Michael Bervell

That makes the theme a better place to look for fixes than treating each page as a separate problem.

In a scan covering 19,581 storefronts, Bervell says the median store broke six rules, each repeated about 5.5 times.

• They explain that an unlabeled cart icon in a header can produce broken links throughout the catalog.
• They also report more failures on collection and product pages than on homepages.

Read original

An apparent compliance measure may not resolve the underlying barriers and could itself be cited in a demand letter. They frame accessibility as a question of both technology and the economics of responding to legal claims.

In a recap of an Honest Ecommerce conversation, Bervell says most settlements cost $10,000–$25,000, roughly what it can cost to prove compliance. They describe demand letters naming widget providers as evidence of non-compliance, including one that listed twelve widgets, and say source-code fixes hold up better.

Read original

They argue that both depend on the information exposed by a site's accessibility tree.

Bervell explains that a button without an accessible name can be invisible to a screen reader and to an agent trying to add a product to a cart; an unlabeled quantity field can also prevent the agent from completing checkout. They note that agentic traffic is small at present.

Read original

It answers a newcomer’s first question: how widespread are ecommerce accessibility problems, and where do they occur in a store?

Bervell reports scanning 19,581 ecommerce storefronts and finding 140 clean, or 0.75% — one store in 134.

• The rest carried 1.77 million broken elements: buttons with no name, links with no text, quantity fields with no label, and controls nested inside other controls.
• Three findings stand out.
• First, the problem is the theme, not the content: the median store breaks six rules and repeats each roughly 5.5 times, so one unlabeled cart icon in a header component ships across an entire catalog.
• Second, the homepage is a store’s best page — failure counts grew about 3.9 times deeper into collection and product pages, where purchases happen.
• Third, nobody is perfect: 27% of stores scored 90 or better, none scored 100, and the best score in the whole set was 97.
• Bervell also argues that AI shopping agents read the same accessibility tree as a screen reader, making an unnamed button invisible both to a blind shopper and to an agent trying to add an item to a cart.
• The cleanest stores named are Graza, OLIPOP, UNTUCKit, Liquid Death, FIGS and OLAPLEX.

For brands planning a campaign

A co-marketing webinar can open with these scan numbers and the theme-not-content framing, shifting the conversation from a vague legal worry to a specific, fixable place in a Shopify theme. The agent-readability argument gives the session a forward-looking second half beyond ADA demand letters.

Bervell offers a $1,000 Amazon gift card, with no cap, to anyone whose two-line introduction to an ecommerce brand turns into a TestParty customer.

• The post pairs the offer with a personal story: Bervell was an intern at Harlem Capital in 2020, stayed in touch through alumni events, congratulations notes and no-agenda coffees, and four years later Harlem Capital co-led a $4M round in TestParty — the first former intern the firm had backed.
• They cite Mark Granovetter’s 1973 paper "The Strength of Weak Ties" and its finding that over 80% of jobs found through personal contacts came from people seen only occasionally.
• The brand-side pitch is that ADA demand letters reached 6,000+ brands in a year and 96% of top websites still fail accessibility standards.
• TestParty is described in one line as using AI to find and fix accessibility issues in a website’s source code, with 30 million+ issues fixed and brands including Eddie Bauer, Magic Spoon and Levain Bakery.

For brands planning a campaign

Bervell works through incentive mechanics and warm introductions rather than paid media. A webinar partnership is most likely to land if it presents two audiences vouching for each other — the agency’s Shopify clients and TestParty’s DTC contacts — rather than a one-way placement.

Summarizing the Honest Ecommerce podcast appearance with co-founder Jason Tan, Bervell lays out three points.

• About 5,000 demand letters were filed publicly the previous year, with an estimate of 10,000 to 15,000 once private settlements are counted; most settle for $10,000 to $25,000, roughly what it costs to prove compliance, so brands pay even when they believe they are right.
• Overlays, they argue, have become a target rather than a shield: demand letters have named widget providers by brand as evidence of non-compliance, and one listed twelve different widgets.
• Source-code fixes hold up better — under 5% of TestParty’s active customers had received a demand letter.
• Bervell notes that the industry extends beyond technology into policy and strategy, and links the full episode in the post.

For brands planning a campaign

These three points make ready-made webinar segments, while the overlay position directly counters agencies already selling overlay tools to clients. The named podcast offers a format template and a host to reference.

Bervell states that more than $4.4 billion in GMV moves through TestParty customer sites annually across 1.1 billion or more accessible sessions, which they put at roughly 1.2% of Shopify’s GMV.

• Brands named include Claire’s, OLIPOP, Hulken, OluKai, Magic Spoon and SUKOSHI, alongside "hundreds more."
• The post is framed as Q3 data.
• Ask for the calculation method before relying on the figures.

For brands planning a campaign

For an agency working on Shopify, this is the most directly relevant proof point in the set — it puts the platform and the customer base in the same sentence. It can anchor a "what's at stake" segment, with the caveat that the underlying calculation is not published.

Bervell describes creating a two-minute "Editorial Motion Graphics" explainer for TestParty in about an hour during lunch for $20, and posts the full prompt.

• It lays out the substance — TestParty as an AI accessibility compliance platform that fixes WCAG 2.2 AA violations in source code rather than applying an overlay widget, built for Shopify and DTC brands, with 200+ customers, backed by $4M from Harlem Capital and Urban Innovation Fund plus NSF SBIR Phase I and Phase II — and four visual directions: a blind shopper in a kitchen failing with a screen reader, a crash zoom into source code rewriting itself clean, a cracked storefront window under a decorative film as the overlay-versus-root-fix metaphor, and a confetti drone flythrough for brand energy.
• The prompt specifies camera motion, subject, lighting and film stock for each, recommends Seedance 2.0, Kling 3.0, Veo 3.1, Wan 2.7 or Sora 2, and closes by asking what other video use cases people have seen.

For brands planning a campaign

Bervell shares reusable production assets, not just opinions. A webinar could offer a shared prompt or creative pack as a giveaway, and the four visual directions provide a ready-made creative brief for the agency’s client pitches.

The headline finding: failures come from themes, not content.

• The median store breaks six rules and repeats each one about 5.5 times, so a single unlabeled cart icon can ship as dozens of broken links across a catalog.
• Product and collection pages were 3.9 times worse than homepages.
• No store scored 100; the top score was 97, and 27% scored 90 or better.
• Stores named among the cleanest include Graza, OLIPOP, UNTUCKit, Liquid Death, FIGS, and OLAPLEX.

A large-sample scan reported through aggregate counts by rule and page type, with specific stores named as positive examples.

For brands planning a campaign

Offers a webinar a data-led hook an agency can reuse: a benchmark for the client's own theme, a page-type finding pointing to product pages, and named peer brands for comparison. The scan result is available to request for free, supporting a co-marketing opt-in offer.

Examples: Source 1

Most settle for $10,000–$25,000 — roughly the cost of proving compliance in the first place. They argue overlays have become a liability rather than a shield, citing one demand letter that named twelve widget providers as evidence of non-compliance, and say under 5% of active customers have received a demand letter.

Podcast conversation; the figures are presented as estimates based on TestParty's view of the market.

For brands planning a campaign

This is the proposed webinar's compliance-risk half, already framed as an argument an agency audience can follow: pay to settle or pay to fix. The critique of overlays offers a clear contrast for positioning source-code remediation.

Examples: Source 1

A button with no accessible name is invisible both to a blind shopper and to an agent trying to add a SKU to a cart.

• An unlabeled quantity field leaves an agent unable to complete checkout, while nested interactive controls can send it to the wrong element and prompt it to abandon.
• They note agentic traffic is small today and frame the theme decision around whether a brand will be in the consideration set later.

A parallel drawn from the scan findings, not a separate measurement of agent behavior.

For brands planning a campaign

A forward-looking segment connects compliance work to agentic commerce, broadening the webinar's appeal beyond legal risk for DTC brands weighing where to spend.

Examples: Source 1

They cite Mark Granovetter's weak-ties research, which found that over 80% of jobs secured through personal contacts came from people seen only occasionally.

• The post turns that into a standing offer: $1,000 for a two-line introduction to an ecommerce brand that becomes a TestParty customer, with no forms and no cap.
• It also describes TestParty as AI that finds and fixes accessibility issues in website source code, with 30 million-plus issues fixed for brands including Eddie Bauer, Magic Spoon, and Levain Bakery.

A personal narrative leading into a commercial offer, supported by a cited social-science paper.

For brands planning a campaign

Shows a simple incentive model an agency partner could adapt for webinar promotion. The story-led, direct offer also gives a sense of how Bervell might carry a live session.

Examples: Source 1

They published the full prompt, including four visual directions: a screen-reader user failing on a store page, source code rewriting itself, a cracked storefront window compared with a taped-over fix, and a celebratory office scene. The post recommends Seedance 2.0, Kling 3.0, Veo 3.1, Wan 2.7, or Sora 2 and asks what other video use cases readers have seen.

For brands planning a campaign

A topic beyond compliance that shows how Bervell tests and explains AI creative workflows. It could provide co-marketing assets or a lighter companion strand if the accessibility angle needs one.

Examples: Source 1

The company has 200+ customers and funding from Harlem Capital and Urban Innovation Fund, plus NSF SBIR Phase I and Phase II.

• A quarter-end post states that more than $4.4 billion in GMV flows through customer sites annually across 1.1+ billion accessible sessions, about 1.2% of Shopify's GMV.
• Named brands include Claire's, OLIPOP, Hulken, OluKai, Magic Spoon, and SUKOSHI.

For brands planning a campaign

Provides credibility markers an agency will want before committing to a co-marketing session: customer count, funding, and a scale figure tied directly to Shopify GMV.

Examples: Source 1 · Source 2 · Source 3

It sells AI accessibility software that fixes WCAG violations in a site's source code rather than adding an overlay widget. Their LinkedIn posts focus largely on the product, data and customers.

View example

At 21, they were the most junior person in every room.

View example

Bervell says it was the first time the firm backed a former intern. They say the firm made the investment, with the anchor check stemming from the relationship built during the earlier internship.

View example

View example

View example

The customer relationship is with TestParty.

View example

View example

View example

View example

View example

View example

View example

View example

The posts don't state that Shopify is a partner or sponsor; ask whether a relationship exists.

View example

Bervell summarizes their conversation about demand-letter economics and overlay widgets.

View example

Both turn compliance language into a service website owners can buy.

Irum works across WordPress and small-business, law-firm, and professional sites, pairing accessibility with SEO and AI automation services. Bervell works at ecommerce scale around Shopify storefronts, publishes aggregate scan data rather than client-level project descriptions, and argues from the economics of legal claims against a platform that fixes source code.

Read original · Original 2

Both write about accessibility decisions as engineering commitments, with public detail on baselines, contrast ratios, focus indicators, screen-reader order, and crisis-handling rules.

Rudolph applies this standard to an AI product for education and mental-health support, writing about duty of care, privacy in crisis situations, and notification rules for minors. Bervell applies it to ecommerce storefronts, measures failures across thousands of shops, and centers the legal and conversion costs for DTC brands.

Read original · Original 2

That aligns with Bervell's argument that convenience features do not make a store functional or compliant.

Gatenby works agency-side on design, brand, and storefront projects — including menswear and heritage retail — and evaluates tools from a client-delivery point of view. Bervell sells the remediation platform itself, publishes scan statistics across tens of thousands of stores, and speaks mainly to legal risk and agent readability.

Read original · Original 2

Gatenby's post on the Shopify Canvas demo works through a concrete release: Canvas lays out a store in one workspace; a theme edited in Canvas stops receiving theme updates; and Sidekick checks that code works and looks right, but not whether it is accessible.

• Gatenby ties that to the European Accessibility Act, which has applied since June 2025 to many stores selling to EU customers, and to whether AI shopping tools can still read a store's content.
• A second post about taking on a Harrogate menswear client shows how Gatenby frames a merchant relationship around why customers buy.
• For a store-owner audience, that is closer to their daily decisions than a compliance scan.

The tradeoff is reach and evidence type.

• Gatenby writes about individual storefront projects and product commentary.
• The posts here contain no scan across thousands of stores, demand-letter or settlement figures, or product that fixes accessibility in source code.
• For a campaign explaining the scale of the compliance problem across merchants or what a remediation product actually does, Bervell's posts carry that material; Gatenby's do not.

Examples: Source 1 · Source 2

Irum's post lists the services in detail: website accessibility audits, WCAG-focused improvements, WordPress accessibility remediation, keyboard navigation fixes, form, link, button and heading structure fixes, contrast and readability work, and manual testing.

• The post frames accessibility as a business problem as much as a legal one.
• A second post describes building a Fluent Forms to Clio Grow intake integration for law firms, pointing to a legal-sector client base.
• A webinar for a legal or professional-services audience that ends with an actual site review suits this evidence better than a merchant-scale compliance scan.

The tradeoff is that the work involves service delivery for individual clients, not ecommerce research or a software product.

• The posts here contain no storefront scan data, figures on demand letters or settlements, or coverage of Shopify themes or DTC brands.
• An agency selling Shopify work to merchants would need to supply the ecommerce framing itself.

Examples: Source 1 · Source 2

Rudolph's posts describe Jean OS launching with WCAG 2.2 AA as a foundational design choice, detailing contrast and focus indicators, minimum target sizes, screen reader and Braille reading order, cognitive and dyslexia-friendly design, and reduced-motion and voice-control support.

• A second post sets out a pledged crisis response: staying present with someone in distress, logging only a category and timestamp, and never routing an adult's distress signal to an employer.
• For a session aimed at teams building assistive or educational AI, that mix of baseline specification and duty-of-care policy is the closer comparison.

The tradeoff is the buyer and setting.

• Rudolph's material is about AI product design and responsible-AI policy in education and health, not ecommerce compliance.
• It has no storefront or theme examples, demand-letter economics, or merchant-facing remediation offer, so a DTC-oriented webinar would need to translate the framework rather than reuse it directly.

Examples: Source 1 · Source 2