How Do I Decide Whether an AI Visibility Gap Needs Schema, Copy, or Digital PR?
Decide by diagnosing the gap from the answers and the sources they cite. If assistants get basic facts wrong that your pages already state, you have a readability problem: blocked crawlers, script-rendered content, or facts buried in PDFs. Schema rarely fixes that on its own. If none of your pages answers the question buyers ask, you need copy: a page that answers it plainly. If your page does answer it but assistants cite review sites, directories, "best of" lists and articles instead, you need third-party evidence. That means reviews, directory profiles and digital PR. The evidence on schema is modest. A large Ahrefs study found adding structured data made no measurable difference to citations in ChatGPT or Google's AI Mode. Most gaps are copy or evidence gaps.
How do you diagnose the gap?
Run the questions where you are missing or misdescribed, several times each, in the assistants that matter to you. For each one, look at the answer and every cited source, then use this table:
| What you see | What it means | Likely fix | How to confirm |
|---|---|---|---|
| Wrong basic facts that your site states correctly | Assistants cannot read or trust your pages | Technical: crawl access, plain HTML text, clear structure; schema as support | Check that AI crawlers can fetch the page and see the text without scripts |
| You are absent, and no page of yours answers the question | Nothing to cite | Copy: a page that answers the question directly | Search your site for the question; if you cannot find an answer quickly, neither can an assistant |
| You are absent, your page answers it, and competitors' pages are cited | Your page is weaker for that question | Copy: more specific, evidence-rich content | Compare your page with the cited ones side by side |
| You are absent, and review sites, directories or lists are cited | Buyers' sources do not include you | Evidence: reviews, directory profiles, list inclusion, PR | Check whether you appear on the cited pages at all |
| You are mentioned but described negatively or outdatedly | Third-party sources carry old or negative information | Evidence plus correction requests | Find the sentence in the cited source |
| You appear for branded questions only | Assistants know you exist but do not associate you with the category | Copy that ties your name to the category, plus third-party mentions in that category | Ask "What is [brand]?" and check whether the category is mentioned |
What does the evidence say about schema?
Structured data helps search engines understand a page, and Google uses some of it for rich results. As a lever for AI visibility, the evidence is weak:
- Ahrefs' 2026 study compared 1,885 pages that added structured data with matched pages that did not. It found changes in ChatGPT and AI Mode citations of about 2%, statistically indistinguishable from zero. AI Overview citations fell slightly relative to the controls.
- SearchVIU's tests found that ChatGPT, Claude, Perplexity and Google's AI features read the visible text of a page, not its JSON-LD, when they fetched it directly.
So use schema to describe what the page already says clearly, such as organisation details, products and prices. Do not expect it to close a visibility gap. When a gap looks technical, check crawl access and whether key text is in the HTML first. How structured data fits AI search is covered in how to use structured data and schema for AI search visibility.
When is copy the answer?
When the question has no clear answer on your site, or your answer is vaguer than the pages that win. Good copy for AI answers:
- Answers the question in the first lines, in words buyers use.
- Is specific: prices, limits, who it suits and who it does not, timelines and steps.
- Has evidence: numbers, examples and named customers where allowed.
- Covers the questions around the main one, such as comparisons, alternatives and objections.
Copy is usually the cheapest fix and the one you control. It is also a prerequisite for the others. PR and reviews send people and assistants to your site, which then needs to answer them. Why clarity beats volume is covered in whether clarity and trust matter more than publishing volume for AI visibility.
When is digital PR or third-party evidence the answer?
When assistants answer from sources other than company websites, which is common for "best" and comparison questions. Muck Rack's analysis of more than 25 million links in AI answers found earned media made up 84% of citations. Its definition is broad, including journalism, academic and government sources, reference sites, social platforms and third-party company content. G2's 2026 survey of more than 1,000 B2B software buyers found review sites (38%) and AI chatbots (37%) were the two top sources shaping shortlists. For local businesses, BrightLocal found Google Business Profiles supplied 28.5% of local AI citations, with Yelp next at 8.53%.
Third-party evidence includes:
- Reviews on the platforms cited in your category.
- Directory and marketplace profiles, complete and consistent.
- Inclusion in the lists and comparisons assistants cite, through outreach, briefings or product access.
- Coverage in trade and news media, where your expertise or data is newsworthy.
How to measure whether PR moved anything is covered in how to measure the impact of PR on AI visibility.
Should I pay for PR, reviews, directories, or content first for AI visibility?
Usually in this order, adjusted by what your diagnosis shows:
- Fix readability and copy on your own site. It is cheap, fully in your control, and every other channel depends on it.
- Complete free or low-cost profiles on the directories and review platforms that assistants cite in your category, starting with your Business Profile if you serve local customers.
- Start a steady review programme, asking every customer for an honest review in their own words on the platforms that are cited, except Yelp, which does not allow businesses to ask.
- Pay for digital PR when the questions you care about are answered mainly from third-party lists and articles, and you have something newsworthy: data, expertise or a genuine story.
Business type shifts the weights:
| Business type | Usually first | Usually later |
|---|---|---|
| Local services | Business Profile, reviews, key directories | PR |
| B2B software | Copy, review-platform profiles and reviews | Analyst and trade coverage |
| Ecommerce | Product pages and feeds, retailer listings, product reviews | Gift guides and roundup outreach |
| Professional services | Expertise pages, directory and professional listings | Trade media and speaking |
If an agency offers an "AI visibility package", ask which gap each item addresses, and how you will know whether it worked.
What might an illustrative case look like?
Consider an illustrative commercial cleaning company missing from "best office cleaning company in [city]" answers. Its agency proposed schema markup and a blog. The diagnosis showed the company's service pages were clear and correct. The answers cited a local business directory, two review platforms and a "top cleaning companies" list from a local business magazine, none of which featured the company. The budget went to completing directory profiles, asking every existing client for an honest review, and pitching the magazine an article on post-pandemic office hygiene backed by the company's client data. Three months later, the company appeared in about a third of answers, cited through the directory and the magazine.
What do people ask about choosing AI visibility fixes?
Is schema useless for AI?
No, but it is supporting, not decisive. Use it to describe clear, correct pages, not to fix gaps.
Can one fix do everything?
Rarely. Most brands need clear copy first and third-party evidence second.
How long before I know whether the fix worked?
Copy and profile changes can show up in search-based answers within weeks. PR and reviews take longer to build up, often a quarter or more.
Published by The Kovmere Team · updated Sep 29, 2026 · More resources