Organic clicks more than doubled, AI visibility nearly quadrupled
A global insurance group gave us one line of business and one quarter to prove that SEO and GEO could move it. Organic clicks more than doubled while the rest of the same domain lost ground, higher-intent search more than doubled its share of non-branded clicks, and the brand's coverage in AI answers multiplied close to four times over.
We do not name the client, a global insurance group, so every figure on this page is relative: a multiple, a percentage change, or a share of a whole.
About the client
A global insurance group, operating in India through a licensed general insurance business. The line of business here is travel: the cover an Indian traveller buys before a trip abroad. It sits on the same domain, the same CMS and the same technical stack as the group's other products, and it competes for search against aggregators, comparison sites and every large domestic insurer.
The group already had an incumbent agency and a knowledge centre full of published articles. What it did not have was movement. Organic traffic had been flat for months, and when someone asked an assistant which travel insurance to buy for a trip abroad, the brand almost never came up.
A flat traffic curve, and no name in AI answers
Flat is harder to fix than broken. Nothing on the site was obviously wrong. The pages loaded, they were indexed, and a fair number of them ranked. The traffic curve simply refused to move, and the incumbent agency could not say why.
The answer was in the shape of the traffic rather than its size. Travel demand in India is won upstream, at the point where somebody is working out whether they need a visa and what the trip will cost, not at the point where they type the word insurance. Almost all of the site's travel content sat at the very top of that funnel, on passport paperwork. It brought people in and then had nothing to offer them.
AI answers were a separate problem with the same root. Assistants name a handful of brands when somebody asks what travel cover to buy, and this brand came last among the ten that surfaced on Indian travel prompts. An AI answer has no page two. Either the assistant names you or you do not exist to that user.
The group wanted proof inside a quarter, on one line of business, before committing anything further. That made the brief unusually clean: move this section, leave the rest of the site alone, and let the difference between them stand as the evidence.
What we ran
A diagnosis in the first week
We pulled page-level Search Console data for every travel URL, classified each one by funnel stage, and set it beside the prompt set that AI assistants answer for Indian travel insurance. Two gaps came out of that. The content sat almost entirely at the top of the funnel, and the brand held a negligible share of the citations assistants draw on when someone asks what travel cover to buy. Both were fixable with content and technical work, and we said so in week one rather than after a discovery phase.
A content map built around the trip
Search volume alone would have sent us straight back to passport admin, which is where the site already was. We built the map around the trip instead: visa rules by corridor, destination planning, what a trip actually costs, then the comparison and cover queries that sit next to a purchase. Roughly three in four of the new pages target the middle of the funnel by design, because that is where a traveller stops reading and starts deciding.
Technical work so the new pages could enter the index
Volume counts for nothing if Google never crawls it. We cleared the indexing and crawl issues holding the section back before the content landed, which is why several hundred new URLs entered search and began earning inside the same quarter instead of queueing behind one another. Pages that went live in July were ranking by late August.
Pages written to be quoted
Ranking and getting cited are different jobs. Every page carries a direct, liftable answer near the top, names its entities explicitly, ships structured data, and uses formatting an assistant can quote without rewriting. Citations decide which brands an AI answer names, so a page that ranks but cannot be quoted leaves the GEO half of the work undone.
A tracked prompt set, run like a keyword set
We track the questions Indian travellers put to assistants across six AI surfaces, fortnightly, alongside the competing brands named in the same answers. The client's team picked fifty of those prompts as the targets. Holding that set fixed gives a clean read on progress: when coverage moves, we know the prompt list did not move under it.
A control group, on purpose
We worked one section of the domain and deliberately left the rest of it alone. Same CMS, same stack, same Google volatility, same fortnight. Every figure on this page is read against that comparison, which is what separates a result from a good quarter.
Results
Twelve weeks in, the section we worked on had more than doubled its organic clicks while the rest of the same domain gave up ground. The audience moved away from passport paperwork and toward people planning a specific trip. And in AI answers the brand climbed from the bottom of the ten that assistants name on Indian travel questions to seventh, with coverage close to four times what it was at the start.
Weekly clicks climbed through the first month, then steepened
We index every week to the fortnight before the work started, which shows the shape of the curve without publishing a traffic number. Flat for the eleven weeks before, a slow climb through the first month as new pages entered the index, then a steeper and steady slope across the final two months as the library matured.
The section we worked on grew. The rest of the domain did not.
Both sit on one domain, one CMS and one technical stack, and both met the same Google volatility over the same fortnight-on-fortnight window. One of them received content and technical work.
The worked section added more clicks over the fortnight than the entire rest of the site gave up, so it accounts for all of the domain's organic growth and then some.
Higher-intent search went from 18% of clicks to 39%
We assign a funnel stage by landing page rather than by keyword text, so the queries Google withholds still count. Both halves grew. Higher-intent search grew several times faster, which moved the audience from people renewing a passport toward people planning a specific trip and shopping for cover.
Non-branded clicks only, with visa rules, destination planning and cover queries counted together as higher intent. Travel search on this domain is almost entirely non-branded, so brand demand carries none of this growth.
The brand went from almost never named to named on one prompt in thirteen
Coverage across the tracked prompt set rose from 0.6% to 7.5%, a 12.5× move. Google AI Overviews and Copilot responded hardest; ChatGPT and Perplexity went from never naming the brand to naming it regularly.
Measured against a prompt set held fixed across both windows, so the denominator never moves. Coverage is the share of tracked prompts where an answer names the brand; share of voice weights that against every other brand named in the same answers.
| Metric | Before | Latest | Change |
|---|---|---|---|
| Organic clicks | 1.0× | 2.1× | +112% |
| Search impressions | 1.0× | 3.9× | +293% |
| Distinct non-branded queries | 1.0× | 3.0× | +203% |
| Higher-intent clicks | 1.0× | 4.6× | +359% |
| Top-of-funnel clicks | 1.0× | 1.6× | +58% |
| Higher-intent share of clicks | 18% | 39% | 2.2× |
| Brand coverage in AI answers | 1.0× | 3.8× | +278% |
| Share of voice in AI answers | 1.0× | 3.4× | +242% |
| Citations to client pages in AI answers | 1.0× | 4.9× | +388% |
| Rank among brands AI assistants name | 10th of 10 | 7th of 10 | +3 places |
| Rest of the same domain, same window | 1.0× | 0.97× | -2.7% |
“We brought in Geology because our organic traffic was stagnating and our existing agency was not able to break the flat traffic curve. Mehul and Sankalp broke down the issues for us in the first week of the engagement. The traffic and revenue from both Google and AI search has multiplied already. Their execution has been really fast even with a small team.”
This is the playbook behind our insurance SEO and GEO program. For the mechanics of the travel line specifically, read the travel insurance GEO guide, or see how a smaller broker ran the same play.
Prove it on one line of business
Run a free audit and see where your organic and AI visibility stands today, then pick the one section worth moving first.