Getting Named When Buyers Ask an AI Assistant

What This Work Involves

A search results page gives a buyer ten links and lets them choose. A generated answer gives them three or four names and a short reason for each, so the shortlist is built before your website is ever opened. Models decide who to name on what they can find that reads as a direct, verifiable answer, and on how consistently independent sources describe you.

The Nine Step Method

• 1. Pick the questions you want to win Write down the ten commercial questions a buyer asks before they ever type your company name.
• 2. Baseline where you stand today Ask each question in a clean session across the assistants your buyers use, and record who is named and which pages are cited.
• 3. Build one answer shaped page per question cluster An H1 that restates the question and an opening paragraph that answers it in full before any narrative.
• 4. Make the facts liftable Comparisons in tables, numbers next to their source, a visible date and one claim per sentence.
• 5. Fix your entity record Identical name, location, service description and founder details in schema, on site, in directories and in press coverage.
• 6. Seed the sources the answers already cite Pursue mentions on the exact directories, roundups and publications your baseline shows assistants trusting.
• 7. Publish something only you can publish Original data, a benchmark or a named method gives other people a reason to cite you.
• 8. Let the crawlers in and tell them what matters Allow the AI crawlers you want, serve full content in the initial HTML, and publish an llms.txt index.
• 9. Re-run the baseline and log what moved Measure the same questions on the same schedule so the next quarter of work is evidence led.

How to Write an Answer Shaped Page

• An H1 that restates the buyer question in the buyer's own words
• An opening summary of roughly 60 to 90 words that answers the question completely
• H2 headings that are questions or clear statements of fact, one idea each
• A table wherever the buyer is comparing options
• Figures with the source named in the same sentence, and a visible date
• An FAQ section backed by FAQPage structured data
• Internal links to the pages a buyer would read next

Where a B2B Scale Up Should Show Up

Plenty of advice tells you to be everywhere, and a company of thirty or forty people cannot be. The shortlist that earns its place for a B2B business with a considered sale is your own site first, then software review platforms such as G2, Capterra and Clutch, trade and industry publications, aligned founder and executive profiles, podcasts and recorded interviews with published transcripts, analyst notes and category roundups, and the niche communities where your buyers actually ask practical questions. Encyclopaedia entries only exist for companies that already meet an independent notability bar, and consumer social platforms rarely decide a B2B shortlist, so treat those as distribution rather than seeding.

A First Ninety Days for a Small Team

• Weeks one to three, baseline and scoping. Agree the ten buyer questions with sales, run the first visibility check and log every cited URL, then audit your entity record for anything that disagrees with itself.
• Weeks four to eight, build the answer layer. Publish or rewrite four to six answer shaped pages, add FAQ structured data, and serve full content in the initial HTML response.
• Weeks six to ten, earn corroboration. Work the shortlist of cited domains, update review profiles, pitch two contributed pieces and record one interview.
• Weeks nine to twelve, publish something original and re-measure. Release data only you can publish, then re-run the baseline and record what moved against what changed.

Who Owns This Inside the Company

• Marketing lead or fractional CMO. Owns the question list, the priority order and the monthly measurement.
• Content or product marketing. Writes the answer shaped pages and keeps claims sourced and dated.
• Whoever owns the website. Handles crawlability, structured data, the initial HTML response and llms.txt.
• Founder or subject expert. Supplies the original view, the data and the interviews nobody can outsource.
• Sales. Feeds back the real questions buyers ask and reports when a lead says an assistant named you.

Where This Work Goes Wrong

Writing for the keyword instead of the question, burying the answer under a story, leaving claims undated and unsourced, chasing volume of content, and treating AI visibility as separate from demand generation.

How to Measure It

Ask your ten questions in a fresh session with no history, then record which providers were named, which URLs were cited and whether you appeared at all. The cited URLs are the most useful column, because that list is your outreach plan.

Common Questions Answered

What is LLM seeding?

LLM seeding is the practice of putting clear, well structured, well sourced information about your brand into the places large language models read, so that assistants such as ChatGPT, Gemini, Claude and Perplexity name you when a buyer asks a question in your category.

How is LLM seeding different from SEO?

Traditional SEO competes for a ranked list of links, while LLM seeding competes to be inside a single generated answer where only a handful of providers get named. Answers reward one liftable paragraph, verifiable facts and agreement across independent sources rather than volume and link count.

How long does LLM seeding take to show results?

New pages can appear in answers within a few weeks once they are crawled. Movement that comes from third party mentions and entity consistency usually takes a quarter or more, since it depends on other people publishing.

Can you pay to be included in AI answers?

No. There is no paid placement inside an organic generated answer, so inclusion comes from being the clearest and best corroborated source for the question.

How do I check whether AI assistants mention my business?

Ask your buyer questions in a fresh session with no prior chat history and no personalisation, then record which brands get named and which URLs are cited, and repeat on a fixed schedule.

Does an llms.txt file help?

It helps as an index rather than as a ranking factor. An llms.txt that links your key pages with a line of description each makes your important content easy to find and read in full.