Last checked: 10 June 2026. Disclosure: we are Fortitude Media. We sell both a tracking platform and a delivery service, so we have an interest in the conclusion of this page. The prices in it come from public sources verified on the date above, no tracking tool is named or criticised here, and the first two options at the end do not involve us at all.
First, the tool did its job
If you have paid for AI visibility tracking for three months and the score has not moved, the tool has not failed; it has told you the truth three months running. This page is not an argument against tracking tools. Knowing your position across the AI engines, seeing which competitors take the recommendations you wanted, knowing which sources get cited: that is necessary information, and before these tools existed nobody had it. We sell tracking ourselves and believe in it.
But a tracker is a set of scales, and nobody ever lost weight by owning scales. The score is produced by what the AI engines find when they assemble an answer: your content, your citations, your coverage, your site. If none of those changed since the baseline, the honest score is the same score. The frustration people feel at month three is not the tool's fault; it is the gap between measurement and work, and it is worth understanding before spending anything else.
What actually moves the number
AI engines recommend brands they can quote, verify and trust, which means four kinds of work.
Content built around buyer questions. Engines assemble answers to questions, so they draw on pages that answer questions: what something costs, how to choose, what the trade-offs are, who fits which situation. Most company websites contain almost none of this. The pages that get cited share a recognisable structure, which we have broken down in the anatomy of an article AI engines cite.
Original data. Engines and the publications they cite both favour sources that add new information rather than recycling existing pages. Surveys, benchmarks and findings from your own operations give engines a reason to cite you as a source rather than mention you in passing. We cover why this works in original research and data: how AI decides what to trust.
Citations and PR. Engines lean on what credible third parties say. Coverage, expert commentary and being referenced by publications the engines already trust all feed the answer. This is slow, compounding work, and it is the part most teams never start.
An AI-readable site. Structure, clarity and machine-readable signals determine whether engines can extract what your pages actually say. The thinking here overlaps with technical SEO but is not identical to it; SEO vs LLM optimisation covers where the two part company.
None of these is secret. Every decent provider in the market would give you a similar list, which leads to the real question.
Why it stalls in-house: capacity, not knowledge
Most teams that bought a tracker already know roughly what to do; they cannot find the hours to do it. The list above is a publishing operation: research, writing, editing, data work, outreach, development. A marketing team that is already running campaigns, events and the sales pipeline does not have a spare publishing operation lying idle. So the recommendations export sits in a folder, the quarter ends, and the score is flat again, not because anyone lacked understanding but because nobody had Tuesday afternoons free for it.
This matters when choosing a fix. If the problem were knowledge, the answer would be a course or a consultant. Because the problem is capacity, the answer is hands: yours, hired, or rented. There are three honest versions of that.
Exit one: hire the team
Hiring gives you permanent capability, at permanent cost. An in-house equivalent of the full function (content, PR, web and tooling) runs £165k to £255k a year. For a company where AI visibility is becoming an existential channel, that can be the right call: nobody knows your subject matter like your own staff, and the capability compounds. The drawbacks are the obvious ones: the cost lands whether or not the work succeeds, recruitment takes months, and a small team carries key-person risk. This route suits companies large enough to keep that team busy indefinitely.
Exit two: an agency retainer
An agency converts the fixed cost into a monthly fee and starts faster. Credible UK GEO retainers typically run £2,000 to £10,000 a month, and a good agency will cover most of the work list above. The structural catch is measurement: most agencies own no platform and no longitudinal data, so the score that judges their work comes from third-party tools, often the very tracker you are already paying for. That is workable, but it means you are the one connecting the work to the number, and the agency starts with no history of your category. If you take this route, do it with eyes open: our ten questions to ask any GEO agency is written for exactly this interview, and tool, agency or both maps the wider decision.
Exit three: a closed loop
The third option puts the measuring and the doing inside one company, so the score you have been staring at becomes someone else's accountability too. In our version, it starts with Discovery, a paid diagnostic at around £1,500 that turns your flat baseline into a prioritised plan, and the fee is credited back if you proceed. Delivery then runs through Forge from £3,950 a month: content, digital PR and, where needed, a website rebuilt on the Fortitude Framework, through a monthly pool of delivery hours quoted as human-equivalent time at roughly half human-equivalent cost. Every Forge plan includes Sentinel Pro, our full platform, so the people doing the work and the chart that judges them are the same company's, on one screen. You keep your existing tracker as long as you like; an independent second opinion on our work is fine by us, and how our own measurement is built is public on the methodology page.
The three exits side by side
| Hire in-house | Agency retainer | Closed loop (measure + deliver) | |
|---|---|---|---|
| Cost | £165k to £255k/year | £2,000 to £10,000/mo | Discovery ~£1,500 (credited back), then Forge from £3,950/mo |
| Speed to start | Months (recruitment) | Weeks | Weeks, after Discovery |
| Who connects work to score | You | You, via third-party tools | The provider, on its own platform |
| Best for | AI visibility as a permanent core function | Execution capacity with your own measurement | One accountable party for the number and the work |
There is no universally right column. A company that can keep a £200k team busy should hire it. A company with a trusted measurement setup and a good agency relationship should use them. The closed loop is for the buyer whose specific frustration is the one this page opened with: a known score, no capacity to move it, and no appetite for owning the gap between two vendors. The wider market context, with every provider named, is in our UK B2B roundup, and the case for our version specifically is on why Fortitude.
FAQ
Why is my AI visibility score not improving? Because tracking does not create anything. AI engines build answers from content, citations, press coverage and site structure, and if none of those changed since your baseline, a flat score is the accurate reading. The fix is work, not a different dashboard.
What actually improves AI visibility? Four kinds of work: content built around the questions buyers ask, original data that gives engines a reason to cite you, citations and PR from sources the engines already trust, and a site structured so engines can read it. All four are publishing work that a tracking subscription does not include.
Should I cancel my tracking tool if the score is flat? No. The tool is providing an accurate baseline and you will need it to judge whatever you do next. The decision is not tracker versus no tracker; it is who does the work the tracker is waiting to measure.
How much does it cost to act on AI visibility data? An equivalent in-house team runs £165k to £255k a year, UK agency retainers typically run £2,000 to £10,000 a month, and our closed-loop route is a Discovery diagnostic at around £1,500 (credited back if you proceed) followed by Forge managed delivery from £3,950 a month with the full Sentinel Pro platform included.
How long does it take for AI visibility to improve? No provider can promise a fixed timeline, because AI answers vary between engines and over time and the work compounds rather than switching on. Judge any provider on concrete monthly output and dated measurement against named competitors, not on a guaranteed date.
Want a second reading of your category? The free AI Visibility Check shows how the five major AI engines answer your category's buying questions today. If it confirms what your tracker says, you have lost nothing; if it shows something different, that is worth knowing too.