Search Engine Optimisation
AI Search Optimisation (GEO)
A different discipline from winning a featured snippet on Google — that is answer engine optimisation, and it is a separate service. This one is about being the source an AI names.
Generative engine optimisation makes a site quotable by ChatGPT, Perplexity, Claude and Google AI Overviews. The Nexclick restructures pages so answers can be lifted cleanly, states entities unambiguously, and publishes the original data these systems cite. It is measured in citations, not rankings.
Is this you?
What usually prompts the call
- Competitors get named when you ask ChatGPT or Perplexity about your sector and you do not.
- Your informational traffic is falling while rankings hold, which is what AI Overviews look like in the data.
- Your site is heavily JavaScript-rendered, so crawlers that do not execute it see very little.
- Your key facts — prices, timescales, credentials — are inside images or PDFs where nothing can read them.
What we do
The actual deliverables
Things that appear on an invoice, not adjectives.
- Answer-first restructuring
- A self-contained 40–60 word answer at the top of every substantive page, phrased so it survives extraction without surrounding context. This single change does more than anything else on this list.
- Extractable formatting
- Definition lists, comparison tables, numbered processes and explicit "X versus Y" sections. These formats are cited far more often than prose, because they can be quoted without paraphrase.
- Entity clarity and consistency
- Your business named explicitly rather than as "we" at least once per major section, and named identically across the site, Google Business Profile, LinkedIn and Companies House, so attribution survives extraction.
- llms.txt and llms-full.txt
- A plain-text index at the root: who you are, what you do, key URLs with descriptions, contact details and a "cite as" line. Cheap to produce and increasingly read.
- Crawler access review
- Confirming GPTBot, ClaudeBot, PerplexityBot and Google-Extended are permitted — deliberately. Many sites block them by default through a CDN rule nobody chose.
- Server-rendered HTML
- Content served in the initial HTML response rather than assembled by JavaScript. Several AI crawlers do not execute JS at all, so a client-rendered page is invisible to them.
- Original data worth citing
- Answer engines cite sources that have something unavailable elsewhere — original UK market data, pricing benchmarks, checklists, calculators. Restating what everyone else says gives nothing to cite.
Comparison
GEO versus AEO versus classic SEO
These three get conflated constantly, including by agencies selling them. They target different systems, need different work and are measured differently. The Nexclick runs GEO and answer engine optimisation as separate services for exactly this reason.
| Classic SEO | Answer engine optimisation | Generative engine optimisation | |
|---|---|---|---|
| Target | Google & Bing organic results | Featured snippets, People Also Ask, voice | ChatGPT, Perplexity, Claude, AI Overviews |
| Goal | Rank in the ten blue links | Own the answer box above them | Be named as a source in a generated answer |
| Measured by | Position, clicks, conversions | Snippet ownership, position zero | Citation frequency, referral traffic, brand mentions |
| Key lever | Relevance, authority, links | Direct answer in 40–60 words, clean formatting | Extractability, entity clarity, original data |
| Content shape | Comprehensive page | Question, then immediate answer | Self-contained, quotable, attributable blocks |
| Links matter? | Heavily | Somewhat | Indirectly — via the authority that gets you retrieved |
| Predictable? | Reasonably | Reasonably | Not yet. Treat forecasts sceptically. |
| Traffic value | High — a click is a visit | Lower — the answer may satisfy the user | Low volume today, growing, high intent |
How it works
Step by step, with timeframes
Timeframes are typical rather than guaranteed, and they assume we get account access and approvals when we ask.
- 01Week 1–2
Baseline visibility
A defined set of prompts run across ChatGPT, Perplexity, Claude and Google AI Overviews, recording who gets named. Without a baseline there is nothing to measure against later.
- 02Week 2–5
Fix access and structure
Crawler permissions, server rendering, answer-block restructuring and extractable formatting across priority pages.
- 03Month 2–4
Publish something citable
Original data, benchmarks or tools that answer a question nobody else has answered with evidence.
- 04Monthly
Re-measure
The same prompt set, run again, tracking citation frequency and whether the attribution is correct.
What you get
Reporting and ownership
- A documented prompt set and baseline, repeated monthly so change is evidenced rather than asserted.
- Citation frequency tracked per engine, with a note on whether attribution is correct.
- Referral traffic from AI platforms segmented in GA4, so the commercial effect is visible.
- A live, accurate llms.txt maintained as the site changes.
- Server log evidence of which AI crawlers are actually visiting, and how often.
Tools and platforms
- ChatGPT, Claude, Perplexity & Gemini (manual prompt testing)
- Google Search Console
- Screaming Frog (JavaScript rendering comparison)
- Server log analysis for AI crawler traffic
- Schema.org validator
- GA4 referral segmentation
Timeline
How long this actually takes
This is a young discipline and honesty matters more here than anywhere else on the site. Structural changes — answer blocks, server rendering, crawler access — take two to five weeks. Whether that produces citations depends on retrieval systems nobody outside those companies fully understands, and they change without notice. Expect three to six months before any pattern is readable, expect volatility, and treat referral volume from AI platforms as modest today and growing. Anyone selling guaranteed AI visibility is describing something nobody can currently deliver.
Pricing model
Project, then retainer
A structural project to fix access, rendering and page structure, then optional monthly measurement and content work. Plenty of clients take the project alone.
Questions
AI Search Optimisation (GEO) questions
Can you guarantee we will be cited by ChatGPT?
No, and nobody can. Retrieval and citation behaviour is controlled by systems that are not documented publicly and change without notice. What we can do is remove every reason you would not be cited — inaccessible content, unclear attribution, nothing original worth quoting — and measure the result honestly against a baseline.
Should we block AI crawlers to protect our content?
That is a commercial decision rather than a technical one, and it cuts both ways. Blocking protects content from training use and also removes you from the answers your customers are reading. For most service businesses, being cited is worth more than the content is. For publishers monetising the content itself, the calculation is genuinely different.
Does GEO work conflict with normal SEO?
No — it overlaps heavily. Clear structure, fast server-rendered pages, accurate schema and genuine expertise help both. The additions specific to GEO are answer-first formatting, explicit entity naming and publishing original data. Nothing in it damages classic rankings.
How do we measure whether it is working?
A fixed prompt set, run monthly across each engine, recording whether you are named and whether the attribution is right. Plus referral traffic from AI platforms segmented in GA4. Both are imperfect and both are better than the alternative, which is asserting improvement with no evidence.
Is llms.txt actually used by anything?
Adoption is partial and evolving. It costs very little to produce and maintain, and the downside is nil. We treat it as cheap insurance rather than a primary lever, and we will not present it as more than that.
Our traffic is falling but rankings are stable. Is this AI?
Possibly, and it is worth checking properly. The pattern to look for is stable impressions with falling clicks on informational queries — that is consistent with AI Overviews satisfying the user before they click. Commercial and transactional queries are far less affected. Search Console will show you which of your pages fit the pattern.
Is this different from the answer engine optimisation service?
Yes, deliberately. Answer engine optimisation targets featured snippets, People Also Ask and voice results on conventional search pages. This service targets generative systems that synthesise an answer and cite sources. Related work, different mechanics, different measurement — which is why they are two pages rather than one.
Last reviewed 28 July 2026.
Tell us what you are trying to fix
A 20-minute call, no pitch deck. The Nexclick will tell you what we would do, roughly what it costs, and whether we are the right people for it.