AI Search Optimisation
Found in Google.Cited by the AI.
Ranking first is no longer the whole job. A growing share of buying research now happens inside ChatGPT, Perplexity and Google AI Overviews, and being invisible there is a slow leak you cannot see in your rankings report. I work on both, because they need different things.
- Technical first
- Entity work
- Citations tracked weekly
Sound familiar
What people tell me before we start
We rank first for it, but ChatGPT recommends three competitors.
Traffic is flat and nobody can tell me why.
Our agency sends a rankings report. That is the whole service.
We publish two posts a week and nothing has changed in a year.
The approach
Two audiences, one foundation
A ranking algorithm picks the best page for a query. An answer engine has already decided what it wants to say and is looking for a source it can attribute. Those reward different things, which is why sites that rank well are often absent from AI answers about their own category.
What they share is the foundation. Both need a site that can be crawled and parsed, an entity that is unambiguous across the web, and claims specific enough to be worth citing. Get that right and you are competing in both places. Skip it and no amount of publishing helps.
So the order is always the same: fix what is broken technically, make the entity coherent, then write content designed to be quoted rather than to hit a word count.
What you get
Every search engagement includes
The same foundation regardless of which spoke you need. No tiers.
-
Technical audit against real crawl data
Not a tool export. I crawl the site, compare it to what search engines have actually indexed, and find the gap between the two.
-
Structured data as an entity graph
A connected JSON-LD graph describing what you are and how the parts relate, not disconnected blocks bolted onto templates.
-
Entity reconciliation off-site
Consistent naming and description across every third-party profile that mentions you, because inconsistency splits one strong entity into several weak ones.
-
Content built to be quoted
Answers in the first two sentences, specific claims with numbers, explicit statements of who something is not for.
-
Answer engine citation tracking
A fixed prompt set run on a schedule across the major answer engines, recording where you appear and where competitors do.
-
Reporting you can act on
What changed, what it moved, what is next. Direct access to the dashboards rather than a monthly summary of them.
The process
Fix, then build, then measure
-
Fix the foundation
Crawlability, indexation, speed, structured data. The unglamorous work that everything else depends on, done in the first weeks.
-
Make the entity coherent
One name, one description, one graph, consistent everywhere you appear. This is what answer engines resolve against.
-
Publish and track
Content designed to be cited, with a weekly prompt set measuring whether it is working in both Google and the answer engines.
If it is not measured, it did not happen
Rankings are a proxy. I report on qualified organic sessions, on answer engine citation share against a fixed prompt set, and on enquiries. Those are the numbers that pay for the work.
Receipts
Search work with the numbers attached
Who you work with
Someone who came to search from infrastructure
I found search sideways. A client had a site their SEO agency could not fix because the problem was a server configuration, not a content one. Diagnosing it was the most valuable marketing work I had done, and I have been doing both ever since.
That background shows up in what I prioritise. Crawl budget is an infrastructure question. Core Web Vitals are an engineering problem wearing a marketing hat. Structured data is a data modelling exercise. Most of what separates a site that performs from one that does not is upstream of anything a content calendar can fix.
You work with me directly. No account manager translating, no junior running the audit tool.
Questions
Search questions, answered plainly
How long before we see results?
Technical fixes can move things within weeks, sometimes days if something was actively blocking indexation. Content and entity work is a three to six month arc. Anyone promising page one in thirty days is either lying or planning to rank you for terms nobody searches.
Is AI search actually worth optimising for yet?
For considered purchases, yes, and the trend is one-directional. The good news is that the work overlaps heavily with what makes classic SEO work: a fast crawlable site, a coherent entity and specific content. You are not choosing between them. You are doing the foundation once and competing in both.
Do you write the content?
I write the technical and commercial pages, and I brief and edit anything longer-form. What I will not do is generate volume for its own sake. If the honest answer is that you need six excellent pages rather than sixty adequate ones, that is what I will recommend, even though it is a smaller invoice.
Will you work alongside our existing agency?
Yes, and it often works well. I frequently handle the technical and entity layer while an in-house team or agency runs content and outreach. What I need is a clear split of responsibility so nobody is undoing anyone else’s work.
What if our site needs rebuilding?
Sometimes it does, and I will say so. A site that cannot be made fast or crawlable within its current platform is a ceiling on everything else. I build sites too, so I can tell you honestly what is fixable in place and what is not, but I will always cost the cheaper option first.
Next step
Send me the domain
I will look at it properly and tell you the two or three things actually holding it back. That is a real answer, not a lead magnet.
- Reply within 24h
- No rankings-report-only retainers