E47: TestParty's Accessibility Journey: Connecting the Dots

In this episode of the United in Accessibility podcast, Ghanaian-American entrepreneur Michael Bervell, Founder and CEO of TestParty, shares his journey in digital inclusion, emphasizing the power of AI in advancing accessibility. 
 
He discusses TestParty’s approach of combining automation with human oversight to enhance server-side improvements, while also recognizing the importance of client-side tools, user training, and hiring people with disabilities to create ethical, user-centered design and meaningful impact in the accessibility space. 

Transcript

00:03 Speaker

Welcome to the IAAP United in Accessibility podcast. Today we're excited to feature Michael Bervell, a Ghanian-American Entrepreneur, Angel investor and Advocate for digital inclusion. Michael is the Founder and CEO of TestParty, a cutting-edge platform dedicated to automating digital accessibility compliance for web and mobile applications. Beyond entrepreneurship, Michael has made significant strides in philanthropy as the Co-Founder of "Hugs for", an international non-profit impacting over 300,000 youth globally. An angel investor in startups collectively valued at over $1 billion Michael also serves as an Elected Director of the Harvard Alumni Association, demonstrating his dedication to fostering innovation and leadership. Join us as we dive into Michael's impactful journey and gain valuable insights on advancing accessibility and creating inclusive technologies on the United in Accessibility podcast.

01:08 Christopher Michael Lee

All right, Michael, welcome to the IAAP podcast. It's great to have you here. Michael is the founder and CEO of TestParty. Michael, welcome.

01:19 Michael Bervell

Thanks so much, Chris, it's awesome to be here, especially after M-Enabling and all it's called conferences.

01:25 Christopher Michael Lee

Well, we love having you at M-Enabling this year, and I'm looking forward to sharing your expertise with our audience. Tell us a little bit about yourself and how you got TestParty started, and why digital accessibility is something that is on your mind now.

01:40 Michael Bervell

Well, for me, my journey into technology started, actually, as a kid, pretty young. I still remember I loved video games, and, like, one of my favorite games ever was RuneScape. And I still remember to this day, in fourth grade, my dad got us an Xbox, and they had the Xbox Connect, which was this, like new type of controller where you could use your body as a controller. And so, I remember, you know, we got the Xbox it was Christmas day around like, 7 or 8pm and, you know, my whole family is from Ghana, and my brother and my sister also love gaming. So, we, we set it up in this kind of, like, darker lighting conditions, and tried to use the Xbox, and it didn't work. The Xbox wasn't designed at that time, at least, like it wasn't common to use to go find diverse testers for products, and so we literally couldn't use the Xbox connect when it was essentially past 7pm in Seattle. So, I grew up loving technology, and I pretty quickly realized, at least with the angle of race, how race and technology intersects. And it really wasn't until I got into college that I started to look at other intersectionalities as well, one of which being accessibility and technology. And so that was kind of a bit of my start, in my beginning, and I ended up writing my senior thesis. I went to Harvard, graduated class of 2019, I wrote my senior thesis about privacy, and how do you preserve privacy rights in a realm of social media, right? It's really easy to think about privacy with physical space. You know you need a search warrant for someone to enter your house, but what does it mean to have a search warrant for your data? And this was right before GDPR and so on and so forth and now those sorts of questions are very common in Europe and across the rest of the world. But I love those types of topics, and so spent some time consulting at Google on accessibility strategy. Then I went to Microsoft, where I worked full time there, and was an Investor on behalf of Microsoft into startups, some of whom were doing accessibility type work and other DEI type work as well. And yeah, then they went back to business school, and now I'm here starting test party. So, it's been it’s been an interesting journey, but I think one of my favorite quotes is from Steve Jobs. He always says "You can only connect the dots looking backwards. You can never connect the dots looking forwards." And so it's interesting how all these kind of curiosities and things that happened in my life have influenced me starting TestParty now focused on trying to bring AI and some of these new ways of thinking into the accessibility space.

04:01 Christopher Michael Lee

Is kind of interesting. How those dots connect over time, right? You do have to look back a little bit, so tell us a little bit more, kind of drill down on TestParty. I mean, how did it get started? What's the business model behind it? What is it?

04:14 Michael Bervell

Yeah, so we started the company about two years ago. Now, it actually started while I was in business school, and I had always known that I wanted to be a founder and entrepreneur, but I knew I wanted to do it in a space that had a broader social impact. I mentioned a bit of my upbringing, but I'll dive a bit deeper into that, because it really influences what we do with the company today. I grew up in Seattle. I was the youngest of three siblings, and both my parents immigrated from Ghana to the US, pretty much knowing nobody, and I think did pretty well for us as their case, we were their main investment. And so, we were very solid, you know, middle class or even lower middle class. But me, my siblings and I really were instilled in this idea to give back, because our grandmother used to always do that with us and do that for us. One of the stories I love to tell, and I told this at M-Enabling on the panel, was the story of my grandmother stealing my teddy bears. Every year, she would tiptoe into our bedrooms during the Christmas holidays and grab some of our teddy bears from, you know, either under our beds or in the closet, and take them back with her to Ghana whenever she was visiting. She was smart that she would always want to, you know, leave Seattle in the wintertime and go to Ghana, West Africa, where it's way warmer. And we never really knew why she was doing this or what she was doing. And as a kid, I always saw her stealing my teddy bears, and I felt pretty protective of it. But when she passed away, and we went to her funeral in Ghana, we realized she had been taking all of these gifts and giving them to kids in hospitals and schools and orphanages, and we turned that into a non-profit. And so, for the last pretty much 15/16, years, we've been running that non-profit called "Hugs for Ghana". I'm actually wearing the sweatshirt "Hugs for Ghana" today. You probably can't see it because it's a podcast, but it's just a cute sweatshirt that has a teddy bear on it, but I bring that all up because I realized pretty early that, two things. One is that the world moves when you push on it, meaning, if you have an idea and you really want to act on it, even if you're you know some kid in Seattle who's, you know, 12 years old, you can put up posters in your high school and collect teddy bears and start to impact, you know, almost half a million children around the world, and now we do work in six countries. The second lesson I learned was that work is so much more fulfilling when you're doing something meaningful. So, in starting TestParty, I wanted to have, you know, a project that I knew could push the world in a direction that I thought, you know, was helpful and was needed. And second was a project that had what I call, you know, triple bottom line, like, of course, you have to, you know, make revenues, and of course, you have to sell a product that can help your shareholders. But you know, the third part of that bottom line is, are you doing something good for society at large? And when looking at industries, you know, we experimented with ending loneliness. That was one of our initial business ideas that me and my co-founder had. We looked at things around corporate meaning. How do you find meaning in work? Could we do and build products for corporations to find more meaning for their employees at work and TestParty was a pivot that we settled into, because my co-founder actually experienced an ADA lawsuit while he was working at Amazon. And so, we started to talk and realized that this really could be a great business opportunity, not just, you know, because there's such a big gap in terms of new technology and accessibility, but mainly because it's just such a big market that has such unmet needs with a huge impact. So, what TestParty does we try to keep it super simple, is we're building spell check for accessibility, and what that means is we have this tool that can automatically read the source code of applications and suggest fixes to engineers, and we focus purely on giving engineers better tools to write better code. And we think that's like the most fundamental reason why 97% of the internet is not, you know, WCAG compliant, because engineers just don't know how to do it. So that's really how the idea of TestParty started and my personal passion for this space beyond just building a business that has a business model.

07:52 Christopher Michael Lee

So fascinating. And thank you for sharing your story. Just curious, how, with the impact of AI nowadays into the space of technology and everywhere, actually, how is that impacting TesParty and the work that you are doing specifically around accessibility?

08:23 Michael Bervell

Yeah, I think it's interesting. I think what I'll do first is define what we think of when we say AI, right? Like, I think when most people say AI, they're thinking particularly large language models, like these, models that take in a bunch of data based on a query or a question will spit out answers related to that query or question, right? And so, people will call, you know, embedding. You take a bunch of data, and you embed it, then you can embed the question, and you can essentially relate the two embeddings to each other. That's one type of AI that's become really popular because of ChatGPT. Another type of AI that we think about a lot at TestParty is actually more around machine learning and computer vision. They're both sub segments of artificial intelligence, where one is, you know, natural language processing, that's the LLM side, and the other is the computer vision type side of the house. And so, when we think of the most useful use cases of AI as it relates to accessibility, we're looking more at the computer vision side of the house versus the large language model side of the house, but we do try to incorporate both tools, so I'll give examples of each. On the computer vision side of the house one thing that we started doing is taking any PDF document, which is generally not a super accessible format, and trying to recreate them into an accessible format. And we've chosen HTML as the output for that. And what that looks like for us is, can we screenshot any page of a PDF and run a bunch of transformations on it to turn any PDF page into an accessible HTML web page. So, whether that's a scan of a textbook, whether that's, you know, a Word document that's been saved that has a bunch of tables, whether it's something even as simple as a calendar, which seems simple to visually look at, but how do you reconstruct a calendar? With no human in the loop, just with computer vision or just with machine learning techniques. So that's one huge aspect that I think there's a lot of AI application within accessibility. Other examples would probably include, you know, smart screen readers. What would it look like to have a screen reader that can actually understand the context of a page by looking at what's actually on the page. So that's a whole area that we've been doing a lot of research in. And the second area, you know, within AI is, like I said, the natural language processing, LLMs, and the way that kind of comes into our business is through thinking at how do we rewrite code basis, right? There's a lot of data in GitHub of, like, really good code. There's also, of course, a lot of bad code. And so how do we clean up the code bases that we have to be accessible? So, if we were to run code transformations, those transformations would also be accessible. So that's how we think of AI. We really try to get down to first principles, to be like, what's most useful here. At the end of the day, AI is a tool we've noticed, and I think anyone would agree with this. It can't solve everything. It hallucinates a lot, especially when there's bad data. And accessibility data is really hard to come by, at least accessible data is hard to come by at the scale that you need for smart AI systems.

11:12 Christopher Michael Lee

So going back to the first example, with the PDFs, you know, I can see your business model in the sense I'm just working in that field making PDFs accessible. Most of the publishers provide PDF files for kids with disabilities going to higher education on K 12 that are not accessible. One of the challenges is around STEM particularly, right? Can you give us a little thought about how TestParty is working, that in that arena? Because, I mean, that's, it's a tough battle.

11:41 Michael Bervell

Yeah, we've built a few pilot programs with a couple textbook companies, and just built a demo to show what we can do with like math equations and science equations and so on and so forth. But STEM I would say it probably has three major issues. The first issue is, how do you represent information without giving away the answer? Like, let's say you want to get an Alt-text for a graph or a table. You want it to be descriptive enough that the student can learn supposing the student is not able to see the table without giving away the answer of what the table is. And a lot of AI, you know, if you put graphs or tables or images and it'll just tell you what it's a graph of, versus giving you the data points for you to into it. So that's one of the areas that we're areas that we're trying to understand, which is interesting, because in almost every other case of AI outside of that use case, you want to be as descriptive as possible, as quickly as possible, whereas in this case, you're trying to be intentionally ambiguous. And so how do you create an intentionally ambiguous AI system? So that's one major question. Second major question is things like math equations. We have a system that's able to take, really any list of equations on a page and turn them into MathML, which is the kind of traditional way of being read out. In our customer research, we realized that MathML is not a super great format like it essentially just reads the math equation from left to right. And as someone who may not have seen a math equation. It's actually quite confusing, right? To hear, you know, x squared over 3x log, like you almost even me saying it out loud, to get lost in the numbers. And so, there's this question of, is that really the best output? And that leads to kind of the third thing that we're thinking of, which is, should the outputs just be a one for one translation of the inputs, or should we try to do some sort of transformation to make it easier to understand or truer to the meaning of the text versus the literal translation of the text. So, these are three interesting just difficulties that we ran into. But at the end of the day, like one of our core metrics, especially when we work through universities, is, are we increasing the graduation rate of students with disabilities at your school? And that's our key metric, because we've noticed that student’s disabilities graduate disabilities graduate at lower rates. They tend to graduate in less STEM heavy fields, which then hurts their long-term outcomes in the job market and so on and so forth and so again, going back to first principles of trying to solve root causes. We think the causes, well, the information is not given to you in a way that's going to lead you to succeed.

13:59 Christopher Michael Lee

You know, it's interesting. Working in academia for years, one of the challenges that disability service officers face at universities is, you know, knowing the textbook or the material that's going to be provided from the faculty and only enough time in order for the conversion. So, I'm really excited to see what you build with this. It’s so needed. I ran a hub at Georgia Tech that focused on converting in accessible PDFs to accessible PDFs, as well as purchasing textbooks and cutting and scanning and doing OCR on them. It's a very, very timely matter, so if we have more solutions in this area, it's going to benefit that outcome, which is students graduating a higher rate than they are right now, and getting the knowledge they need to do that, which is huge, all right, so that's great. And MathML, I mean, it's, it's so expensive. I mean, it wasn't 4 or $5 a page to convert STEM related content. So, there's a lot here that needs to be solved. And I'm glad TestParty is in the pool. I don't quite understand the second example that you gave a little bit in regard to, I mean, I understand what you're trying to do is clean up code. Can you dive a little bit more into what exactly TestParty is doing in this area? Are you going to GitHub and you clean things up? Or is it, is it a little bit more strategic?

15:15 Michael Bervell

Yeah, it tends to be a bit more strategic. And this is kind of on the, you know, like we're saying, there is AI, there’s the computer vision, PDF stuff like that, and then there's LLMs, which goes into code bases. And so, what we're doing within code bases is, let's use a super simple example, right? You have 10 images hidden on your website and different pages, and maybe only two of those images have Alternative text, and the other eight do not, right? What current tools will give you is, will allow you to scan all the pages, and they'll flag where there might be images that don't have Alternative text, which is super helpful. The next step from that is now that you've found the image that doesn't have Alternative text, go back to your code base and add the right Alternative text into that image, and then redeploy right. And so, what we're hoping to do at TestParty is to be at essentially the source code level. So as engineers are writing code, we're scanning the whole code base and finding all 10 images, and not just saying, you know, this doesn't have Alt text, but actually trying to provide the Alt text. So, our whole philosophy is, there's been a lot of work done already, specifically with W3C and WCAG. And our question is, how can we make sure that that work is actually getting into the hands of people who want to use it right? What happens today is there's all this information, and then there's all these developers and designers, and there's a gap like the developers and designers are not using the information, even though the information is being really robustly updated. And so, our question is, how can we solve that gap in a way that's reliable? I think, when it comes to code remediation and source code remediation, the big question is, is coverage, how much of WCAG can you actually automatically identify by scanning, whether you're scanning in the source code or scanning on, you know, end pages, and how much can you actually remediate? And when we first started the company two years ago, we were like, oh, we will be able to remedy 100% of WCAG within three years. And this is like two naive accessibility guys coming into the industry, not really knowing much, thinking like, there's a rule set why can't you just automate the rule set? And there's a lot of things that are contextual. Or even last night, I was reading a blog post about the accessibility of election maps on election night. And that blog post had a lot of really interesting edge cases that I don't know if an AI tool would be able to capture. And there's so many different ways of making things accessible that there is some sense of art to the science, right? And so, you know, I think with the science, you can probably get to 40% remediation coverage, and the last 60% is that art, which no computer today is able to do. So, I think that's kind of where we think, when we think of, how can you use AI within the source code, it really comes down to, what can you identify and fix automatically? With regards to WCAG.

17:57 Speaker

With the adoption of WCAG, 2.1 in many countries, there is an increased demand for web developers, designers and other professionals with knowledge of web accessibility standards and guidelines. With this growth comes the need for an objectively verified level of expertise. The Web accessibility specialist exam will provide individuals and employers with the ability to assess Web Accessibility Competence. Complete the WAS and CPACC exam to earn the special designation of Certified Professional in Web Accessibility.

18:36 Christopher Michael Lee

I mean, you mentioned a lot about, you know, you've brought in the example of images and description. There's a science behind that. In a sense, long description was a short description. How you're addressing that through the technology.

18:49 Michael Bervell

Our approach and strategy, just from a philosophy perspective, has been always having humans in the loop. And so, what we do is we'll suggest something, but we won't assume that it's correct. And even whenever I post on LinkedIn, I'll upload my picture to ChatGPT or Anthropic and say, give me an Alt text for this. And even I'll adapt it. I don't think there's ever been a case where I've just copied pasted and hit Enter, because there's some context that I want to come through the image and the reason why I'm sharing it. When I think of some other examples of that sort of automation, with and without human the loop, I think one of the biggest examples is maybe even Closed Captions, right? You can go to, really, any YouTube video, or even zoom or Google Meet, and hit the closed caption button. And it's some percent accurate, right? I think for every 1% of packet loss, you know, you lose, maybe 5 or 10% of understandability is the numbers that I've heard. Packet loss being, you know, 100% accurate translation versus 99% translation, so 1% packet loss. So, you would assume from going 100 to 99 you would only lose 1% understandability, but it's actually much more severe. So even if a tool is 95% accurate at translation, you might be losing 40% of the actual understandability of it, right? So that's a really interesting nuance, but I think the same can be applied to Alt text, right? Your Alt text might be 99% perfect, but that extra 1% is going to come through, that human in the loop is going to approve a change or something. And so, we always think, how do you reduce packet loss, in addition to things like, how do you reduce false positives, and all the kind of rules that we that we suggest.

20:02 Christopher Michael Lee

Yeah, that's helpful. Thank you. Are you doing the thing with audio description?

20:06 Michael Bervell

So, we have seen some audio descriptions. I think Amazon released something two weeks ago that we were tinkering around with to try to get active. There's also a lot in sign language, automated sign language translation. I think once we get to a lot more media heavy environments, that requires a lot of data, right? And like I said, our strategy right now has been, how do we take existing information and make sure people are getting connected to it, versus trying to create net new data enriched products? And so for the most part, we've been leaning on the open source community for audio description stuff. I think Microsoft released something as well a couple weeks back that we're playing around with. This is what's beautiful about AI is that there's just so many people tackling it from so many angles that it really becomes a fun game of like Frankenstein creation and then massaging to create products that are actually scalable. But I think that audio description is something that is required right on sites. And so, the question is, how do we make it easier? But I imagine packet loss with audio description might be even higher than what we're describing with captions or just still images.

21:07 Christopher Michael Lee

Yeah, and, you know, audio description can be so subjective, just like, you know, text description. So, it's an interesting you almost, psychology comes into play a lot, right? So, let's dive into Overlays. So, Overlays are often criticized, often I put in parentheses, as you know, just superficial fixes. How does TestParty address these criticisms? And how are you approaching an accessibility standpoint?

21:37 Michael Bervell

Yeah, when we first started looking at this industry as a place where we wanted to build a company, right. Because we're looking at trying to build a generational company that uses our technology, not just in accessibility, but also in other industries, and then makes accessibility, like, seem cool and fun, so that people who are, like, my age, you know, just turning 30, want to work in accessibility for the next 10, 15, 20 years, right? And so, when we first looked at this industry and we saw a lot of the Overlay tools, we almost didn't decide to make the company here, because we're like, oh, there's already a solution. And when we started digging into what the solution was, we realized a lot of the criticisms around how they, you know, interrupt and hurt some of the experience of blind or low vision users. So, I think when we when we look at the market, we see kind of two sides to it. We call it server side and client side. Server side is what can the company be doing to make sure that their websites are the most accessible. And client side is, what should individual users be doing or adding such that they can be making sure that their experience using a site is accessible? I think we can think of a lot of really successful client-side businesses. NVDA, screen reader, JAWS, Voiceover, right? These are all client-side innovations, even when in the non-technical realm, walking sticks right? These are all client-side innovations that an individual user adapts. The burden is on that person to make their experience better. I think Overlays are interesting client-side adaptions. Yes, the business is the one paying for it and adding it to the site, so the financial burden is on the business, but the actual usability burden, or the usability benefit, comes through a customer's interaction and engagement. And I think some of the gaps that I've seen with overlay tools is that, if a customer is not well trained on how to engage with the tool and or if they're really well trained, but the tool is not adapted to their needs, there is a gap. I have talked with a lot of the Overlay tools like C suite teams and so on and so forth. And that's their focus is, how do you make this client-side tool better for the client and the customer in a way that it becomes scalable? I think that's a gap that they're trying to address, but I do think that they have a lot of scale in a way that a lot of accessibility agencies or consultancies haven't reached yet, and so I think that's, you know, last year around this time, Level Access and UserWay announced their partnership. I'm sure that was part of the reason, as a way of creating this pipeline of, you know, initial, maybe not so, accessibility, aware, educated consumers or businesses to now buy subsequent products. We try to focus purely on the server side, thinking, what should businesses be doing to be building better websites, so regardless of what happens on the client side, whether you're using an Overlay or whether you're using voiceover or screen reader, you can have a better experience. We have seen that if you just do better server-side improvements, you may need less of the client side tools, because the server side is just like, that's where most of the websites being built. Essentially, if you just build better sites, what's the effect? And that's where we're trying to focus on so giving tools to engineers, to designers, to companies, since you know, they're the ones creating the products. So, the burden should be on them to be. Create better products versus the user to consume it in a different way. So that's a bit esoteric and philosophical, but we do try to, like, bucket, at least most of our industry into those sorts of frameworks.

24:57 Christopher Michael Lee

So, on the client side, obviously, and that was very well described. Michael, thank you for that. Your thoughts on that. I mean, I think, you know, training and you know, is important, right? And just having the users understand, because at different levels of users and different levels of use of assistive technology, for example, whether you're, you know, using JAWS or another screen reader software. So, I think that's very helpful. And you know, the other bucket, the way you'll focus on is, you know, more the server side, looking at businesses. How does test party, you know, kind of address, you know, you know, legal requirements. And is this Overlay phenomenon being that part of what you're trying to address with, with your clients?

25:42 Michael Bervell

I mean, all the explanations that I gave me was very much like the like the engineer in me, right? And then the artist and philosopher in me is thinking, okay, well, what is the right thing to do? Because I think a big reason why Overlays grew the way they did is because they said, If you install this, it'll fix your problem, right? Which, you know, startups constantly oversell. And so, you know, I think overselling is okay, but if the consequence of overselling is hurting a population, you can almost argue it's probably not as great, right? I think what's interesting about accessibility as an industry is your AB tests relate with people like your AB testing on people's lives and their experience using technology and navigating the world. And that causes into question is that an ethical AB test? So that's a bit of the philosophical or the more kind of human side, but I think test parties philosophy is always just to be honest in advertising and honest in marketing, and saying that we can get you to this percentage of compliance, we think, with pretty high probability, and to go the rest of the way, you should have humans in the loop, whether you're hiring through an existing accessibility consultancy or hiring people with disabilities to test your product and to give solutions. You really do need to pair both sides of the coin. And the question becomes, what are the low hanging fruit that you can do with automation? And then how can you build on top of that with consultancies or agencies, or whatever it might be who are experts in this? I think that's very different than what you saw between 2019 and 2022 with some of these startups in the space, because they tried to sell their tool as the one overall, end all be, all solution, which, to be fair, that's how we came into the market as well, thinking that that would be really easy solution. And now we're realizing two years in, you know, that these partnerships are really worthwhile beyond technology, but also into giving people disabilities, jobs that they really are well positioned to crush. Right? Like they're amazing, they're super talented and this is one of the areas where we can actually be thinking about employment as well, and really high employment options.

27:42 Christopher Michael Lee

So, you've peppered in this manual versus automatic, you know, testing, and it's how the importance of having people involved of steal at this time that may change down the road as technology changes. Talk about your philosophy, or test parties, philosophy kind of around the user centered approach with, especially with all these standards and guidelines are out there.

28:04 Michael Bervell

We have, I guess, two approaches as it comes to user centered design. One is like internally before we release a product, and the second is like externally after we release a product. And so, we certainly think at all times it's like automation and people, it's never like an “or” it's always like it's a “and” statement. So internally, like, we've tried pretty hard to hire people with disabilities to be on our team, and so we did our recent survey in September. We do like an annual employee survey, we have 25% of our team as people with disabilities or impairments, which is awesome, and 75% of our team has an immediate family member, meaning a parent or a sibling or a child who has a disability or impairment, even if you look across, you know, the population at large, I think what that means is people are coming to work here because they have a mission and a passion for it, and so they're bringing to every product that idea of is this accessible before we release it, and even after we release it like we love being criticized and told this didn't work for me here's the reason why, because we take those learnings and roll it into future product releases. So, I'll give you a pretty clear, concrete example. We work with a lot of Shopify sites to make their sites accessible and compliant. I think today, we're influencing over half a billion dollars of revenue on an annual basis, just with our automated scanning tools, with the sites that we work with, and oftentimes when we're auditing these sites and doing some of these manual fixes, we'll find that there’s a lot of third party tools on the sites that aren't accessible, and so now we have this habit of writing the emails for the company, for them to send to the customer service rep at the third party tool to fix, which has been really interesting. So, for instance there's a reviewer tool that we've seen four or five times, and we've had four or five different customers that we've worked with email that reviewer tool and say, "Hey, this isn't accessible. Here's how the test party team recommends that you fix it." And that's something that you wouldn't be able to get through just automation, but once we do that at nausea, you know, over and over and over again, now we're starting to influence other companies to change their practices, and even beyond our customer, customer base of four or five, we're influencing now millions of potential customers around the world with more accessible products through these really large third parties. So that gets a bit to like our philosophy, which is, start with yourself, be really intentional about having a diverse team and disabled team, and then go from there to know that this is even a problem and how they should fix it. You usually won't see our logo or our brand a lot of the sites that we work with, but we try to kind of influence through that sort of those sort of tactics and means.

30:35 Christopher Michael Lee

Yeah, third party going that direction is always a smart direction if you can make that happen. So, I can commend y'all for, for going that, that route when you can. So, you started with this, moving down this route about, you know, as a young company, two years old, you know, you make mistakes, things happen differently. What would be your recommendation to other companies, though, whether they're startups or they've been around for a lot longer, when something does happen within the company, that you're like, okay, particularly in the accessibility arena, how do you put roadblocks? How do you make sure that when you're releasing something, whether it's marketing, any kind of communication, that you're aware of the community that you're reaching out to?

31:19 Michael Bervell

Personally, I'm going to at least one or two accessibility meetups in person per week, right? Because I think a lot of these conversations just happen like, you know you should find a water cooler. And, of course, in this next now virtual first world, it's a lot harder to hear some of some of those conversations. But I think that's one thing is to be where the conversation is happening so that you can know, you know, either anticipate or react quickly to you know what you might have done wrong? Of course, the second thing is to like, always be like, on the lookout for like advocates, and also for people who are like, vocal about not being advocates and engaging with both types of people, right? I think the only way that you can learn is through friction, or I guess the only way that you can build a fire is with friction. And, of course, fire is super useful. And so like, what can we be doing to actually go out and find people who have intentional frictions and intentional disagreements and try to understand what the disagreement is? I think almost everyone in the industry agrees that the world is not as accessible as it should be, and that ableism is really at the root of a lot of technologies and just thinking. And so, no matter you know what people would say about how we operate or how we do business, I think they would agree that if we can eliminate ableism, that would be a net benefit. So, knowing that we have that point of agreement, the question is, what could we be doing to actually get better get to that point? So, we try to focus on that ultimate end state, which is an always active battle. Like, I wish I could be like next year will end ableism. It's one of our key performance indicators for the year. That's really hard to get to, but I do think that everyone in the space is thinking about that sort of stuff, and that's what we try to focus on, and seek out disagreement.

32:56 Christopher Michael Lee

I really like the analogy with the fire. I mean, you've got to have friction, right? And you've got to have both sides coming together, having that dialog. So, such an important thing in many areas of society. So, projects, what's on the horizon for TestParty, what's next?

33:15 Michael Bervell

Yeah, we're trying to do a lot more Open-Source project. I think we were inspired by some of the Open-Source tools in the space. I think ExCourse score is a really good one. Wave, Alpha, these are all really great examples of how people can build on top of the work of others to create a better a better industry, a better market. I think ultimately our strategy as it looks towards 2025 is, how do we expand the pie, right? I think a lot of people ask me, you know, what do you think about your competitors? What do you think about competition? And my answer is pretty much always the same, like our competition is the 97% of websites that are not accessible. It's not, you know, fighting for the 3% that are accessible and trying to get the biggest dollar value from those customers. If there was any other industry where someone said you have a 97% market share that you can go after, everyone would be sprinting towards that, right? And so, I think that that's what we're trying to push the industry into. And so, we're doing that through a lot of Open Source tools. We just released one this week around color pickers. So, it's a color picker that actually has a flag if it's accessible or not accessible based on the background and foreground. We're doing a lot more research, hiring interns and people in college who have disabilities to write about their experiences and also to write about what they've seen in the market and what we should be doing, and we're hoping maybe even to start a podcast in the next like 9, 10, 11 months. And these are all areas where we're thinking, what could we be putting out into the world to expand the potential pie of people who know about accessibility and care about accessibility? And of course, like, we're focusing on the growth within our own business as we target, you know, trying to be $100 million company within the next five years.

34:51 Christopher Michael Lee

Those are great goals. A lot of exciting things on the horizon. Michael, I've really enjoyed having you today, and the work that you're doing is fascinating. I'm really excited about the PDF optimize. I lean that way right now, but there's such good things that we'll definitely keep watching you. Are there any takeaways that you want to leave our audience today that you want to share?

35:13 Michael Bervell

Yeah, I mean, I think my biggest takeaway so far in working in this space, in accessibility, is there's so much good already out there, and the question is, how do you find it and package it in a way that people aren't able to use it really well. And that could be something as simple as like, you know, if you're listening to a podcast that you really love, you know, take two or three bullets and send it to your boss and ask them to send it to their boss something that's really simple and bite size that gets to the point of how important this market is, right? The spending power of people with disabilities is massive, I think at the end of the day, a lot of people will develop into having a disability. So, I almost see the work that I'm doing today as helping my future self. Who knows what might happen in 20 to 30 years. It's already helping my immediate family, some of whom have disabilities. And so, what could I be doing to help myself in the future if I do develop a disability that affects how I go about the world day to day. So those are some of the things that I that I think the audience, hopefully will go away from this with. Is, you know, what can I be doing day to day to make a small change, and how do I pack in information in a way that benefits and already builds off the work of so many smart people who've come before us?

36:21 Christopher Michael Lee

Sharing right? Michael, thank you so much. It's been a pleasure. We're looking forward to following you and your career and learning how test party will move to the next level. And thank you for your commitment to accessibility and working towards it.

36:33 Michael Bervell

Thanks so much, Chris, it's awesome to be here again, guys. Thanks so much.

36:37 Speaker

Do you want to continue the discussion from this podcast? Members can access the Strategic Leader in Accessibility Community of Practice in our Member connections platform. If you are not a member, please check our website for all IAAP membership benefits, or email us at info@accessibilityassociation.org, and we will be happy to talk about membership and help get you engaged.