SEO · Local SEO · AEO · GEO · AI Search
Turn search change into measurable visibility.
An SEO agency that also measures AI search. Technical SEO, local SEO, content and off-page work, then we measure what ChatGPT, AI Overviews and the map pack actually say about you. From $900 a month.
SEO audit
$99, three days
Measurement
Per platform, never blended
Delivery
Founder-set, team-run
Scope
SEO, AEO, GEO, local
01What changed
Ranking and citation have come apart
For twenty years, ranking in the top ten meant being seen. That link has weakened quickly, and the numbers are not subtle.
Verified Published third-party research, not Searlixa data. Sources named above and linked on our methodology page. Every number on this site carries its status.
Read those four rows together and one conclusion follows. A brand can hold position three on Google and still be absent from the answer a buyer actually reads. It can also be cited constantly on one platform and invisible on the next. Search Console will not tell you either thing.
The second row is the one that changes budgets. If more than half of AI citations come from outside the traditional top ten, then the page you have spent two years pushing toward position four is not automatically the page that gets used. Something else is being selected, and until you measure it you are guessing at what.
The fourth row breaks a different assumption. Nearly three quarters of the AI-cited pages in that sample had no page-level referring domains at all. Links were not the entry condition. That does not mean authority stopped mattering — it means the mechanism selecting sources for an answer is not the mechanism ranking them in a list, and treating them as the same thing produces a plan aimed at the wrong target.
None of this makes search work obsolete. It makes the reporting obsolete. The dashboard that told you how you were doing now describes a shrinking part of how buyers actually find you.
→Why this matters
Being found is more than a Google ranking
A page-one position used to be the whole contest. The same question now gets answered inside an AI Overview, a ChatGPT reply, a map pack or a YouTube result, often with a competitor cited and you nowhere in it.
Discovery is not one channel to rank in any more. It is a set of surfaces to be present across, and Search Console reports on one of them.
02Where we stand
Optimisation is largely SEO. Measurement is not.
In May 2026 Google published documentation stating that optimising for generative AI search is optimising for the search experience, and is therefore still SEO. It named llms.txt files, content chunking and AI-specific schema as things you can ignore for Google Search.
We agree with that, and we think most of our category has been slow to say so. The work that earns citations on Google is the work that earns rankings on Google. There is no separate engine to game, and an agency selling you a distinct AI playbook is usually selling you the same technical and content work with a new invoice line.
That position costs us something, which is why it is worth stating. It removes three easy deliverables from our own proposals. A file you can write in ten minutes looks like progress on a monthly report; explaining that it does nothing is a harder conversation and a shorter scope.
But Google's guidance answers a question nobody was really stuck on. The hard question is the other one. Are we cited, on which platforms, for which buyer prompts, against which competitors, and is it moving? Nothing in Search Console answers that. Platform behaviour diverges too far, and more than half of AI citations sit outside the traditional top ten.
There is also a limit worth naming. Google is describing its own surfaces. ChatGPT and Perplexity run their own retrieval over their own corpora, and the two most similar platforms in that 2026 research shared only 24.1% of cited pages. Advice that is correct for AI Overviews does not transfer automatically to an assistant.
That gap — between what is optimised and what can be seen — is where we work.
03What we do
SEO, AEO and GEO: measurement first, because you cannot prioritise a gap you have not sized
To say it plainly, because the rest of this page is about measurement: we are an SEO agency. Technical SEO, local SEO, on-page and content, entity work and earned off-page placement. The AI visibility measurement sits on top of that work, not instead of it, and for most clients the SEO is still where the results come from.
Almost every engagement starts the same way: a $99 audit covering technical SEO, on-page, AEO, GEO, LLM visibility, entity accuracy and off-page, back within three working days. Not because it is a clever funnel, but because the alternative is committing a quarter of budget to a guess.
What the audit finds is rarely what clients expect. Content is almost never the first item. A crawler that was meant to be allowed is blocked. A comparison site is quoting pricing you changed two years ago. Critical copy renders only after JavaScript, so some retrieval systems have never seen it. Three of your own pages are competing for one intent. Those take days to fix and they move first.
The genuine content gap — nothing authoritative connecting you to a query family at all — is real and it is the slow one. Months, not weeks. Knowing which of those five you actually have is the difference between a quarter well spent and a quarter wasted.
After the audit, an engagement runs on a loop rather than a launch: diagnose, prioritise, engineer, measure, repeat. The prompt set is frozen so trend data means something. Sampling happens on a fixed cadence. Every platform is reported separately, because a blended score would average away the finding it exists to surface.
Prices are published in full, including what each tier delivers under every discipline. Programmes start at $900 a month.
04How we prove it
Every number carries its status, including the ones that weaken our case
Five labels, applied to every figure we publish. Verified means third-party research with a named source. Observed means our own sampling, with sample size and date attached. Estimated means modelled, and the model is described. Demo means illustrative, and nothing real is implied. Updated means the figure changed and the change is dated.
If a number on this site has no label, we have not published it. That rule applies to the inconvenient ones too. The research on this page includes a finding that nearly three quarters of AI-cited pages had no referring domains — which sits awkwardly next to an off-page service line, and is on the page anyway.
The same standard governs what we refuse to sell. There is a published list, and it is longer than most agencies would put in front of a buyer: no llms.txt as a billable service, no content chunking, no AI-specific schema packages, no guaranteed rankings or citations, no bought links or paid guest posts, no page-count content programmes, and no single blended AI visibility score across platforms.
Each refusal has a reason published beside it. Most of them cost us revenue. That is the test of whether an evidence standard is real: it has to remove something you would otherwise be paid for.
The full measurement methodology is public, including the sampling procedure, the coding rules and the things the method cannot see.
05Who you would work with
The agency behind it: strategy and verification are not delegated
Searlixa is led by Shrawan Swami, who has worked in search since 2013 and sets the strategy on every engagement. The prompt set, the competitor set and the recommendations are his. Sampling and coding are run by the delivery team against his documented rules, and he checks every report before it is sent.
The sector depth is easiest to check on the work itself. Twelve named medical and hospital engagements since 2018, from national hospital groups to single-practitioner clinics, across India and Australia. Every client is listed and verifiable. No outcome figures are published, because publishing a performance number without a documented baseline is exactly what this site tells buyers to distrust.
Six industries, seven markets, and both lists are shorter than they could be. The source plan for this project contained 41 industry hubs and 76 city pages. Most would have been the same service description under a different noun, which is the pattern Google's spam policies name directly. We publish a page when we have something specific to say about that vertical or market, and we say plainly on the markets page which markets we serve without one.
06Common questions
The first questions buyers ask
What is the difference between SEO, AEO and GEO?
SEO is optimising to be found in search results. AEO is answer engine optimisation, aimed at being used inside AI assistants. GEO is generative engine optimisation, aimed at AI summaries inside search engines. Google's position is that all three are the same discipline for its own products. The practical work overlaps heavily; the measurement does not.
How do you measure AI visibility?
We build a fixed set of buyer prompts from your real sales language, then sample it repeatedly across each platform. From that we report citation presence, share of voice against named competitors, and which of your pages get used. Because AI answers are non-deterministic, every figure carries its sample size and date range.
Which AI platforms do you track?
ChatGPT, Google AI Overviews, Google AI Mode, Gemini and Perplexity as standard, with Copilot and Claude on request. We report each platform separately. Published research found the two most similar platforms shared only 24.1% of their cited pages, so a blended average hides more than it shows.
How quickly can we expect results?
The $99 audit comes back in three working days. Meaningful movement on a site with existing authority usually appears in three to six months. A brand-new domain in a competitive B2B category should plan for twelve to eighteen. We would rather tell you that before you sign than after.
Do we need an llms.txt file?
Not for Google. Its documentation states the file receives no special treatment, and adoption sits under 10% of the top 1,000 websites. There is a reasonable argument for it with some non-Google agents. We will help you decide and record the reasoning, but we do not bill it as a visibility service.
Do we have to stop doing traditional SEO?
No, and doing so would be a mistake. Traditional top-three results appeared in AI citations 98.9% of the time in one 2026 study. Ranking is no longer sufficient, but it is still the strongest single gateway into AI answers. The two programmes share most of their work and should share a budget.
Find out where you actually stand
An assessment gives you a baseline: which prompts your buyers use, where you appear across search and AI answers today, who is taking the citations you are not, and what we would do about it in what order.
It comes back in three working days. You get the prompt set and the raw sampling either way, so the work is useful whether or not we go further.