Search results are changing shape. Instead of a page full of links, people often see a ready-made answer first, with sources sitting underneath. Google says AI Overviews and AI Mode can run a “query fan-out”, issuing multiple related searches across subtopics to build a response and surface a wider set of supporting pages than a classic search might. That’s handy for readers, but it also means brands can shape decisions without earning a visit.
In the click-first era, search engine optimisation reporting was a scoreboard: impressions, rankings, CTR, sessions, conversions. Those metrics still matter, but they miss a growing slice of visibility. A user can read an AI answer, decide what to do, and never click. When your brand is cited or named inside that answer, you’ve been included in the part of the journey that used to happen on your site.
GEO in plain language
GEO, short for generative engine optimisation, is about appearing inside AI-generated answers, not only alongside them. Industry definitions describe GEO as optimising content so it can be used and cited across tools such as ChatGPT, Gemini, Copilot, Perplexity, and Google’s AI features. Academic work also argues for impression-style metrics that capture visibility in generative outputs, because the interface composes an answer rather than listing ten links.
What “inclusion” actually looks like
“Inclusion” comes in flavours. You might get a direct source link from best SEO services, you might be named, or you might be paraphrased while someone else gets cited. Google’s guidance says there are no extra technical requirements for AI Overviews or AI Mode beyond being eligible for Search snippets, and it frames these features as widening the range of supporting links. Citations are attributable; mentions can still sway perception but won’t show up neatly in analytics.
Build a prompt set that mirrors real behaviour
A keyword list is useful, but it’s a blunt tool for generative answers. Start with a prompt set: 30 to 50 questions your audience genuinely asks, written in plain language. Mix in comparisons, objections, local intent, and the follow-up questions people type after reading an answer. Keep the set steady for a few months so changes are meaningful, not noise. A SEO consultant can help you map prompts to decision stages (learn, shortlist, buy) and tag each one, so reporting shows where inclusion grows and where it stalls.
Four metrics to report each month
Start with inclusion rate and citation rate. Inclusion rate is the share of prompts where your brand or domain appears anywhere in the answer or its citations. Citation rate is stricter: you only count it when your site is directly linked as a supporting source. The gap between those numbers tells you whether engines are comfortable naming you, or comfortable relying on you as evidence. Include a stress-test prompt like SEO experts near me because list-style answers make it obvious who gets recommended.
Prominence is where you show up with the help of best SEO agency in Sydney: first screen of citations, mid-pack, or buried behind a “view sources” click. Share of voice compares you with the competitors you actually lose deals to, counting mentions and citations across the same prompt set. Walker Sands defines share of voice in AI responses as a percentage based on brand mentions and citations relative to others across AI platforms and Google’s AI features.
Then track sentiment and factual accuracy. Record the tone the engine uses for you (specialist, budget, premium), and flag anything that’s simply wrong: services you don’t offer, suburbs you don’t cover, outdated staff details, or invented pricing. If your site is vague, models can fill the gaps with assumptions, and a phrase like affordable SEO packages can get attached to a promise you never made. Treat accuracy fixes as content maintenance, and re-test the same prompts after updates.
Link inclusion to outcomes
Inclusion shouldn’t become a vanity chart. Tie it back to behaviour. Some interfaces make links prominent; others hide them behind icons or expandable panels. Microsoft’s Copilot Search highlights cited sources and lets users view the sources that informed the response. When you do earn visits, compare their quality to classic organic sessions: time on page, enquiry starts, assisted conversions, and which pages show up repeatedly as citations. Google also notes that sites appearing in AI features are included in Search Console’s Performance reporting under the standard Web search type.
A simple rhythm that holds up
Run your prompt set monthly, save the full responses, and record a handful of fields: included (yes/no), cited (yes/no), prominence, sentiment, and accuracy notes. Then pick one improvement action for the next month based on the clearest pattern you saw. The GEO research paper reports that content changes such as adding relevant facts, quotations, and statistics can lift the likelihood of a source being used or cited in generative responses. After a few cycles, you’ll have a clearer picture of inclusion and a shortlist of changes that are moving it.