How can I make my business more visible in AI search?
Most writing on this subject is technical. This one is about the decision: whether it is worth funding, who should run it, and what a realistic first year looks like.
Most writing about AI search visibility is technical, which is useful for the person doing the work and useless for the person deciding whether to fund it. This is the other view: what it is worth, who should own it, what it costs, and how to sequence it against everything else competing for the same budget.
What it is actually worth
The honest answer is that it varies enormously by business, and the industry discussion is unhelpfully averaged.
The scale is not speculative. Google reported 2.5 billion monthly active users for AI Overviews and 1 billion for AI Mode in May 2026, and AI Overviews appear on roughly a third of monitored queries. But scale is not the same as relevance to your funnel.
Three questions determine whether this matters to you.
Does your buyer research before choosing? If purchases in your category involve comparison, shortlisting or asking around, assistants are now part of that process and being absent from it costs you entry to the funnel. If your category is impulse, local, or entirely brand-driven, the effect is smaller.
How much of your traffic is informational? Informational queries attract AI answers far more often than transactional ones. A business living on how-to and comparison traffic is structurally exposed; a business living on branded and transactional traffic much less so.
Is your category one where a shortlist gets made? When an assistant answers “what’s the best X for Y” by naming three companies, the shortlist is formed before anyone visits a website. If that is how buying starts in your market, presence on the shortlist is not a marketing channel — it is qualification.
Working out your exposure
Twenty minutes gives you a defensible answer, and it is worth having before the budget conversation rather than after.
- Write fifteen questions a real buyer would ask before choosing a supplier in your category. Full questions, not keywords.
- Ask each of them to ChatGPT and Perplexity. Record whether you are named, who is named instead, and which sources are cited.
- Count. Named in zero of fifteen is a finding. Named in twelve is a different conversation entirely.
- Check the citations. If the cited sources are review sites and forums rather than company websites, that tells you where the work is.
This exercise is uncomfortable and it is the single most persuasive artefact available for an internal case. A screenshot of an assistant confidently recommending three competitors does more than any market-size statistic.
Who should own it
Whoever owns organic search. The foundations are shared — retrieval depends on the same signals as ranking — and splitting ownership produces two teams doing contradictory work on the same pages.
What changes is that the owner now needs two things they may not have had.
Occasional engineering access. The highest-value fixes are rendering and crawler configuration, and neither is a marketing task. A team that cannot get a rendering change prioritised will be stuck at the first gate indefinitely.
A route into communications or PR. The corroboration half of the work — being accurately described by third-party sources — sits outside a conventional SEO remit and is the highest-ceiling lever available.
What does not work is creating a separate AI visibility function. The overlap with SEO is around eighty percent, and the duplication produces two backlogs competing for the same engineering time.
Get the artefact for your business case
A free scan produces exactly this: the real answers assistants give in your category, who they name, and what they cite.
Run my free scanWhat it costs
| Item | Typical cost | Notes |
|---|---|---|
| Baseline measurement | Free to low | Free scans exist; manual testing costs an afternoon |
| Rendering and crawler fixes | Days of engineering | One-time; frequently the entire problem |
| Tooling | ~$29–$99/mo, up to enterprise | Monitors are cheap; platforms that audit and produce cost more |
| Editorial restructuring | The largest real cost | Usually underestimated relative to tooling |
| Corroboration work | Ongoing, PR-shaped | Highest ceiling, longest timeline |
The pattern most teams get wrong is spending heavily on tooling and lightly on capacity to act on it. A dashboard nobody has time to respond to is a cost rather than an asset, and this category is unusually prone to it because the dashboards are compelling.
A realistic first year
Month 1. Baseline twenty prompts across three engines. Fix rendering and crawler access. These two things frequently produce the largest single improvement available, because they move you from invisible to visible.
Months 2–3. Write one plain sentence stating your category, audience and alternatives, and make it consistent everywhere. Restructure your ten most important pages answer-first. Publish one piece of original data.
Months 4–6. Work the citation map — get accurately described on the specific third-party domains the assistants already cite in your category. This is slow, unglamorous and where the durable gains are.
Months 7–12. Expand the prompt set as you win the narrow questions. Maintain the original data. Keep scanning weekly and reading trends rather than single runs.
What is not realistic in year one: being named on broad category questions dominated by market leaders, or a step change in ChatGPT specifically, which lags the other engines by the widest margin.
The non-technical levers
Three of the strongest inputs are not marketing activities at all, which is why they are frequently missed.
Being describable. If your own staff describe what you do three different ways, the web will too, and a model has nothing stable to state. Agreeing one sentence — category, audience, alternatives — and using it everywhere is an internal alignment exercise with an unusually direct external payoff.
Being reviewed. Independent reviews and comparison listings are cited heavily. Asking satisfied customers to review you where your category actually gets reviewed is cheap and compounds.
Being present where your buyers talk. Forum threads, particularly Reddit, are cited disproportionately often. Genuine participation — answering questions in your area of expertise without pitching — builds the footprint models read. Our guide to Reddit visibility covers doing this without being counterproductive.
Three objections worth answering
“This is a fad — the traffic is not there.” The traffic genuinely is smaller than the discussion implies, and that is not the argument. The argument is shortlist entry. When an assistant answers “what’s the best X” by naming three companies, the discovery step has happened and most of the market has been eliminated before anyone visited a website. That is a qualification event rather than a traffic event, and it does not show up in a traffic model at all.
“We cannot measure the return.” Partly true and improving. Referral traffic from assistants is increasingly visible as its own analytics channel, branded search tends to rise when assistants begin naming you, and sales teams can usually report whether prospects arrive having already shortlisted. None of that is a clean attribution model, and the honest position is that this is measured the way brand marketing has always been measured rather than the way paid acquisition is.
“We should wait until it settles.” The surface has largely settled — AI Overview presence has held in a narrow 31–34% band through 2026, which indicates systematic deployment rather than experimentation. What has not settled is who gets cited inside it, and that is the part you can influence. Waiting means letting competitors accumulate the corroboration that is slowest to build.
What the team actually does differently
The practical changes to a marketing team’s week are smaller than the strategic framing suggests, which is worth saying because it lowers the perceived cost of starting.
The content brief changes. Every brief gains two fields: the specific question this page answers completely, and the one thing it will contain that nobody else has. That is a fifteen-minute addition to a process that already exists.
The weekly review gains one chart. Mention rate across the prompt set, by engine, against two named competitors. Fifteen minutes, alongside the existing traffic review.
One recurring engineering ticket appears. Confirming that new templates and components render server-side. Once this is established as a definition-of-done item, it stops being a recurring problem.
PR gains a target list. The specific third-party domains assistants cite in your category, which is a considerably more focused brief than “get coverage”.
When to wait
Three situations where this is not your next investment, and saying so is more useful than selling the work.
Your organic search is weak. Retrieval depends on being indexed and relevant. A site with no organic presence is not in the candidate set, so the fix is foundational SEO first. AI visibility work on that base is premature.
Your buyers do not research. If purchases in your category are local, urgent, or driven entirely by existing relationships, the assistant is not in the path. Measure once to confirm, then deprioritise honestly.
You cannot act on findings. If your engineering backlog means a rendering fix takes nine months, buying a monitoring tool produces a year of watching a number you cannot change. Fix the capacity problem first.
Everyone else should at minimum baseline it. The measurement is cheap, and the most common outcome is discovering something specific and fixable that has been quietly costing you shortlist entries for a year.
Sources
- Google — 2.5 billion monthly active users for AI Overviews and 1 billion for AI Mode, stated 19 May 2026.
- SERP feature tracking, 2026 — AI Overview presence averaging ~32.5% of monitored queries within a 31.02%–34.40% band.
- Vercel and MERJ — analysis of 500M+ GPTBot fetches finding zero JavaScript execution; same behaviour for ClaudeBot and PerplexityBot.
Frequently asked questions
How can I make my business more visible in AI search?
Start by measuring whether assistants name you for your category’s buying questions. Then fix the mechanical barriers — content that only renders after JavaScript, blocked AI crawlers — which are usually the whole problem and take days. Then work on being described consistently by the third-party sources those assistants already cite, which is slower and has the highest ceiling.
Is AI search visibility worth investing in?
It depends on whether your buyers use assistants for research and how much of your funnel starts with discovery. Google reported 2.5 billion monthly active users for AI Overviews and 1 billion for AI Mode in May 2026, so the audience is not speculative. The strongest case is for businesses whose buyers currently arrive via informational search, which is exactly the traffic AI answers absorb.
Who should own AI search visibility in a company?
Whoever owns organic search, because the foundations are shared and split ownership produces contradictory work. What changes is that they now need occasional engineering support for rendering and crawler access, and a route into PR or communications for the third-party corroboration half.
How much does AI visibility work cost?
The mechanical fixes usually cost days of engineering time. Tooling ranges from free scans to around $99 a month for platforms with auditing and content production, up to enterprise contracts in five figures annually. The largest real cost is editorial capacity to restructure existing pages, which most teams underestimate relative to the tooling.
How long before AI visibility work shows results?
Perplexity often reflects changes within days, Gemini within a few weeks on Google’s recrawl cadence, and ChatGPT typically longest because it also depends on third-party sources updating. Plan on four to eight weeks for a readable trend and one to two quarters for corroboration work to compound.
Can a small business compete with large brands in AI answers?
On specific, constrained questions, yes and often more easily than in classic search. Broad category questions favour the largest, most-discussed brands. The winnable ground is the narrow version — a particular industry, size, use case or constraint — where a large competitor’s generic page is not the best answer and three accurate third-party descriptions can make you the named option.