For twenty years, getting found meant ranking on Google. That is changing fast. Over 800 million people now ask an AI engine what to buy or who to hire before they ever open a search box, and the answer they get is a short, confident recommendation - usually naming one or two brands. If you are not one of them, you are invisible at the exact moment the decision is made.
The good news: the signals these models use are knowable, and most of your competitors are not optimising for them yet. The practice of earning those recommendations is called generative engine optimisation, or GEO. Here is the checklist we run for clients, in the order that matters.
- AI engines recommend brands based on structured data, third-party citations and content depth, not ad spend.
- Start by auditing what the models say about you today, then fix the foundations before chasing volume.
- Schema and technical fixes can move within weeks; citations and authority compound over months.
- Visibility is not a one-time fix - it has to be tracked and maintained as engines and competitors move.
Why this matters now
AI search rewards early movers more than traditional search ever did. When a model has thin information about a category, it leans heavily on whatever credible signals exist. The brand that establishes clear entities, earns the first citations and publishes the reference content becomes the default answer - and that position is hard for latecomers to dislodge. Acting while your category is still sparse is far cheaper than fighting for an established slot later.
The five-step GEO checklist
Audit how AI engines see you today
You cannot fix what you cannot see. Run the buyer questions in your category across ChatGPT, Perplexity, Gemini and Google AI Mode, and record where you show up, where a competitor shows up instead, and which sources the models cite. This baseline tells you whether your problem is recognition, credibility or content. If you would rather not do it by hand, our GEO Audit does exactly this across 50+ prompts and four engines.
Fix your structured data and entities
Language models read structured data to understand who you are and why you are credible. Implement clean JSON-LD using Organization, Person and Service entities, linked together in an @graph, and keep your name, address and details consistent everywhere they appear. This is the single highest-leverage fix because it is fully in your control and engines re-crawl it quickly.
Earn citations in the sources AI trusts
AI engines do not invent recommendations - they synthesise them from third-party sources. The work is to find which industry publications, forums, review sites and reference pages get cited in your category, then earn genuine mentions there. A handful of citations in the right sources beats hundreds of generic backlinks that no model is reading.
Publish content written to be quoted
Map the real questions buyers ask AI in your category, then publish clear, structured pages that answer them directly. The goal is not to rank a 2,000-word essay; it is to give a model a clean, confident sentence it can lift about you. Close the gaps between the questions being asked and the pages you actually have.
Track, measure and keep shipping
AI visibility decays. Engines update, competitors publish, and new buyer prompts appear every month. Track your mentions, share-of-voice and citations weekly, watch for sudden drops, and keep shipping schema, citations and content. The brands that win treat GEO as an ongoing position, which is the whole idea behind a GEO retainer.
Want to see what AI says about you?
Our GEO Audit shows exactly how ChatGPT, Perplexity, Gemini and Google AI Mode talk about your brand today, with 12 prioritised fixes ranked by impact. From $297, yours to keep.
How long until it works?
Set expectations by signal type. Schema and technical fixes can show movement within two to four weeks as engines re-crawl. Content additions typically start influencing answers six to twelve weeks after they are indexed. Citation and authority work is the slowest but most durable, compounding over three to six months. A good programme sequences the quick wins first so you see early signs of life while the compounding work builds underneath.
The brands that get recommended by AI are not the ones spending the most. They are the ones the models can most easily understand and verify.
Where to start
If you only do one thing this month, do step one: find out what the models actually say about you. Everything else follows from that baseline. From there, fix the foundations in your control, earn a few well-chosen citations, and keep shipping. If you would rather have it handled end to end, that is what we do - see how the GEO retainer works or book a free strategy call.