Our WorkServicesRevenue EngineeringZatrovo BookingAI VisibilityStart a Project →
Generative Engine Optimisation

BethebrandAIrecommends.

Search has changed. Your guests now ask ChatGPT, Perplexity and Google AI Mode which hotel to book, which spa to visit, which club to join. We engineer the answer.

ChatGPTPerplexityGeminiClaudeAI OverviewsAI ModeCopilotMeta AIGrok
Book an AI Visibility Audit →
$1,150 · fixed scope · two-week turnaround

900M

weekly active users now ask ChatGPT before they ask Google

OpenAI, Feb 2026

8%

click through rate on results with an AI summary, half the traditional rate

Pew Research, 2025

4.4×

higher conversion rate from LLM-referred visitors than organic search

Semrush, 2025

9%

of AI responses about a brand contain a fabricated fact

Four Dots Hallucination Index, 2025

The Shift

The result page is being replaced by the answer.

The page is now the answer itself. AI Overviews trigger on roughly half of all Google searches; Perplexity, ChatGPT and Claude have eaten the discovery layer for high-intent buyers. Generative Engine Optimisation decides whether your brand is named, fairly described and recommended inside those answers — through entity engineering, schema architecture and citation strategy. Done late, your competitors own the answer.

The Surfaces

Eight engines. Eight ranking models.

Citation overlap between ChatGPT and Perplexity is roughly 11%. AI Overviews and AI Mode share citations 13.7% of the time. A single channel strategy fails by design. Every engine needs its own playbook.

ChatGPT

900M weekly users. OpenAI index + Bing fallback.

Google AI Overviews

Triggers on ~48% of searches. Citation overlap with organic now 54%.

Google AI Mode

Fan-out querying, 16 parallel sub-queries. 13.7% citation overlap with AIO.

Perplexity

Velocity-weighted. Fresh content (<14 days) lands in top citations 72% of the time.

Claude

Skews technical. Rewards primary sources and structured documentation.

Microsoft Copilot

Inherits Bing ranking. Over-indexes on LinkedIn and Microsoft properties.

Meta AI

Llama + Bing. Pulls from Reddit, Facebook and Instagram public content.

Grok

Real-time X firehose. Recency on X is the dominant signal.

How We Build It

Diagnose. Engineer. Earn. Monitor.

Six modules across four disciplines. Each one produces a named artefact your team can inspect, review and own.

01Diagnose

AI Visibility Index + Prompt Universe

A named, scored diagnostic across ChatGPT, Perplexity, Gemini, Claude, AI Overviews, AI Mode, Copilot and Meta AI. Mapped to the actual prompts your buyers type — discovery, comparison and high-intent queries.

02Engineer

Entity reconciliation + Knowledge Graph

Wikidata claims, Knowledge Panel acquisition, sameAs linking, entity disambiguation. Brands with verified Wikidata are 3.2× more likely to appear in AI Overview citations.

03Engineer

Content + schema architecture

Passage-level content engineered for retrieval (80–100 word answer blocks). JSON-LD layered for sector — LodgingBusiness, SpaBusiness, GolfCourse — with Review, FAQPage and Speakable nodes.

04Earn

Citation + source placement

Tier-1 PR (Condé Nast Traveler, Robb Report, FT, Bloomberg), trade press, YouTube transcripts, Reddit and review-platform corroboration. Unlinked brand mentions now correlate ~3× more strongly with AI visibility than backlinks.

05Engineer

Crawler + render substrate

GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot governed in robots.txt. Critical content shipped server-rendered — these crawlers do not execute JavaScript. The plumbing most agencies miss.

06Monitor

Share-of-voice + perception drift

Weekly tracking across all eight engines. Visibility, position, sentiment, hallucination patrol. Perception-drift alerts when descriptors shift. Monthly report tied to the prompts that drive direct bookings.

The Technical Substrate

The plumbing most agencies miss.

An AI visibility programme that does not address crawler policy, render guarantees, schema validity and entity reconciliation is a content programme with a new logo. Here is what we audit and ship at the infrastructure layer.

Crawler policy

GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Google-Extended and ChatGPT-User are each governed separately in robots.txt. Block training crawlers, allow retrieval crawlers, never the other way around.

Render guarantees

GPTBot, ClaudeBot and PerplexityBot do not execute JavaScript. Critical content must ship in the initial HTML response. Server rendered, not hydrated client side.

Schema layering

JSON-LD remains the only structured data format every major engine ingests reliably. We layer Organization, Service, FAQPage, Speakable, Review and sector specific types, and validate them against each engine's extraction model.

The Cost of Silence

If you don’t shape the answer, the model invents one.

Hallucinated facts

Roughly one in eleven AI responses about a brand contains a fabricated detail. Wrong pricing, wrong location, a service you don’t offer. At 10,000 monthly mentions, that is 900 prospects misinformed.

Competitor capture

When users ask about your category, or worse your name, and the AI cites a competitor, the answer is the booking. The retraction window is the next training cycle.

Direct booking erosion

AI assistants increasingly route to OTAs by default. Without a deliberate citation strategy, every AI-mediated enquiry costs you commission you should never have paid.

Live Case Study

Zatrovo

Our flagship SaaS reaches 13,000+ members across multiple studios. Structured data, passage level content and a citation programme built for the way modern search actually works. Surfaced when buyers ask AI for “the best booking platform for boutique studios.”

Read the Zatrovo case study →
Zatrovo · Built for studios
Engagement Models

Three ways to begin.

AI Visibility Audit

$1,150

one-off

Full AI Visibility Index across 8 engines, prompt universe map, hallucination audit, competitor share-of-voice, prioritised remediation plan. Fixed scope, two-week turnaround.

Most chosen

AI Visibility Project

from $3,000

project

The full Diagnose → Engineer build. Entity and schema architecture, content engineering, citation source plan, technical infrastructure, and a 90 day authority brief handed to your PR team, or ours.

AI Visibility Retainer

from $950

per month

Ongoing monitoring, monthly content cadence, citation outreach, perception-drift alerts and quarterly strategy reviews. Built for brands that intend to defend their category.

Indicative pricing. Project and retainer scopes are sized to the brand, the category and the competitive set. Larger groups, multi-property estates and bespoke engagements are quoted individually.

Or bundled into our flagship engagement, Revenue Engineering →

Frequently Asked

The questions buyers ask first.

How is this different from SEO?+

SEO optimises for the ranked list of blue links. GEO optimises for the answer the model writes. Different ranking models, different inputs, different artefacts. There is overlap. Both reward authoritative content and good schema. But citation logic, surface coverage and measurement systems are distinct. Most agencies branding "AI SEO" are doing the same on page work they always did and hoping it lands.

How long until I see results?+

The Visibility Index baseline ships in two weeks. Schema, entity reconciliation and crawler policy take effect four to six weeks after deployment. Citation share typically begins shifting after 60 to 90 days of corroborated signals across 8 to 12 independent domains. Anyone promising rank inside 30 days is selling air.

Do I need an existing PR or content team?+

No. We can run the citation programme, brief your team if you have one, or hand the 90 day authority brief to your PR partner. Most clients start with us running it and shift to internal ownership over months six to twelve.

What if the AI engine has wrong information about my brand?+

Hallucinations are common. Roughly one in eleven AI responses about a brand contains a fabricated fact. We engineer the substrate (entity reconciliation, schema, authoritative source corroboration) so the model has a stronger prior to draw from. Wrong descriptors fade as new corroborated ones compound. Typical drift correction window is 60 to 120 days.

Is llms.txt worth implementing?+

No major model provider has confirmed production ingestion in 2026. We implement it because it is cheap insurance, not because it moves the needle. Belt and braces, not strategy.

Can this be bundled with broader marketing work?+

Yes. The full GEO programme is included inside our flagship Revenue Engineering engagement at the Growth tier and above, alongside website, booking, SEO, reviews and reporting. Useful when you want one team accountable for the whole funnel.

Theanswerisbeingwritten.Makesureyourbrandisinit.

Start with a fixed scope AI Visibility Audit. Two weeks. One named report. A clear view of where your brand stands across every engine that matters.

We respond within 24 hours.

Generative Engine Optimisation · GEO · AEO

ChatGPT · Perplexity · Gemini · Claude · AI Overviews · AI Mode · Copilot · Meta AI · Grok

Start a Project