November 18, 2026Last updated: November 18, 20269 min read15 sections
13 Tactics to Take Your AI Search Campaigns to the Next Level
By Tom Hunt, Founder & CEO, Fame
Tom founded Fame in 2019 and leads the company today, including Fame Search, its AI search agency for B2B companies. LinkedIn
Key Takeaways
- Start with crawl access and weekly tracking, because nothing else can be cited or measured without them.
- Pages for follow-up questions, fair comparison pages and consistent company facts are fully in your control.
- Third-party lists, review profiles and original data take longer and depend on other people.
- llms.txt, visible refresh dates and FAQ schema are cheap housekeeping with the weakest evidence.
- No tactic guarantees a citation, so ask every tracked question several times and read trends over weeks.
Most AI search campaigns stall after the first batch of answer pages, because the team runs out of obvious things to do. These 13 tactics are what we would add next. For each one I explain what it is, why it tends to work and what you can do this week, and I say so where the evidence is thin. They are the same moves we make as an AI search agency, and none of them needs us.
One warning before the list. Nobody outside the AI companies knows exactly how the engines choose sources, and most published research comes from vendors with something to sell. Each tactic below improves your odds and guarantees nothing. For the wider picture, start with the complete guide to B2B AI search.
1. Build pages for the follow-up questions engines generate
What it is. When a buyer asks an AI engine a question, the engine often runs several searches of its own behind the scenes. Google calls this query fan-out: breaking a question into subtopics and issuing many queries at once. Your page can be retrieved for one of those sub-questions even if it never ranks for the original.
Why it tends to work. Ahrefs' analysis of 15 million fan-out queries found "price" rose from 6.04% to 8.88% of fan-outs in a month, and "official" from 10.1% to 13.1%. Ahrefs warns that a model update could shift these patterns. Still, if an engine goes looking for prices and official information, a site that hides both gives it little to work with.
This week. Take your five most important buyer questions and write down what a careful buyer would ask next: what it costs, who it suits, how long it takes, what the alternatives are. Check whether a page on your site answers each one in its first lines. Publish the missing ones, starting with pricing.
2. Publish comparison and alternatives pages
What it is. A page that sets you against a named competitor, and a page for buyers who are leaving one.
Why it tends to work. Buyers ask engines head-to-head questions. If you have not published a fair comparison, the answer gets assembled from pages written by other people, including the competitor. Keep it fair, because readers and machines can both tell when a table has been rigged.
This week. Pick the competitor you lose the most deals to. Write one page using facts from their own public site, note the date you checked them, and say plainly where the other option is the better fit.
3. Do honest outreach to third-party lists
What it is. Getting your company onto the "best of" lists that engines already cite for your category, by asking the editor and giving them a reason.
Why it tends to work. In Ahrefs' study of 750 prompts, "Best X" lists were 43.8% of cited page types, and brands higher on third-party lists were more likely to be recommended. The prompts were written by the author and the result is a correlation, so treat it as a pointer.
This week. Export the third-party URLs cited in your tracked answers. Choose the lists where you honestly fit. Send each editor something they can verify, such as corrected pricing or a customer proof point. Do not pay for placement on low-quality sites.
4. Put full episodes and corrected transcripts on YouTube
What it is. Uploading long-form video with accurate captions, a real description and chapters, and pairing it with a page on your own site.
Why it tends to work. The data here conflicts, and I would not rebuild a content program around one study. What we can say is that long-form videos took 94% of YouTube citations in OtterlyAI's study, against 5.7% for Shorts, and description length had the strongest correlation at 0.31, which is weak to moderate. We go through the disagreements in our piece on Gemini and YouTube.
This week. Take your best existing video. Correct the names and product terms in the captions, write a useful description, add chapters, and publish the transcript on your site so you have two chances at a citation.
5. Guest on podcasts that publish text
What it is. Appearing on other people's shows, chosen partly for whether they publish transcripts, show notes and full YouTube episodes.
Why it tends to work. Ahrefs, analyzing 75,000 brands, found branded web mentions correlated with AI visibility at 0.656 to 0.709. Correlation is not proof, and we found no independent test showing guesting causes citations. So judge a show first on whether it reaches your buyers, and treat any AI benefit as a bonus. The full argument is in podcast guesting as an AI search strategy.
This week. Run your buyer questions through the engines and note any podcasts or interviews that get cited. Pitch one of those shows.
6. Write an llms.txt file, then move on
What it is. A proposed Markdown file at your site's root that gives AI agents a short guide to your most important pages.
Why it might help. I am including it because people ask, not because it is proven. No strong evidence shows it earns citations, and Google has said it is unnecessary for its AI features. The most plausible use today is an agent already on your site trying to complete a task. Our llms.txt guide for B2B websites has the evidence.
This week. Give it an hour at most. List your name, a one-sentence summary, core services, pricing and key proof pages, then go back to tactics with stronger support. Be wary of anyone charging separately for it.
7. Set a refresh cadence with visible dates
What it is. A fixed schedule for reviewing your important pages, with the date of the last real update shown on the page.
Why it tends to work. Prices, product details and "best of" answers go out of date, and an engine that retrieves a stale page may repeat a stale fact about you. I cannot point you to a clean B2B study on how much freshness matters, so treat this one as good housekeeping with a probable upside. We refresh pages every 90 days.
This week. List your 20 most important pages with the date each was last reviewed. Show an "Updated" date on each. Change that date only when the content really changed, since Google's guidance warns against changing dates to look fresh.
8. Keep your company facts consistent everywhere
What it is. Making sure your name, description, category, pricing and audience read the same on your site, your LinkedIn page, directories and review profiles.
Why it tends to work. If your pricing page, your LinkedIn profile and a directory listing all disagree, you make the engine's job harder and any quoted answer less reliable.
This week. Write a five-line fact sheet: company name, one-line description, category, price and who it is for. Check it against every profile you control and fix what differs.
9. Claim and correct your review site profiles
What it is. Owning your listings on the directories and review sites that appear in answers for your category.
Why it tends to work. Ahrefs' fan-out study found ChatGPT increasingly searching for prices, reviews and official information. Analyze AI reported a 60.2% mention rate on answers citing a directory page, against 36.3% without, though that comparison is unadjusted and comes from a vendor.
This week. Find which review and directory pages are cited in your tracked answers. Claim those profiles, correct the description and pricing, and ask customers for honest reviews where that platform's rules allow.
10. Publish original data
What it is. A number only your company has, published with its method, sample size and date.
Why it tends to work. In the GEO paper, adding quotations, statistics and cited sources produced improvements of roughly 30-40% on its main visibility measure. Its main test used a custom engine built on GPT-3.5, so the numbers do not transfer directly to today's engines, but the direction is a sensible default.
This week. Find one figure in your product data, customer survey or sales records that a buyer would want. State the number, its source and its date together in one sentence that stands alone. Never invent a number to sound quotable.
11. Add FAQ sections, with schema as a cheap extra
What it is. Question-led headings with a direct answer underneath, plus FAQPage structured data that matches the visible text.
Why it tends to work. The visible questions and answers are the part with a case behind them, because they match how buyers phrase things. The markup is less certain. Ahrefs tracked 1,885 pages adding JSON-LD and found a ChatGPT effect that could not be told apart from zero, and Google says no special schema exists for its AI features. So add it where your CMS makes it nearly free, and do not sell it internally as a citation tactic.
This week. Add three to five real buyer questions to each of your top pages, each answered in two or three plain sentences. If you add schema, make sure it says exactly what the page says.
12. Let AI crawlers in
What it is. Checking that the bots behind AI search products can actually reach your pages.
Why it works. This is the one factor the engines document directly. OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, and Perplexity documents PerplexityBot in the same way. Training crawlers such as GPTBot are a separate setting, and a separate policy call.
This week. Open yoursite.com/robots.txt and read it. Then ask whoever runs your CDN or firewall whether bot rules block these visitors, because a clean robots file does not help if a firewall rejects the request. For the other engines, check each company's own crawler documentation.
13. Track weekly, and ask every question more than once
What it is. A fixed panel of buyer questions, asked several times a week in each engine, with mentions, citations and competitors logged.
Why it works. One answer is a sample. In Ahrefs' stability study, 59% of ChatGPT citations were gone by day two, and about 75% of citations that dropped out came back within the 20 days. Without repeated runs you cannot tell whether any of the 12 tactics above did anything.
This week. Fix a panel of 20 questions, most of them unbranded. Run each three times per engine in a fresh chat and log the results in a spreadsheet. The routine is in how to track AI visibility.
Where should you start?
Do the two that make the rest measurable and possible, then work down by evidence. This is my own ordering, and your tracked answers should overrule it.
| Order | Tactics | Why here |
|---|---|---|
| First | 12 and 13: crawl access and weekly tracking | Nothing else can be cited or measured without them |
| Next | 1, 2 and 8: follow-up pages, comparisons, consistent facts | On your own site, fully in your control |
| Then | 3, 9 and 10: lists, review sites, original data | Off-site sources take longer and depend on other people |
| When it fits | 4 and 5: YouTube and guesting | Worth it if you already make video or the shows reach your buyers |
| Housekeeping | 6, 7 and 11: llms.txt, refresh dates, FAQ schema | Cheap, but the weakest evidence |
The work is consistent more than it is clever. Campaigns improve when someone owns the list every week.
More on B2B AI search
- B2B AI Search: The Complete Guide (2026)
- The 13 Best AI Search Marketing Agencies for B2B (2026)
- What to Look For When Hiring an AI Search Agency (12 Questions to Ask)
- Your Buyers' Shortlist Is Written Before They Visit Your Site
- Why Most AI Visibility Dashboards Measure the Wrong Thing
- Third-Party Lists Are the New Backlinks
- What We Learned Running AI Search on Our Own Site for 150 Days
Want your company named in AI answers? Fame Search is our AI search agency for B2B companies: weekly tracking across four AI engines and one action on your site every day, for $2,000 a month.
Frequently Asked Questions
Check that AI search crawlers can reach your site, then set up weekly tracking with repeated runs. Without those two, nothing else can be cited or measured.
The evidence is weak. Ahrefs tracked 1,885 pages adding JSON-LD and found a ChatGPT effect that could not be told apart from zero. Write the visible questions and answers for buyers, and add the markup only where it is nearly free.
Give it an hour at most. No strong evidence shows it earns citations, and Google has said it is unnecessary for its AI features.
Track a fixed panel of buyer questions every week, asking each one several times per engine, and read trends over several weeks. A single answer is a sample and can show a win or a loss by chance.
Want us to run this for you?
Fame starts and grows B2B podcasts that generate pipeline. Tell us who you are trying to reach and we will send a costed proposal within 1 working day.
Related content
- Read moreNovember 18, 2026
B2B AI Search: The Complete Guide (2026)
B2B AI search is the way business buyers use ChatGPT, Perplexity, Claude, Gemini and Google's AI features to research problems and shortlist vendors, and the work of getting your… - Read moreNovember 18, 2026
The 13 Best AI Search Marketing Agencies for B2B (2026)
The best AI search marketing agencies for B2B in 2026 are Fame, FirstIndex, Arobis AI, RevenueZen, Optimist, DerivateX, TripleDart, LoudFace, Discovered Labs, Omniscient Digital,… - Read moreNovember 18, 2026
Your Buyers' Shortlist Is Written Before They Visit Your Site
More B2B buyers now appear to ask an AI assistant who can solve their problem before they open a single vendor website. - Read moreNovember 18, 2026
Why Most AI Visibility Dashboards Measure the Wrong Thing
Most AI visibility dashboards report a single score built from one run of each prompt, counted in mentions.