When your customer asks ChatGPT, Claude, or Perplexity for a recommendation, they do not scroll through ten blue links anymore. They read a direct answer with three or four citations. Those citations are the new front page.

Generative Engine Optimization (GEO) is the discipline of becoming one of those citations. This piece explains what GEO actually is, how it differs from classic SEO, what we do inside a GEO engagement, why being early matters, and the honest tradeoffs.

The shift, in one paragraph

In 2022, a buyer Googled “best CRM for a small services business in Kenya” and skimmed five articles before deciding. In 2026, they ask ChatGPT or Perplexity the same question and get a synthesised answer that names three CRMs, with one short paragraph each, and citations. They click maybe one citation. Often zero. The decision is informed by what the AI said, which is informed by which sources the AI chose to read.

If you are not one of those sources, you are invisible. Not lower-ranked. Invisible.

What GEO actually is

GEO is the set of choices that make a brand citable by AI engines. It overlaps with SEO at the technical level (clean HTML, structured data, fast loads) but the optimisation target is different.

SEO optimises for being ranked. GEO optimises for being quoted.

The implications:

Short, factual passages win. AI assistants pull quotable sentences, not paragraphs. A page full of long marketing prose is less citable than a page with crisp definitions and specific numbers.
Citation-worthy authority matters more than backlinks. A mention on Wikipedia, in a podcast transcript, or in a respected industry directory carries more weight than ten generic backlinks. AI engines triangulate trust from sources humans already trust.
llms.txt has emerged as a soft standard. It tells LLM-powered systems how to read and summarise your site. It is not a silver bullet, but ignoring it is leaving signal on the table.
AI crawler access is a binary choice. GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended either can read you or they cannot. Most sites have not made an explicit decision; the default is often blocked.

Why being cited matters more than ranking, for some queries

Not all queries shift to AI search equally. The shift is strongest for:

Comparison and recommendation queries (“best X for Y”, “X vs Y”, “alternatives to X”)
How-to and explanation queries (“how does X work”, “what is X”)
Local recommendation queries (“best X in [city]”)
Buying-research queries (“things to consider when picking X”)

For pure transactional queries (“buy X”, “X pricing”), classic search engines still dominate. For navigational queries (“acme corp login”), search engines also win because the user knows where they are going.

The mistake is treating GEO and SEO as competing investments. They are not. GEO captures the comparison and research stages of the buyer journey, where the buyer is forming opinions. SEO captures the buyer once the opinion is formed and they are ready to act. You want both.

What we do inside a GEO engagement

A standard engagement runs 4 to 12 weeks for the initial work, then ongoing if the client wants monthly tracking and authority work.

Phase 1, Audit

We run a panel of representative prompts (the prompts your buyers actually ask) across ChatGPT search, Claude, Perplexity, Gemini, and Google AI Overviews. We measure:

Where are you cited?
Where are competitors cited and you are not?
What sources do the AI engines pull from when answering in your category?
What are the gaps between your content and what gets cited?

The output is a one-page citability scorecard plus a prioritised opportunity map.

Phase 2, Access and structure

The plumbing work that determines whether AI engines can even read you. Mostly invisible to humans, mostly the highest leverage:

robots.txt updated to explicitly allow AI crawlers
llms.txt written and deployed at the root
Schema.org JSON-LD: Organization, Service, FAQPage, Article, BreadcrumbList
Internal linking patterns that help retrieval
Canonical URLs, sitemap, redirect cleanups

Phase 3, Citability rewrite

The content work. We identify the 10 to 30 highest-value pages, then rewrite each so the most quotable passage is:

Short (one to three sentences)
Specific (with names, numbers, dates, places)
Factual (verifiable, attributable)
Clearly your brand (not generic)

This is the part most clients underestimate. Generic marketing copy is almost never quoted. Specific, opinionated, factual passages are.

Phase 4, Off-site authority

AI engines cite sources that humans already trust. The off-site work targets:

Quality industry roundups and listicles (“best X for Y” articles)
Podcasts in your category (a guest appearance with a clear transcript)
Directories AI engines actually index
Comparison sites
Wikipedia (where genuinely warranted; we do not abuse this)

This compounds over 3 to 6 months. It is the slowest part of GEO and the part with the longest tail.

Phase 5, Measure and iterate

Monthly tracking of:

Citation appearances across the major AI engines
Citation share versus competitors
Referral traffic from AI search (still small but growing)
Movement on the prompts we identified as priorities

The report is one page. The action plan is two lines. We double down on what is working.

A concrete scenario: the niche B2B SaaS

A small B2B SaaS in the HR-tech space came to us frustrated. Their Google SEO was strong, they ranked top-three for their bread-and-butter keywords. But when buyers asked ChatGPT “what’s the best HR software for a 50-person African company”, they did not appear. Two younger, smaller competitors did.

The audit found three problems:

Their site blocked GPTBot via the default robots.txt of their CMS.
Their best-performing pages had hero-banner marketing copy at the top. The quotable, factual content was buried below the fold and inside long paragraphs.
They had almost no off-site mentions in the directories ChatGPT cites for HR-tech recommendations.

We fixed all three over 8 weeks. Robots.txt updated. The 12 highest-value pages rewritten so the first 200 words contain the specific, citation-ready facts. A guest spot booked on a respected HR-tech podcast with a published transcript. A listing in three industry directories.

Six weeks after the rewrite, ChatGPT and Perplexity started citing them by name for the priority prompts. Three months later, citation share for their category prompts had gone from 0% to roughly 35%. The marketing director told us it was the highest-ROI quarter she had run.

Why being early matters

The training data and ranking heuristics of major AI engines bake in patterns that are sticky. Brands that are cited often today tend to be cited often tomorrow, because:

AI engines lean on prior citation history as a trust signal
Authority sources (Wikipedia, major directories, podcasts) update slowly, so an early entry stays in place
Each citation is a small positive feedback loop, increasing the probability of the next one

The cost to claim a slot now is a fraction of what it will be in 18 months when the field is crowded. This is one of the very few channels where being early is a real, measurable advantage rather than a vague “first-mover” cliche.

Frequently asked questions

Is GEO replacing SEO?

Not yet, they overlap, but the targets differ. Google AI Overviews still uses much of the SEO graph. ChatGPT and Perplexity rely on web search plus their own preferences. We optimise for both, but the GEO-specific work is what most agencies miss because they do not understand the new optimisation target.

How do you measure GEO results?

We run a panel of representative prompts monthly across the major AI engines and track citation share. We also use third-party citation trackers (Profound, Otterly) and direct attribution from referral traffic where it exists.

What is llms.txt?

An emerging standard, similar in spirit to robots.txt, that tells LLM-powered systems how to read and summarise your site. Adoption is incomplete (not every AI engine respects it yet), but the leaders are starting to. The downside of having one is essentially zero; the upside is real and growing.

How long until I see results?

On-site changes (access, structure, citability rewrite) can move citations in 4 to 8 weeks. Off-site authority work compounds over 3 to 6 months. The combined picture usually starts showing in months 2 to 3.

Does WPfoss still do classic SEO?

The technical, structural, and content fundamentals of SEO are folded into our GEO engagements and our AI-Powered Website Development service. We do not sell SEO as a standalone retainer because we believe the future of search is hybrid (AI plus classic) and serving only the classic side is leaving most of the value on the table.

What does a GEO engagement cost?

A one-time audit and rewrite engagement runs $5k to $15k depending on site size. Monthly authority and tracking runs $1.5k to $4k per month and is where the compounding happens. We give a firm quote after the audit.

Will this affect my Google rankings?

Almost always positively. The technical work for GEO overlaps heavily with technical SEO. We have not seen a case where GEO work hurt Google rankings. We have seen many where it helped.

The honest pitch

GEO is one of the very few channels where being early matters. The brands that are cited inside ChatGPT, Claude, and Perplexity in 2027 will mostly be the brands that started working on it in 2025 and 2026. The cost of being late is exclusion from a growing share of buyer conversations.

Order a GEO engagement or book a free call and we will run the audit and show you your starting position.

GEO: how to get cited by ChatGPT, Claude, and Perplexity