009 Agency’s GEO services are one continuous loop rather than a menu. The order is not a preference: entity work on a site an AI crawler cannot read is invisible, and citation work on a brand a model cannot verify does not hold. So everything starts with measurement and ends with measurement.
the loop
| stage | what happens |
|---|---|
| 01 · audit | A manual scorecard of how six engines see, describe and recommend you, with a dated 90-day plan. GEO audit → |
| 02 · strategy | The prompts worth winning, the competitors already winning them, and the order of attack for your category. |
| 03 · technical | Crawler access, rendering, the JSON-LD entity graph, llms.txt, headers, Core Web Vitals. technical GEO → |
| 04 · content & SXO | The page types models actually quote: definitions, comparisons, pricing, lists. Written with QA before you see a draft. |
| 05 · authority | Getting into the sources engines read: directories, roundups, industry lists, communities. |
| 06 · measurement | The same prompt panel re-run every review, reported as a delta against your logged baseline. |
| 07 · compounding | What moved gets more budget, what did not gets dropped. Decided on the numbers, not on taste. |
Stages 02, 04, 05 and 07 are delivered inside the retainer and are not sold separately — there is no useful way to scope them before the audit has measured where you actually stand. Stages 01 and 03 have their own pages because they are the two things people search for by name.
what makes this different from an SEO retainer
Two things, and both are checkable.
We measure the thing you are buying. Not rankings as a proxy for visibility, but the answers themselves: a fixed panel of your buyers’ prompts, run live through ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek, logged verbatim with the sources each engine cited. That panel is your baseline and every review re-runs it.
We ship files, not advice. Each sprint hands over things a developer applies without interpreting them: the llms.txt, the schema blocks with identifiers wired, the server config, the page briefs. Then we re-check that the fix landed, because a fix nobody verified is a claim.
And because many of the sources engines quote are themselves ranked pages, classic SEO is inside the loop rather than beside it. One team, two scoreboards: rankings in Google, recall and citations in the engines.
how engagements start
Every engagement opens with the manual GEO audit. It is complimentary, it takes five to ten working days, and you keep the scorecard and the 90-day plan whether or not we work together. The full-cycle retainer starts from $2,000/mo (August 2026) and runs in bi-weekly sprints.
We are referral-first and capacity-bound, so the honest constraint is how many audits we can run by hand at once — not price.
who this is for
Fifteen niches, the ones our audit dataset covers: Web3, FinTech, AI companies, SaaS, cybersecurity, cloud & DevOps, IT outsourcing & software development, LegalTech and law firms, iGaming, e-commerce, real estate technology, ArchViz, GameArt, consulting, and B2B service companies. In each of them we already know what the AI shelf looks like before the first sprint.
The method, the sample and every mistake we catalogued are public in 100 GEO audits: the report, and the vocabulary is defined in the GEO glossary.