Can AI search tell who your company is? Schema, profiles, pricing and about pages on 1,586 software company websites

Software company websites do better at telling AI search engines who the company is than what the product costs. Of 1,361 commercial software company websites, only 22.7% state amounts on a pricing page, while 68.8% of all 1,586 sites use one consistent brand name, 47.8% link three or more profiles, 44.8% have both an about and a contact page, and 43.1% have complete Organization JSON-LD. The data is from one run of version 2 of my AI search readiness check on 22 October 2026: 1,601 software company websites attempted, 1,586 scored. This article covers the five entity checks. The full study covers the rest of the 22 checks.

By Ronan Liedmeier, fractional head of growth. Last updated 23 October 2026.

What do the entity checks measure?

The entity checks measure whether a crawler can identify the company behind a website and describe it without guessing. There are five of them, worth 17 points of the score.

They look at the parts of a website that tell an engine which organization it is reading about: a machine-readable name and logo, links to the same company's profiles elsewhere, one name used everywhere, pages that say who the company is and how to reach it, and what it charges. Their job is to make sure the engine attaches what it reads to the right company. Making a page quotable is a separate job, covered by the content checks.

The profiles D2 counts are LinkedIn, Crunchbase, G2, Trustpilot, Clutch, Wikipedia, X, GitHub and YouTube. D5 is marked not applicable when a site does not appear to sell something with a list price, which applied to 225 sites, so pricing figures use a base of 1,361 commercial sites. Wikidata is part of the tool but scores no points in version 2, so it is not reported here.

How did software company websites do on the five checks?

Brand name is the best-scored check, with 68.8% of the 1,586 sites passing, and pricing the worst, with 22.7% of the 1,361 commercial sites passing. As a group, entity is the second-strongest of the four, with sites earning a mean 62.5% of the applicable entity points (median 64.7%), behind Access at 93.1%.

Across all 1,586 sites, the entity group accounts for 9,604 of the 51,121 points lost, or 18.8%. Organization schema loses the most of the five (3,214 points), then linked profiles (2,218), about and contact (1,601), pricing (1,574) and brand name (997). Answerability and evidence lose far more: the full breakdown is in the main study.

Organization schema splits the sample almost in half, with 43.1% complete and 44.5% having none at all.

Does Organization schema get a software company cited?

There is no evidence that it does. In my 209-vendor Perplexity test, Organization structured data had a within-answer correlation of 0.02 with being cited and sameAs links 0.07, and neither was significant.

Ahrefs found the same with a controlled design. They tracked 1,885 pages that added JSON-LD between August 2025 and March 2026 against matched pages that never did, and concluded that "Adding schema produced no major uplift in citations on any platform" across Google AI Overviews, AI Mode and ChatGPT (Ahrefs). Their pages were already cited before the change, so the test says less about sites engines have not yet picked up.

So the case for the markup is identification. Google's documentation says organization markup on the home page "can help Google better understand your organization's administrative details and disambiguate your organization in search results", and that some properties affect which logo appears in Search results and the knowledge panel (Google Search Central). The same page says "There are no required properties" and recommends placing the markup on the home page or a single page that describes the organization, such as the about page. My check asks for three properties, name, url and logo, because those are the minimum that tie a domain to a named company with a recognisable mark.

Sites with complete Organization markup score higher overall: a mean of 73.8, against 71.2 for incomplete markup and 52.3 for none. I read that as a marker of how much care went into the site as a whole. Sites whose homepage did not load for the checker also land in the no-markup group, which pulls its average down. The gap gives no reason to expect the markup itself to move anything.

US and European software company websites are level here: 43.5% of US sites and 42.9% of European sites have complete markup.

Why do linked profiles and one brand name matter?

They let an engine connect what other sites say about a company to that company's own domain. A buyer asking about a tool often gets an answer built from review sites, LinkedIn and directories, and the engine has to decide whether the company named on a review site is the same one that owns the domain it is reading.

Schema.org defines sameAs as the "URL of a reference Web page that unambiguously indicates the item's identity" (Schema.org). My D2 check accepts the links in sameAs markup or as plain links on the homepage, because a human reader and a crawler can both follow either.

Of the 1,586 software company websites, 758 (47.8%) link three or more profiles, 547 (34.5%) link one or two, and 281 (17.7%) link none. The sameAs result from the 209-vendor test applies here too: linking profiles has not shown a citation effect. It gives the engine a way to check that the profiles and the site belong to the same company.

Brand name consistency is the best-scored entity check. 1,091 sites (68.8%) use the same brand string in og:site_name, the Organization name and the title suffixes, 244 (15.4%) are partly consistent, and 251 (15.8%) are inconsistent or have no brand string to compare. The check marks a site down when the brand strings on different pages do not contain each other, for example when the title suffix carries an old product name the rest of the site no longer uses.

How many software companies have reachable about and contact pages?

Fewer than half have both. 711 of 1,586 software company websites (44.8%) had both an about page and a contact page the checker could reach, 512 (32.3%) had one, and 363 (22.9%) had neither.

An about page is where an engine finds the plain statement of what the company does, where it is based and who runs it. A contact page confirms the company is reachable and is the obvious place for an address. In the 209-vendor test, a reachable about and contact page was weakly associated with citation and did not reach significance, so I treat it as an identification fix with a possible small side effect rather than a way to get cited.

Some of the 363 will have these pages under paths or labels the check did not recognise, or behind a contact form loaded with JavaScript. If your site has them and the check says otherwise, a crawler following links is likely to have the same trouble. The JavaScript article covers that failure in more detail.

How many software companies state their prices?

Fewer than one in four. Of the 1,361 commercial software company websites, 309 (22.7%) have a pricing page that states amounts, 791 (58.1%) have a pricing page without amounts or show prices only somewhere else, and 261 (19.2%) have no pricing page and no prices.

Pricing is the one entity item that answers a buyer question directly. "How much does X cost" is a prompt buyers type, and when the vendor's own site gives no figure, the engine can only answer from what other sites say or report that the price is not public. A pricing page with amounts gives it a first-party figure to use. That is the reason I rank it first below, and it is still a readiness item: I have not measured whether stating prices changes how often a vendor is cited.

The partial group is the large one. It includes "Contact sales" pricing pages with plan names and no numbers, and sites that mention a price on a product page or in a blog post without a pricing page.

The US against Europe gap on pricing is real (z 4.2). FAQ sections and comparison tables show the same kind of gap, while question headings and Organization schema show none, covered in the content formats article. The overall score is about the same: a mean of 64.9 for US sites and 63.3 for European sites, a difference that is not statistically meaningful.

How was this measured?

I pulled software publishers (NAICS 51321 or 5112) with 11 to 200 employees, founded 2015 or later, from Apollo on 22 October 2026, across 13 countries: the US, UK, Ireland, Germany, Austria, Switzerland, the Netherlands, Belgium, Luxembourg, Denmark, Sweden, Norway and Finland. Agencies, consultancies, PR firms, online shops and staffing firms were excluded by NAICS code. 65 companies were dropped while pulling (no domain, mislabelled large brands, IT outsourcers, agencies, recruiters, one shop), and 32 were removed before the run: 12 I had already contacted and 20 very large, well-known brands. HQ country is whatever the Apollo filter returned; I did not check it per company.

Of 1,601 sites attempted, 1,586 were scored (1,583 complete, 3 partial scans that ran out of the 45-second time budget) and 15 failed. The run was one continuous batch on 22 October 2026, 18:03 to 18:28 Pacific time, using the same scoring code as the free readiness check, run locally on a residential connection. Every site got one check, and the median site had 9 URLs read.

Limits that apply to this article:

• A high score says nothing about how often a site is cited, and the entity checks in particular have shown no citation effect in the tests I know of.
• 105 homepages did not load for the checker, many of them because the site blocks automated requests in general. Those sites fail most entity checks for that reason, not because the pages are missing.
• D4 can miss about and contact pages under unusual paths. D5's "commercial" filter is the tool's judgment of whether a site sells something with a list price, and 225 sites were marked not applicable.
• The sample is software companies with 11 to 200 staff. The results should not be read as true for other kinds of websites.

What I'd do

For a software company site, I would start with pricing and finish with schema:

1. Publish a pricing page with amounts. Only 22.7% of the 1,361 commercial sites in this sample do it. If list prices are impossible, give a starting price or a price per seat with what changes it.
2. Make the about and contact pages reachable from the main navigation as plain links. The about page should say in its first paragraph what the company does, for whom, where it is based and who runs it.
3. Use one brand name everywhere: the title suffix, og:site_name, the Organization name, the LinkedIn page and the review profiles. Drop the legal entity suffix from page titles unless that is the name buyers use.
4. Link three or more profiles from the homepage footer, starting with LinkedIn and the review site buyers in your category actually read. That lets an engine match third-party mentions to your domain.
5. Add Organization JSON-LD last, on the homepage, with name, url, logo and sameAs pointing at the same profiles. It helps Google identify the company and pick the logo. I put it last because it is the item most often sold as a citation fix, and the evidence from my own test and from Ahrefs says it is not one.

None of these five replaces the checks that did predict citation within an answer in the 209-vendor test: FAQ sections, named authors with an about page, article depth, comparison tables and visible dates. The hub guide on getting cited covers the rest.

To see whether engines can identify your company and what they say about it, start with the free AI search readiness check, which runs the same five entity checks on your site. The AI Visibility Snapshot goes further: ten prompts in your category, your site scored on the 22 checks, and a one-page readout in three working days. The readout ends with the order I would fix things in, the same way this article does; fixing them in that order is the work I do as a fractional head of growth.

Re-check by 23 January 2027: the pricing pass rate and the US against Europe gap on a fresh pull, whether the about and contact check needs to recognise more paths, and whether Wikidata is scored in the next version of the checker.