Learn

The whole method, written out so you could run it without us.

20 guides covering one full loop: baseline, site fixes, content and sources, re-measurement. Method details link out to the blog rather than being repeated here.

Start with the baseline Have us run it

The path

One loop, in order.

Each stage produces what the next one needs, which is why the order is not a suggestion.

All The 90-day GEO implementation path The overview guide: what to do in which order, and the three places implementations usually stall. 01 Measuring your baseline Step one. An afternoon of work, no tools required, and the only thing that makes later claims checkable. 01 Choosing which questions to measure Where the question list comes from, and the one category that will inflate every number if you let it in. 01 Baselining competitors, not just yourself Share of voice only means something if competitors are counted the same way you are. 02 Making your site readable to machines Step two. Small fixes, disproportionate effect, and the stage that stalls most often. 02 Making your brand unambiguous to a model Before a model can recommend you it has to be sure which "you" is meant. Short and generic names make that hard. 02 Structured data that actually gets used Four types carry almost all the value. Markup that contradicts the visible page is worse than none. 02 Letting AI crawlers in, deliberately The most common cause of zero citations is not bad content. It is a page the crawler never received. 02 Writing an llms.txt that earns its place Not a sitemap in another format. A short, opinionated index of the pages that answer something. 03 Content and sources, together Step three. The stage where most programmes go wrong — by doing only half of it. 03 Building the fact base your content stands on The difference between content that gets quoted and content that reads like everyone else is whether there is anything specific in it. 03 Writing FAQ answers that get quoted The structure a model needs is the opposite of the one most FAQ pages use. 03 Third-party sources: what actually corroborates you A claim only you make is a claim. The same claim found somewhere independent is evidence. 03 Choosing where to place, and what to expect An outlet that is never indexed cannot be cited, no matter what its metrics say. 03 Working with the differences between platforms One content strategy, several retrieval behaviours. The differences matter most for measurement. 03 Local businesses: the questions are different A nearby question is answered from a short list. Getting onto it depends on facts most local sites never publish. 03 Regulated industries: where the line is The constraints are real and they are not only legal - models themselves answer these categories differently. 04 Re-measurement, acceptance and reporting Step four. The step most often skipped, and the only one that proves the first three worked. 04 Who does what, and on what cadence Most implementations stall for organisational reasons, not technical ones. 04 Reading a recheck, and deciding what to do next Most rechecks produce a small movement and an attribution problem. Both are normal and both are workable.

How this differs from the blog

The blog covers one thing per piece — how structured data works, how to write a FAQ that gets cited, how generative engines pick sources. These guides do not repeat that. They put the pieces in order and tell you what to do first, linking out to the blog wherever the method detail lives.

If you want to understand one mechanism, start in the blog. If you want to run a programme, start here.

FAQ

Questions teams ask before they start

Why publish the whole method?

Because the part clients actually buy is not the method — it is the placement relationships and the discipline to keep re-measuring every month. Both are hard to sustain in-house, and neither becomes easier by keeping the method secret.

Should I read the blog or these guides?

Blog for one mechanism in depth — how structured data works, how to write a FAQ that gets cited. Guides for the order of operations: what to do first, what each stage produces, where implementations stall.

Can one person run the whole loop?

Baseline measurement and re-measurement, yes. Site readability fixes need engineering time, and source placement needs channel relationships and a steady publishing cadence; those two are the parts a single person rarely sustains internally.

Do the guides require paid tools?

No. Every step in the five guides can be carried out with a spreadsheet and a browser.

How long until changes appear?

On platforms with live search, new content usually takes days to appear in answers; platforms relying more on training data follow their own release cycles. A full loop takes about 90 days, which describes how long the stages take to run, not a commitment about results.

Which step fails most often?

The third. Producing content without placing it is the most common reason an engagement ends with no measurable change. The second most common is re-measurement lapsing at round three, after which nothing can be established either way.

Find out where your brand actually stands in AI answers.

An AI visibility diagnosis across the platforms your customers use, on your real questions. No charge for the first look.

Get a free diagnosis 17722562362

The whole method, written out so you could run it without us.

One loop, in order.

How this differs from the blog

Questions teams ask before they start

Keep exploring

Find out where your brand actually stands in AI answers.