Generative engine optimisation:
be in every source AI reads.

They aim at the same moment, an AI-written answer naming you, from different directions. AEO works on your own pages: answer-first formatting, schema and structure so engines can quote you directly. GEO works on the wider corpus those engines retrieve from: the directories, press, reviews and profiles where your brand is described, made consistent and worth citing. If AEO is making yourself quotable, GEO is making the rest of the web agree about who you are. Most engagements pair them. Each stands alone when the other half is already handled.

The honest answer: nobody outside the model companies knows the exact weighting, and anyone claiming a formula is guessing. What is observable is the mechanism: for most commercial questions, assistants retrieve live web content through search indexes and their own crawlers, then write from the passages they fetched. Retrieval visibly favours content that is reachable, clearly structured, explicitly sourced and consistent with what the rest of the web says. Those are levers you can actually work. GEO is the discipline of working them systematically rather than hoping.

No, and nobody can sell you that. What trains a model is decided by the model companies on their own schedules, and by the time a training snapshot matters it is already historical. The practical lever is retrieval: when a buyer asks a commercial question today, most assistants search the live web and answer from what they find. That is the corpus you can influence this quarter, and it is where this service does its work. There's a slow, compounding side-benefit too: a consistent, well-cited brand is also better represented wherever future snapshots are taken.

In the same two columns as our AEO reporting: this is a young discipline, and no engine publishes citation logs. What we can track: scheduled spot-checks of the major assistants on an agreed question set, recorded over time so direction is visible; AI referral sessions in your analytics; every piece of coverage and every mention earned; and the input work itself: profiles aligned, schema live, assets published. What we cannot: impressions inside assistants, or how answers vary between users. Every report separates the measured from the unmeasurable, so you always know which is which.

Engagements are scoped to the state of your existing footprint. A brand with consistent profiles and steady press needs a lighter programme than one starting from contradiction. Every proposal itemises the deliverables (the corpus audit, profile fixes, schema work, PR assets, reporting cadence) and lands priced within 48 hours of the first call. On timing: consistency fixes take effect as engines re-crawl, typically within weeks. Earned coverage builds over months and compounds after that. This is a programme, not a toggle.

Yes, the programme ships under partner agencies' brands like every other service here: your reporting templates, your client calls, our delivery. It is NDA-backed, with a contractual commitment never to approach your clients and a 12-month post-engagement blackout. 100+ agencies work with us this way, and the average partnership runs past the two-year mark.