AI SEO done right: we get your site found in Google and Bing, and your brand cited in ChatGPT, Perplexity, Claude, and Gemini answers. By the end, you own a content production system one in-house person can run.
Ten years ago there was one channel: Google. Today the same person gets answers from five different systems, each with its own rules for who gets cited. A strategy built for one platform covers about 11% of the traffic that's actually out there.
When an AI Overview sits at the top of the page, the CTR for position #1 drops. Gartner expects traditional search traffic to fall further by the end of 2026.
AI Overviews now show up on a quarter of all queries. Brands cited inside those answers see organic CTR climb by 35%.
800 million weekly users. 87% of ChatGPT citations overlap with top Bing results, which is why being indexed in Bing is non-negotiable now.
780 million queries a month, real-time RAG. Every answer pulls 5–8 sources from about 7 unique domains and links each one. Recency is heavily weighted.
Each system favors its own set of sources. ChatGPT alone now drives 10% of new sign-ups on Vercel. That's an acquisition channel, not an experiment.
The industry hasn't settled the vocabulary. The same work goes by four names: GEO, AEO, LLMO, AI SEO. In practice we run all four as services — GEO, AEO, AI SEO and LLMO — because they only work together. Skip the SEO foundation and the other layers give you nothing back — AI systems pull from the Google and Bing index, so if you're not in there, you're invisible to them too.
The classic. Optimization for Google and Bing. Goal: top-10 positions and clicks from the SERP.
Optimization for direct answers inside Google and Bing. Goal: own position zero — the featured snippet.
Optimization for brand mentions inside ChatGPT, Perplexity, Claude, and Gemini answers. Goal: become a source the AI cites.
The technical sub-layer of GEO. Structuring data so LLMs can pull it cleanly and cite it without distortion.
Without a foundation, the finish falls off. With a foundation but no finish, you lose every buyer who would have picked you through an AI answer.
Princeton / Georgia Tech (Aggarwal et al., KDD 2024): 40% of GEO ranking factors overlap with classic SEO. The other 60% are specific to extraction, authority, and brand mentions.
The foundation. Without it, AI systems can't see your brand at all. B2B and SaaS SEO: keyword research, technical SEO, site architecture, Schema.org, hreflang, link building.
As a GEO agency we run generative engine optimization services — structuring content so LLMs can extract it cleanly. Answer capsules, FAQ blocks, declarative tone, dated citations.
The files that let AI systems read your site correctly. "LLM-readability" becomes an engineering job.
We use off-the-shelf tools where they work and build custom ones where they give a real edge. The custom tools stay with the client when the project ends. JavaScript SEO is part of the routine: we render JS content the way crawlers do and verify the bots can actually read it.
Alongside the off-the-shelf platforms, we build internal tools for the client's specific workflows. They live inside the client's stack and stay there after the project closes.
Some projects need extra tooling beyond this. We work that out during the architecture phase.
Hover a row to expand the phase.
Data-driven keyword research. Technical audit of the existing site, or planning from scratch if there is no site. Schema.org markup, hreflang, robots.txt configured with explicit access for AI crawlers.
Content production against the priority clusters from the keyword research. Every page is structured for extraction: answer capsule right after each H2, FAQ blocks, tables, Schema.org Article + FAQPage.
Expand content coverage. Programmatic landing-page generation from the client's own data, where it fits the category.
Standard project length depending on scope and complexity. When the project ends, the client runs the system in-house — without a subscription or agency lock-in.
Structured data for Google and Bing, access for AI crawlers, fast indexing.
Published pages on your priority queries, written in the formats AI systems cite the most.
Dashboard, rank tracker, AI brand monitoring across AI answers — inside the client's stack, with documentation.
An internal tool: someone on the team clicks "write a guide on topic X" and gets a draft in the brand voice, structured for AI citation. Style guide and a library of proven formulas live inside the tool.
Runs the process after we step out. Two training sessions and a final report close the handoff.
The old SEO metrics stay: positions, CTR, organic traffic. We add new KPIs for AI visibility on top. We don't promise specific traffic or rank numbers. Anyone guaranteeing "top-3 in 3 months" either doesn't know how SEO works or is lying.
The share of category answers that mention your brand. Calculated against a fixed list of priority queries, run regularly across every target AI platform. The king metric for GEO.
Raw count of brand mentions and links to your site inside AI answers. Broken out per platform.
Whether AI gets your services, focus area, and region right. Hallucinated facts are a signal to fix entity consistency.
Share of target search intents covered by content the AI can actually extract and cite.
Real clicks from AI platforms, captured through GA4 custom dimensions. The visitor lands already past the decision — shorter sales cycle, higher average order value (Semrush, June 2025).
The old SEO metrics aren't going anywhere. We add AI visibility on top of them.
Six things we skip by default on every project. Most are mistakes we see repeated across clients who come to us after a failed first attempt.
robots.txt
Sites built a few years back often ship with robots.txt that blocks GPTBot, PerplexityBot, and Google-Extended by default. AI systems physically can't read the content. First check on any audit: open robots.txt and make sure AI bots are explicitly allowed.
A company orders 100–200 AI-generated articles with the brand name sprinkled in. AI systems detect the template and ignore that content when building answers. Real example from 2025: 150 articles, zero results, not a single one cited by ChatGPT or Perplexity.
Company name spelled one way on the site, another on LinkedIn, a third on Crunchbase. AI cross-references entity data across sources. When the data doesn't line up, the entity reads as ambiguous and doesn't get cited. Reconciling brand data is a required step in our process.
ChatGPT and Perplexity weigh recency and authority heavily in 2026. Content with no datePublished, no author with a verified bio, no updates — it loses its shot at being cited.
An update within 30 days drives 3.2× more citations. ConvertMate GEO Benchmark 2026
"Let's crack ChatGPT first, then think about the rest" doesn't work. Only 11% of domains get cited in both ChatGPT and Perplexity. Multi-platform work from day one gives the best composite result.
You can hold the #1 spot in Google and still never appear in the ChatGPT answer for the same query.
Overlap between Google top results and AI-cited sources dropped from 70% to under 20% across 2024–2025. Brandlight / 5W Research, May 2026
Prices are locked at the time of booking. Bundle a new site with optimization and you get a discount. We agree on the amount once we know the scope.
at +60% traffic growth, 1.5% conversion, €300 per customer · An estimate, not an offer
If your question isn't here, write the CTO or CEO directly. We reply within 24 hours.
GEO (Generative Engine Optimization) is optimization for brand mentions inside ChatGPT, Perplexity, Claude, and Gemini answers. SEO works on positions in Google and Bing; GEO works on showing up inside the AI answer. The Princeton/Georgia Tech paper (KDD 2024) found 40% of GEO ranking factors overlap with classic SEO — the other 60% are extraction, authority, and brand-mention signals. GEO sits on top of SEO; it doesn't replace it.
No. AI systems pull from the Google and Bing index plus real-time crawling — without classic SEO your content never reaches them. What's changing isn't SEO's importance but how answers get consumed: part of the time the user reads the answer inside the AI and never visits a site. So SEO stays the foundation, and GEO adds visibility inside AI answers on top of it.
SEO — positions in Google and Bing. AEO (Answer Engine Optimization) — landing in direct answers and featured snippets inside the search engine. GEO — brand mentions in generative answers (ChatGPT, Perplexity, Gemini). LLMO — the technical sub-layer of GEO: Schema.org, semantic HTML, crawler access. In practice we run all four — they only work together, not apart.
If you're wondering how to rank in ChatGPT search, the mechanics are different from Google. ChatGPT cites top Bing results 87% of the time, so being indexed in Bing plus fast indexing through IndexNow is critical. After that: content structured for extraction (answer capsules, FAQ, tables, dates and authors) and brand mentions on authoritative sources. Nobody can guarantee placement — same as a #1 in Google — but you can systematically raise the odds and track Share of Model.
First LLM mentions can show up 4–6 weeks after the technical foundation and a handful of core pages are live. Stable presence in AI answers takes 2–3 months. Serious lift on AI citations and classic SEO positions usually lands by the end of the project. Full payback on the investment: 7–12 months. SEO and GEO buy an asset. They don't rent traffic.
llms.txt, and do I need it?
llms.txt is a pointer file for AI crawlers — a working draft, not an approved standard. We ship it on request, or when it makes clear sense for the client — documentation-heavy sites, technical resources. We don't treat it as critical and we don't build strategy around it. No large study has shown a real lift from it yet.
Through custom referral channels. In GA4 Admin we create channel groups for specific domains: chatgpt.com, perplexity.ai, claude.ai, copilot.microsoft.com, gemini.google.com. After that, your reports show a dedicated AI Traffic channel broken out per platform. For deeper analysis we layer in custom dimensions for query type, entry point, and conversion path.
Good — they cover part of the work. We can run in parallel and focus on GEO and LLMO as the layer on top. At kickoff we do a short audit of what's already been done in SEO and align with their process. No duplication: they hold the Google positions, we add the AI visibility.
Yes. aibot.pro is a live example of the techniques we ship: Schema.org via JSON-LD, FAQ blocks in the right format, an answer capsule after every H2, entity data matched across LinkedIn and the directories. It's the work we'd do for a client, running on our own site.
A technical SEO audit of your site plus 10 priority queries for your category run through ChatGPT, Perplexity, and Google Gemini. We show you where you're cited, where your competitors are cited, and the gaps in between. Free, no strings.