assurance

[95] Mr. Featherstone summarizes his assessment of Mr. Gupta’s report by asserting that the testing and report was completed by those who are inexperienced in accessible technology. Mr. Featherstone asserts that the flaws in Mr. Gupta’s report call into question the reliability and accuracy of the report. He notes that the accessibility errors in the Excel spreadsheet itself that Mr. Gupta attached to his report also demonstrate that Mr. Gupta does not have a good understanding of website accessibility. Some of the issues Mr. Featherstone noted with Mr. Gupta’s report are listed below.

Duplicate issues

[96] Mr. Featherstone submits that while Mr. Gupta’s report contained 181 accessibility errors, a more nuanced reading of the information in Mr. Gupta’s report would narrow that number down to 26 unique and identifiable errors. Mr. Featherstone explains that he suspects that part of the reason there are only 26 unique errors is that manual testing and automatic testing can result in the same errors, but these should only be counted once. Additionally, if there was an issue identified with VoiceOver testing on iOS that is also identified as part of TalkBack testing on Android, one would expect those issues to be counted as one accessibility issue with the website. While there is a column for duplicates in Mr. Gupta’s spreadsheet, as would be expected in an accessibility testing report, Mr. Gupta did not flag any duplicate issues in the 147 issues found with manual testing despite various methods being used.

Performance-related issues

[97] Mr. Gupta identified seven instances of performance-related issues, such as the website not loading properly or quickly enough. For example, if the page did not load properly, it may have seemed like an accessibility-related issue to persons who use screen readers because information was missing; however, Mr. Featherstone asserts this is not an accessibility issue because this issue affects everyone.

Inexperience with using screen readers

[98] According to Mr. Featherstone, Mr. Gupta identified some issues in his report that demonstrate a misinterpretation of how a screen reader normally behaves. For example, reading out sentences in “chunks”, pronouncing the Canadian dollar sign, written as CA $, as “CA dollar sign”, or reading out cursor placement are not accessibility-related errors with the website because this is behaviour one would expect of a screen reader and would not be confusing to a typical person who uses a screen reader.

[99] Mr. Featherstone further explains that when one uses a screen reader with a touch screen, the user must apply the correct amount of pressure and speed to navigate through a webpage while using a touch screen. If one presses too hard, the device may interpret that as a desire to move elsewhere on the page, rather than to keep reading from the previous location. Mr. Gupta’s report acknowledged that one tester may have used too much pressure, sending them elsewhere on the screen. Experienced testers should be aware that these user errors exist—even experienced users sometimes press too hard—and not record them as accessibility issues.

[100] Finally, Mr. Featherstone notes that Mr. Gupta submitted that the screen reader did not properly focus on the prices in a given screenshot. Mr. Featherstone asserts that this is normal screen-reader behaviour. While one should be able to tab from one input field to the next when interaction with the website is required, information that is being provided in text form should not be “focusable” by a screen reader. In other words, persons using screen readers should have to use the arrow keys, rather than the tab key, to get to that information.

Miscategorizing accessibility issues

[101] According to Mr. Featherstone, Mr. Gupta’s team miscategorized 49 of the reported 181 accessibility issues. In order to provide meaningful information to a client, accessibility testers must be able to point to which WCAG success criterion the error corresponds to. Miscategorizing errors, according to Mr. Featherstone, indicates that the testers are not experienced.

[102] To illustrate, Mr. Featherstone points out that one of both Mr. Gupta’s and Ms. Liskovoi’s key findings was that there was an area on Air Canada’s website where information was organized into columns, but the screen reader read it in rows, rather than column by column, which resulted in the user not being able to understand the information. That particular example was a legitimate failure of the WCAG success criterion, 1.3.2 Meaningful Sequence. Mr. Featherstone went on to explain that in his experience, failures of 1.3.2 are relatively rare, and account for less than 0.01% of the accessibility errors Level Access has identified over the past 10 years of testing websites and applications. In Mr. Gupta’s report, however, the Meaningful Sequence criterion was used to categorize approximately 31.5% of the 181 issues found, indicating that this application of 1.3.2 was likely improperly used as a default categorization any time the screen reader output did not make sense to the testers.

[103] The opposite happened with WCAG success criterion 4.1.2, where Mr. Gupta’s team only identified 2% of the errors as 4.1.2; whereas the average use of 4.1.2 among Mr. Featherstone’s team over the last 10 years was 29%.

Calendar pop-ups

[104] Mr. Featherstone points out that while Mr. Gupta claimed the mouse needed to be used to focus inside the calendar widget, Ms. Liskovoi submitted the widget worked with command and the arrows. Mr. Gupta also stated that the screen readers read aloud irrelevant information, but Mr. Featherstone notes that this information was not noted in any of the issues identified by the screen-reader testers, so it is not clear from where Mr. Gupta got this information.

Error messages not read aloud

[105] Mr. Gupta asserted that error messages were not read aloud properly on the Passenger Information Page only, whereas Mr. Featherstone argues that there were error messages on that page. He hypothesizes that Mr. Gupta’s team may have missed error messages read out inline as many persons who use screen readers do if they go too quickly.

No accessibility statement

[106] Finally, Mr. Gupta’s report stated that there is no accessibility statement on the website. Mr. Featherstone points out that this is not a requirement of WCAG and expresses his opinion that this does not mean that the website is inaccessible.

Expert report by Lisa Liskovoi, Inclusive Design Research Center

[107] Mr. Featherstone concedes that Ms. Liskovoi’s report showed a higher level of understanding of accessibility issues than Mr. Gupta’s report; however, he asserts that the main issue with Ms. Liskovoi’s report is that it did not include detailed findings and only included a summary. Ms. Liskovoi did not set out how many accessibility issues occurred, how frequently each accessibility issue she found occurred or how severe the issues were. There are not enough details in the report to be able to reproduce the accessibility issue or to determine what to do next.

Mr. Featherstone notes that Ms. Liskovoi’s report contained four key findings of errors that he points out are common with a website this complex belonging to an organization as large as Air Canada:

1. customers were given incorrect instructions for operating components;
2. elements behaved unexpectedly;
3. incorrect information was given about values within inputs; and
4. markup did not appear within the correct reading order.

[108] In contrast, Mr. Featherstone states that Ms. Liskovoi did not identify some issues that are also often found on large and complex websites, such as keyboard access to the booking flow, unlabelled form fields, text equivalents for images and overall semantic markup. In other words, these errors were not present on Air Canada’s website. Mr. Featherstone asserts that this means Air Canada is progressing on its accessibility journey, even since Level Access reviewed the website for Mr. Featherstone’s December 2020 report.

[109] Mr. Featherstone points out that both Ms. Liskovoi and Mr. Gupta noted that they tested the date picker with a keyboard rather than with a screen reader. Mr. Featherstone points out that this would not be reflective of the experience of persons who use screen readers because, for example, Mr. Greco typed in the date rather than using the date picker during his demonstration. Mr. Featherstone argues that an important piece of accessible design is having multiple ways for a user to provide information that is requested. Mr. Featherstone submits that being able to enter the date manually, especially where the format for the date is provided by the website and read aloud by the screen reader, is an acceptable accessible format of data entry.

Expert report by Jonathan Duncan, Professional Quality Assurance, “Plausible Root Cause Analysis”

[110] Mr. Featherstone posits that Mr. Duncan’s report about the root causes for the inaccessibility of Air Canada’s website was largely conjecture and speculation. Mr. Featherstone points out that Mr. Duncan conceded that he did not have any knowledge of Air Canada’s practices with regards to design, development or testing, nor did he have some of the “bare minimum” information that would be required for a root cause analysis.

[111] Mr. Featherstone also argues that Mr. Duncan’s report was unreliable because it was based solely on Mr. Gupta’s and Ms. Liskovoi’s reports. Mr. Featherstone asserts that the first problem with this foundational basis is that it is not clear whether Mr. Gupta and Ms. Liskovoi were given the same mandate, which would have made it difficult to draw conclusions combining the two reports. Mr. Featherstone also suggests that Mr. Duncan did not verify that any of the information provided by Mr. Gupta was correct because Mr. Duncan did not mention any of the errors and inconsistencies Mr. Featherstone identified in his report. Mr. Featherstone believes that had Mr. Duncan been presented with a report from Mr. Gupta containing 26 errors—the number of errors Mr. Featherstone submits are verifiable and unique—rather than the 181 errors Mr. Gupta reported, Mr. Duncan would have come to a different conclusion.

[112] Finally, Mr. Featherstone asserts that while Mr. Duncan has 25 years of experience in website development, he did not list any experience with accessibility or WCAG on his resume. Mr. Featherstone thus remains uncertain as to what experience Mr. Duncan is relying on to provide assertions about the accessibility of Air Canada’s website.

Analysis

[113] The Agency found in the Obstacle Decision that Mr. Greco faced an obstacle to his mobility due to unequal access to Air Canada’s website. At this point of the proceeding, the burden lies with Air Canada to demonstrate either how it can remove the obstacle of unequal access to its website or that it cannot do so without experiencing undue hardship. In the Scope Decision, the Agency clarified that, in light of the ongoing and evolving nature of the obstacle and the fact that the accessibility issues identified throughout the proceeding are manifestations of the obstacle rather than the obstacle itself, the accessibility of Air Canada’s website at the time of the oral hearing would be the most relevant for the Agency’s analysis. The Agency accordingly defines the removal of the obstacle as whether persons who use screen readers have equal access to Air Canada’s website as those who do not use screen readers at the time of the oral hearing.

[114] As expanded upon below, the Agency finds that the evidence presented during the oral hearing and in the expert reports demonstrate that persons using screen readers are now able to independently complete a booking in a similar amount of time as persons who do not use screen readers. The Agency finds that the question of whether Air Canada is technically compliant is not dispositive of the issue of whether Mr. Greco has equal access to Air Canada’s website. In light of the technological complexities of a booking website, the Agency accepts that the presence of minor issues for persons who use screen readers or the lack of an accessibility statement are not sufficient to establish unequal access.

Demonstrations performed during the oral hearing

[115] The demonstrations performed during the oral hearing showed that previously identified accessibility issues have been resolved and while Air Canada’s website still had some minor accessibility issues, none would prevent persons who use screen readers from completing a booking independently. The information on the website could be understood with surrounding context and tasks could be completed with some workarounds.

Mr. Greco’s booking demonstrations

[116] Mr. Greco attempted three bookings throughout the course of the hearing. Mr. Greco’s first booking demonstration occurred during his examination-in-chief and a second occurred during cross-examination, where Mr. Greco repeated his first demonstration while being questioned by Air Canada. His third booking was completed between the second and third hearing dates but entered into the record through testimony.

[117] During his first demonstration, Mr. Greco was able to type the date into the input value field without having to use the date picker, which had been identified as an accessibility issue in Mr. Featherstone’s December 2020 report. Although Mr. Greco entered the date in the wrong format, the Agency notes that Mr. Greco moved quickly through some instructions provided through his screen reader. The Agency recognizes that persons who use screen readers must, in the interest of time, occasionally skip predictable cues, but this does not mean the instruction is missing. Further, taking into consideration Mr. Greco’s submission that Air Canada’s website should have error messages informing the user of the mistakes made, the Agency notes that an error message informed Mr. Greco that the second date he entered was not compatible with the first and Mr. Greco was able to correct his mistake.

[118] Mr. Greco also encountered problems by incorrectly assuming that an origin city was not needed for the second leg of a multi-city. This issue was due to Mr. Greco’s own incorrect assumption and was unrelated to the accessibility of Air Canada’s website. Mr. Greco was eventually able to deduce and remedy this issue on his own.

[119] With respect to city selection, Mr. Greco entered the letters V-A-N in the origin city field for the second flight and selected the first option, erroneously assuming that he had selected Vancouver. The Agency accepts Mr. Greco’s submission that persons using screen readers will make mistakes entering information, as any user would, but all users bear some responsibility for confirming that they have entered correct information. The Agency expects that persons using screen readers would do so by listening to the information being relayed. Indeed, during cross-examination by Air Canada, Mr. Greco was prompted at this point to let the instructions play out, and, with some navigating, his screen reader announced that his search for VAN had returned six different destinations. He was then able to move down through the list to select YVR Vancouver, demonstrating that the information was available to him. In addition, evidence submitted by Air Canada in Issues Resolved Report 2 shows that entries are now read to persons using screen readers without navigation. The Agency therefore finds that city selection was not an accessibility issue.

[120] Flight selection had been identified in Mr. Featherstone’s December 2020 report as a task requiring assistance from the facilitator. However, during his demonstration, Mr. Greco was able to successfully choose flights that would suit his purposes from the list of available flights during both his examination-in-chief and cross-examination.

[121] At the time of the hearing, the time limit for completing a booking also did not constitute an accessibility issue because Mr. Greco was able to successfully extend the time limit during his demonstrations without any difficulty. The Agency accepts Air Canada’s explanation that time limits on the booking process are needed for technical reasons and that all users face them.

[122] Prior submissions, including Mr. Featherstone’s December 2020 report, indicated that entering passenger information was a task which raised issues for persons using screen readers. Mr. Greco testified during his demonstration that the input fields were not properly labelled until he pressed the up or down arrows to navigate through the list. This enabled him to determine that he was being asked for a date of birth, though he needed to tab back and forth between passengers in order to determine which passenger’s birth date the website was asking him to enter. Similar issues occurred when he arrived at the country input field. Evidence submitted by Air Canada in Issues Resolved Report 1 confirms that this issue has been resolved.

[123] Although Mr. Greco entered the seat selection overlay screen during his first demonstration and was unable to exit, ending his booking demonstration, this did not pose a problem during his second demonstration during cross-examination. Mr. Greco was unable to complete the second booking transaction during his cross-examination, but only due to time constraints related to other witnesses scheduled to give testimony. Mr. Greco testified that the only additional evidence to be gathered was the PDF confirmation document that would be sent by email, and he offered to complete a third booking between the second and third oral hearing dates, a proposition that was accepted by Air Canada and the Agency. Because of a timeout, Mr. Greco was required to restart the booking process over a third time to obtain an email confirmation. Mr. Greco completed this third booking in 34 minutes, which is only one minute longer than Air Canada’s average global booking time. This evidence demonstrates that Mr. Greco’s ability to complete a booking improved with usage, which, as Air Canada indicated, is true of all users.

[124] In light of the foregoing, the Agency finds that Mr. Greco’s demonstrations showed that he was able to complete a booking in a similar amount of time as the average user. Mr. Greco himself acknowledged that the issues he encountered did not prevent him from completing a booking.

Mr. DiNero’s booking demonstration

[125] The Agency accepts Air Canada’s argument that the accessibility issues Mr. DiNero encountered in his demonstration were minor and did not prevent him from completing a booking because there were workarounds for these issues or information could be understood from the context. For instance, Mr. DiNero encountered one of the same problems Mr. Greco did: when tabbing to the input field for date of birth, the website did not announce through the screen reader what information was required. Mr. DiNero was able to deduce from the context of the drop-down boxes what information he was being asked to fill in, and as previously mentioned, this issue has since been resolved by Air Canada. In addition, the abbreviation “m” read as “meters” instead of “minutes”, but this issue has since been resolved, as submitted in Issues Resolved Report 2, along with the issue of the seat selection table not reading in the right order.

Expert evidence

[126] The Agency determined in the Scope Decision that its role in these proceedings is to determine, based on the evidence before it, whether Air Canada can remove the obstacle or whether it had removed the obstacle at the time of the oral hearing. Although the Agency therefore places more weight on the evidence presented during the hearing, it recognizes that the expert reports are particularly helpful in terms of identifying potential issues.

[127] The Agency finds that Mr. Gupta and Ms. Liskovoi’s reports have limited reliability and probative value in terms of providing an assessment of the accessibility of Air Canada’s website for persons using screen readers. The Agency accepts Air Canada’s submission through its expert, Mr. Featherstone, that although Mr. Gupta’s report contains 181 accessibility errors, a more contextual analysis of the information on which Mr. Gupta relied would narrow that number down to 26 unique and identifiable errors. This is but just one example that highlighted that Mr. Gupta and his team lacked accessibility testing experience. The Agency also notes, as highlighted by Mr. Featherstone, that Ms. Liskovoi’s report did not contain the underlying data on which she relied. Therefore, while the report showed an understanding of accessibility, it is difficult for the Agency to evaluate its accuracy. Although the reports are helpful for the Agency in identifying potential accessibility issues, the Agency does not give the conclusions in these reports much weight given that these experts appear to lack significant accessibility testing experience.

[128] By comparison, the Agency finds that Mr. Featherstone’s reports have greater probative value and reliability because Mr. Featherstone has over 20 years of experience in accessible web design and accessibility testing. The initial report provides an analysis of the website approximately 9 to 10 months prior to the oral hearing. While the report identified no blocking issues at that time, it did identify three tasks requiring the help of a facilitator to complete. The Agency notes that all of these issues have since been resolved. For instance, Mr. Greco was successfully able to enter dates for flights instead of using the date picker and did not have difficulty choosing a flight to meet his criteria. Both he and Mr. DiNero were able to understand what information to enter when fields did not have labels, and this issue has since been resolved. Finally, evidence submitted by Air Canada in Issues Resolved Report 1 show the issue associated with use of the radio buttons for payment choice as having been resolved as of April 2021. The Agency also notes that although it did not witness Mr. Greco complete the payment information, Mr. Greco did not raise any concerns with this page.

Technical compliance

[129] Although Air Canada argues that WCAG are merely guidelines and should not be used as an assessment tool, WCAG has both guidelines and related success criteria that can be used to test whether digital content is meeting each of the guidelines. Furthermore, pursuant to section 9 of the Accessible Transportation for Persons with Disabilities Regulations^{Note 4}(ATPDR), Air Canada’s compliance with Level AA WCAG is required as of December 31, 2020^{Note 5}.

[130] Nonetheless, the Agency finds that technical compliance is not dispositive of whether Air Canada can remove or has removed the obstacle given that the obstacle was not defined in relation to technical compliance. Mr. Greco’s application framed the issue as functional usability by focusing exclusively on his inability to use Air Canada’s website with a screen reader. Other than his expert reports, Mr. Greco’s submissions continued along this vein throughout the proceedings. Only in his closing statement does Mr. Greco bifurcate website accessibility into functional usability and technical compliance and assert that Air Canada did not demonstrate the latter.

[131] The Agency further notes that Mr. Greco does not dispute that Air Canada demonstrated the functional usability of its website at the oral hearing and that witnesses from both parties—namely Mr. Lucchini, Ms. Liskovoi, and Mr. Featherstone—testified that a website may be functionally accessible even if it is not technically compliant.

Accessibility statement

[132] The Agency finds that an accessibility statement is not an essential component of an accessible website, and its absence does not result in unequal access. The Agency acknowledges witness testimony from both parties that accessibility statements can provide helpful accessibility-related information to persons who use screen readers and other persons with disabilities, but the presence or absence of an accessibility statement does not shed any light on whether there is still an obstacle or whether the obstacle could be removed up to the point of undue hardship.

Determination

[133] The Agency found in the Obstacle Decision that Mr. Greco faced an obstacle to his mobility due to unequal access to Air Canada’s website. The Agency finds that the obstacle was undue given that Air Canada has since been able to render its website accessible to Mr. Greco such that he and other persons who use screen readers are now able to independently complete a booking in a similar amount of time as persons who do not use screen readers. In light of the technological complexities of a booking website, the Agency accepts that the presence of minor issues for persons who use screen readers is not sufficient to establish continued unequal access.

Issue 2: What remedies, if any, should the Agency order?

The law

[134] Subsection 172(3) of the CTA, at the time the application was filed, read as follows:

On determining that there is an undue obstacle to the mobility of persons with disabilities, the Agency may require the taking of appropriate corrective measures or direct that compensation be paid for any expense incurred by a person with a disability arising out of the undue obstacle, or both.

Air Canada’s position

[135] Air Canada submits that it has removed the obstacle by making its website more accessible and integrating accessibility into its website development lifecycle to prevent future issues and that it cannot do more without experiencing undue hardship. Air Canada argues that, if the Agency decides that there is more that Air Canada can do to remove the obstacle, Air Canada should be afforded the opportunity to comment on the Agency’s proposed corrective measures. Air Canada states that Mr. Greco has not made any proposals or suggestions with respect to how the obstacle can, could or should be removed, other than his sole proposal that its website revert to its pre-January 2017 condition. It states that this is impossible because the code that generated that version of the website no longer exists. In addition, although the current user interface—the way the website looks—is the same as when the 2017 version of its website launched, much of the architecture or platform underneath has since changed and the current website incorporates regulatory standards that have come into effect since 2017.

Mr. Greco’s position

[136] Although Mr. Greco initially requested that Air Canada’s website revert to its pre-January 2017 condition, in his closing statement, he requests instead that Air Canada be required to submit to the Agency and to Mr. Greco a quarterly report on the usability and technical compliance of its website for an indefinite period of time. He also requests that he be allowed to make written submissions to the Agency within one calendar month of the submission of Air Canada’s quarterly reports about the accessibility of Air Canada’s website for persons using screen readers for an indefinite period of time. As discussed above, Mr. Greco argues that the Agency may retain jurisdiction indefinitely for the purposes of this order under section 24 of the Charter.

Analysis

[137] Having found that Mr. Greco faced an undue obstacle, the Agency will now consider whether it should order corrective measures. The Agency will not consider Charter remedies in this case, as indicated in the preliminary matter.

[138] In the Scope Decision, the Agency indicated that “[e]vidence on the procedures and safeguards, including testing, that Air Canada has in place to address the accessibility of its website may … inform the formulation of any corrective measures the Agency may order.” Thus, in deciding whether to order corrective measures, the Agency will consider the measures that Air Canada has implemented, as well as regulatory requirements that have come into force since the application was filed.

Measures that Air Canada has implemented to remove the obstacle

[139] Air Canada has removed the obstacle by integrating accessibility into its website development lifecycle and implementing measures to both prevent and address accessibility issues as and when they arise. The Agency is satisfied that these measures will continue to assist Air Canada in ensuring that its website remains accessible.

[140] Nonetheless, it is worth noting that Mr. Greco’s application contributed to Air Canada’s journey toward an accessible website. In 2017, Mr. Greco was unable to complete a booking on Air Canada’s website using a screen reader. The evidence submitted during the oral hearing suggests that Air Canada made website design choices during its website redesign in early 2017 that negatively impacted accessibility.

[141] For example, Air Canada did not engage independent expertise in accessible web design until well after the 2017 release of the redesigned website. In addition, Air Canada chose Angular as the framework for its new website in 2017, which witnesses from both parties—namely Mr. Duncan and Mr. Faxas—testified was initially not particularly well suited for coding accessible content. Witnesses from both parties, Mr. Featherstone and Mr. Duncan, also testified that organizations often select a technology based on how it will be useful to the organization, and these reasons may not include accessibility. While the Agency notes Air Canada’s position that Mr. Duncan’s report about the root causes for the inaccessibility of Air Canada’s website is largely conjecture and speculation, a point Mr. Duncan himself was careful to acknowledge, the Agency finds Mr. Duncan’s expertise on this point salient, and notes that Mr. Featherstone’s testimony was consistent in many respects with that of Mr. Duncan. These witnesses also testified that it can take much more time and work to address accessibility when unsuitable technology has been selected. The Agency therefore concludes that Air Canada did not prioritize accessibility when it overhauled its website in 2017.

[142] Since then, however—perhaps due in part to this application and the regulatory requirements that came into effect in 2020 —Air Canada has significantly changed its practices. Air Canada has integrated accessibility into its website development lifecycle and implemented measures to address and prevent accessibility issues, such as providing training to employees and establishing a system to identify, triage, and resolve accessibility issues. The Agency is satisfied that Air Canada has incorporated accessibility into its processes to the point that an order requiring it to do so is unnecessary.

[143] Air Canada also provided detailed evidence about the complicated process by which its website is developed and maintained. The Agency recognizes the need for Air Canada’s website to be flexible enough to respond to emerging technology, business practices, and regulatory requirements, which, as Mr. Featherstone testified, makes ensuring the website also reflects accessibility best practices even more difficult.

[144] In order to address this challenge, Air Canada created its Centre of Excellence to keep abreast of emerging accessibility best practices. Although the evidence supports Mr. Greco’s assertion that the Centre of Excellence is not yet well established, the Agency finds that Air Canada has established the Centre of Excellence to the point where an order mandating its operation would be superfluous. Further to Mr. Greco’s point that the Centre of Excellence does not have any members with a visual impairment other than UsableNet’s employees, regardless of whether these are UsableNet or Air Canada employees, the fact that individuals with visual impairments are engaged in Air Canada’s processes is sufficient.

New regulatory requirements regarding website accessibility

[145] Notwithstanding Air Canada’s repeated assertion that 100% compliance with WCAG is unachievable, section 9 of the ATPDR requires that large carriers, including Air Canada, ensure that every website that is made available to the public and that they own, operate or control meets the requirements for Level AA conformance with WCAG 2.1.

[146] In addition, as set out in Part 4 of the Accessible Canada Act^{Note 6}and detailed in the Accessible Transportation Planning and Reporting Regulations^{Note 7}, transportation service providers such as Air Canada now have obligations with respect to planning, reporting, and soliciting feedback on enumerated priority areas, one of which is communication technologies.

[147] None of these requirements were in force when Mr. Greco filed his application, and the Agency is satisfied that compliance with them will contribute to website accessibility such that corrective measures are not needed at this time. The Agency has similarly held in other decisions that the coming-into-force of a regulatory provision designed to remove the obstacle identified in a proceeding may obviate the need to order corrective measures. In Decision 32-AT-A-2021, for example, the ATPDR were cited at paragraphs 46-47 as one of the reasons for reviewing and rescinding the corrective measures ordered in a series of earlier decisions referred to as the Allergy Decisions. The Agency explained:

The complaint-driven process of section 172 often addresses circumstances where a person with a disability has already encountered a barrier, and focuses on the appropriate remedy that should be granted as a result of the incident. Conversely, the focus of regulations is to address barriers before, instead of after, they occur. As a means to establish consistent industry standards for the reduction, removal or prevention of barriers to persons with disabilities, regulations are more appropriate than adjudication on a case-by-case basis involving individual applicants and limited numbers of carriers.

[148] The Agency notes that compliance with these regulations is monitored by the Agency’s Determinations and Compliance Branch, as set out in its Compliance and Enforcement Policy.

Determination

[149] The Agency will not order Air Canada to implement corrective measures given that Air Canada has rendered its website accessible to Mr. Greco and other persons who use screen readers by incorporating accessibility into its website development lifecycle and implementing measures to both prevent and address accessibility issues as and when they arise. The Agency is satisfied that these practices, in conjunction with compliance with regulatory requirements that have come into effect since Mr. Greco filed his application, should ensure continued equal access. In these circumstances, the Agency finds that additional corrective measures are unnecessary and declines to order them.

Conclusion

[150] The Agency finds that the obstacle faced by Mr. Greco was undue given that Air Canada has since been able to remove the obstacle by rendering its website accessible to Mr. Greco and persons who use screen readers. However, the Agency will not order any corrective measures because Air Canada has removed the obstacle, has measures in place that should prevent its recurrence and is subject to new regulatory requirements regarding website accessibility.

Direction

[151] In accordance with subsection 25.1(1) of the CTA, the Agency has all the powers that the Federal Court has to award costs in any proceeding before it. Parties have until 5 pm Gatineau local time on January 11, 2023, to file any request for costs with the Agency’s Secretariat and copy the other party. Parties should keep in mind the principles which the Agency may consider when addressing costs as set out in Decision 58-AT-R-2018 (Murphy and Anderson v VIA Rail).

[152] Should a party submit a request for costs, the other party will have five business days after receipt of the request to submit their response. If no request for costs or to extend the time limit to file such a request is received by 5 pm Gatineau local time on January 11, 2023, the Agency will close the file.

---