Showit Website Designers vs AI Builders: What the Data Shows

Showit website designers draw close to 600 searches a month in the United States, according to Semrush, in the year the web’s largest annual audit found websites getting measurably worse for the first time in six years, and specialists who build on the platform say the two are linked.

AI website builders have made publishing a site a matter of minutes. The evidence now arriving from independent researchers is that the sites being published are heavier, less accessible and less secure than the ones they replaced. The businesses which depend on their websites, those specialists say, are responding by hiring people rather than prompts.

The clearest measure comes from WebAIM, the Utah State University accessibility group that audits the home pages of the top one million websites every February. Its 2026 report found that 95.9 percent of home pages had detectable failures against the Web Content Accessibility Guidelines, up from 94.8 percent a year earlier. In WebAIM’s words, the figure is “reversing a trend of small improvements each of the previous 6 years”.

The rest of the report points the same way. The average home page carried 56.1 detectable errors, up 10.1 percent on 2025. The average number of elements on a home page rose to 1,437, “a 22.5% increase in only one year”, and has nearly doubled in seven. Pages are getting bigger and more complicated faster than at any point in the audit’s history, and the failures are growing with them.

Share of the top one million home pages with detectable WCAG failures. Source: WebAIM Million, 2019 to 2026.

That period coincides with the arrival of AI site generation as a standard feature. Wix, Squarespace, Hostinger and Framer all now offer to produce a complete website from a short description, and a newer class of tools will turn a paragraph of text into a working site with a database behind it. None of them is responsible for WebAIM’s numbers on its own. Together, they have removed the moment at which someone who knew what a heading, an alt attribute or a form label was for would have looked at the page before it went live.

The builders make the opposite case: that a generator applies a consistent structure a first-time owner would never manage alone, and removes the mistakes people make. The audits published so far do not separate AI-built sites from the rest, so that claim has not been tested either way. What they do show is the direction of the whole.

The code underneath is not faring better. Veracode, the application security company, has spent two years testing what happens when large language models are asked to write code with no security guidance. Its March 2026 update, part of a programme that has now evaluated more than 150 models and covering the latest releases from OpenAI, Google and Anthropic, found that “in 45% of cases, the model introduces a known security flaw into the codebase”. The models’ ability to produce code that runs has climbed past 95 percent over the same period; their ability to produce code that is safe has stayed flat at roughly 55 percent since 2023.

“Models have become excellent at writing code that compiles,” wrote Felix Brombacher, Veracode’s senior director of product management, in the report. “They’ve failed at writing code that’s safe.” For websites in particular the detail matters: the models passed Veracode’s cross-site scripting tests, the class of flaw that lets an attacker inject content into a page, only 15 percent of the time.

Google, for its part, has said there is no shortcut. Its guidance on AI features in Search states that “there are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary”, and lists instead the fundamentals it has always asked for: content in textual form, structured data that matches the visible text, internal links, page experience and crawl access. Those are the parts of a website a generator does not see and a first-time owner does not know to check.

For most businesses a mediocre website is an inconvenience. For photographers, therapists, wedding venues and other businesses whose site is the business, it is the difference between being booked and being invisible, and it is in those markets that the reaction has been sharpest.

Adil Makhdoom, a website designer and SEO specialist who builds on Showit, Squarespace and WordPress, audited 51 live business websites earlier this year to see what their platforms actually emitted for search engines to read. One of the eleven Showit sites in the sample carried structured data, and that one came from the WordPress blog attached to it. Squarespace produced markup on every site, but six of the ten local business nodes it generated had neither a name nor an address. WordPress with a free SEO plugin was the only configuration that consistently described the business in full.

“What the AI builders sell is the part that was already easy,” Adil said. “Anyone could get a good-looking page in an afternoon three years ago too. What nobody has automated is the structure: the headings that tell Google what the page is, the markup that says which business this is and where it works, the text under the images, the form a screen reader can use. That is still done by a person or it is not done. The WebAIM numbers are what it looks like when it is not done at scale.”

The market for that work is measurable. Semrush puts United States searches at 590 a month for “Showit website designer” and a further 260 for “Showit designer”. A smaller category of Showit SEO services and Showit web design that accounts for search from the first page has grown alongside it. The searches now come from therapists, venues and florists as well as photographers, the businesses for whom SEO for photographers and its equivalents are not a marketing channel but the only one.

Showit website designers occupy a particular position in that shift, because Showit is a canvas rather than a template. It gives the designer complete visual control and, in exchange, assigns none of the structure automatically. In Adil’s view, that makes it the platform where the difference between a designer who handles the technical layer and one who does not is most visible, and where AI generation has the least to offer, since there is no template for it to fill.

Adil said the tools themselves are not the problem. “I use AI every day. It drafts, it checks, it saves hours. What it cannot do is take responsibility for the result. A site that WebAIM would flag and Veracode would fail can still look perfect in the builder’s preview, and the owner has no way of knowing until the bookings do not come.”

For a business about to hire a web designer, he said the questions have not changed, only the urgency. Ask for a live site the designer built and run it through Google’s Rich Results Test, which shows in seconds whether it carries structured data. Ask how heading levels are assigned. Ask what the site looks like with the images turned off, because that is what a search engine and an answer engine read. For sites on Showit specifically, free checkers such as Showit Site Analyzer read a site’s search structure, speed and AI readability without a signup and report what a builder or a designer left out.

WebAIM cautions that the errors its tools can detect are only a fraction of the accessibility problems on the pages it audits, since most failures need a human to find. That, in the end, is the whole story. The web has never been easier to publish to. It has never had fewer people looking at what was published before it went live.

Disclaimer: This article is for informational purposes only and reflects general observations about Showit website designers and AI website builders. Results may vary depending on individual needs, skills, budgets, and project requirements.