What Should I Do When AI Engines Recommend Nobody for My Highest-Value Questions?
When AI engines recommend nobody for your highest-value questions, first find out why nobody is named, because the cause decides what to do. Assistants may be cautious for safety or professional reasons, as with legal, medical or financial questions. They may have too little reliable information about providers, or the question may need clarifying first. They may give criteria instead of names. Or buyers may use words that providers never use. Then act on the cause. Publish pages that meet the criteria assistants list, and become the clearest source for the exact question, in the buyer's own language. Get listed in the directories and registers assistants send people to. An empty answer is also an opportunity. With no incumbent, a well-documented provider has a real chance of being named.
Why do assistants recommend nobody?
| Cause | What the answer looks like | What it tells you |
|---|---|---|
| Caution | "I can't recommend a specific adviser, but here is how to choose one." | The topic is sensitive; names are withheld on purpose |
| Thin evidence | "I don't have reliable information about providers in [area]." | Too few sources connect providers to this need |
| Criteria instead of names | "Look for a certified installer with at least five years' experience." | Assistants know what matters but cannot match providers to it |
| Clarification | "Which city are you in, and what's your budget?" | The recommendation comes after the follow-up |
| Vocabulary mismatch | A generic answer that misses the specialist service | Buyers and providers describe the service differently |
SparkToro's research on AI recommendations found that the length of recommendation lists tracks how many entities a topic contains. In narrow or new topics, where few providers are documented, the lists are correspondingly short. The distinction between refusals and other no-name answers is covered in what refusal means in AI search and how it differs from a no-recommendation.
What should you do for each cause?
Criteria answers. Take the criteria the assistant lists and show, on your own pages, that you meet each one: certifications with numbers, years of experience, guarantees, prices, turnaround times. Make each criterion a sentence an assistant could quote.
Thin evidence. Become the clearest source for the exact question:
- Publish a page that answers it in the buyer's words, with specifics about your service, area and customers.
- Get listed in the directories, marketplaces and associations that cover your niche.
- Earn a few independent mentions, such as local coverage, partner pages or an industry list, that connect your name to the need.
How to target questions that assistants answer poorly is covered in how to target questions AI answers poorly or where your site is the primary source.
Caution. Assistants may never name individual providers for some questions. Be the factual source they draw on for the educational part of the answer. Make sure you appear correctly in the registers and directories they point people to, such as licensing lookups and professional directories.
Clarification. Test the follow-up. Give the location, budget or need the assistant asked for, and see who it names then. Optimise for that second answer.
Vocabulary mismatch. Find the words buyers use in calls, emails and searches, and use them in page titles, headings and first lines, alongside your trade terms.
How do you find which questions matter?
Start with the questions closest to revenue: those your best customers asked just before buying. Pull them from sales calls, enquiry forms and support tickets. Test each in ChatGPT, Gemini, Perplexity and Google's AI features several times, and classify the outcome. Prioritise questions where nobody is named and the cause is fixable: criteria, thin evidence or vocabulary. Those are where you can become the default before competitors do. How to turn customer conversations into test questions is covered in how to turn customer interviews, sales calls and support tickets into AI search prompts.
How do you measure progress?
- No-name rate: the share of answers to each question that name nobody. It falls as providers, including you, become documented.
- First appearance: the date you first appear in answers to the question, and in which assistants.
- Share when named: once names appear, how often you are among them.
- Competitor entry: whether other providers start appearing, which shows the window is closing.
Track these monthly on a fixed question set, with several runs per question.
What should you avoid?
- Inventing evidence. Unsupported claims about certifications, awards or results can backfire when assistants or customers check them.
- Mass-producing thin pages for every variation of the question, which search engines treat as scaled content abuse.
- Assuming the answer will stay empty. Competitors see the same gap. Move while it is open.
What might an illustrative case look like?
Consider an illustrative company that restores listed-building windows, a specialist service that sells for high prices. Assistants answered "who can repair sash windows in a listed building in [county]?" with advice on checking conservation experience, and named nobody. The company's site said "window repairs" in general terms. It published a page titled "Sash window repair for listed buildings in [county]". The page listed its conservation accreditations, the local authorities it had worked with, typical costs and before-and-after case studies. It also joined a heritage trades directory. Two months later, assistants named the company in most answers to the question, as one of two providers, and enquiries for listed-building work rose.
What do people ask about empty AI answers?
Is an empty answer good or bad?
Both. It means buyers get no help, but it also means no competitor has claimed the space yet.
Will assistants ever name providers for legal or medical questions?
Sometimes, depending on the question and the assistant. Where they will not, focus on the sources they point people to.
How quickly can we become the named provider?
It depends on how quickly the sources assistants use pick up your pages and listings. For narrow questions, search-based answers can change within weeks, but there is no fixed timeline, and broader questions take longer.
Published by The Kovmere Team · updated Sep 29, 2026 · More resources