The 2026 AI Visibility Playbook: how to rank in ChatGPT, Gemini and Perplexity

Generative search optimisation (GEO) is the new SEO. In 2026 an increasing share of high-intent traffic never touches a blue link – the answer arrives directly, cited from a handful of sources the model trusts. If you are not one of those sources, you are invisible.

This playbook is a 47-point checklist for getting your brand cited across the four LLM search surfaces that matter right now: Google AI Overviews, ChatGPT search, Gemini, and Perplexity. It is structured as three passes – foundations, prompt-shape content, and measurement – so you can start where the leak is worst.

Why AI visibility is different from SEO

Classic SEO optimises for a click. GEO optimises for a citation. The moves overlap – clean semantic HTML, real authorship, structured data – but the failure modes are different. A page can rank first in Google and never appear in an AI answer if the model cannot cheaply extract a self-contained claim from it.

• AI Overviews cite roughly 27% of English searches globally – treat that as the new organic top-3.
• ChatGPT search and Perplexity favour sources with a stable canonical URL and llms.txt.
• Gemini rewards structured data and pages with a clear author + publication date.

The three passes

Pass 1 – Foundations

Before you rewrite anything, make sure the model can find and trust you. Publish an llms.txt at the site root, declare your author schema, and add a stable canonical to every article. If any of the three are missing, downstream work leaks.

1. Publish /llms.txt at the site root.
2. Add Article + Person schema with a real, resolvable author URL.
3. Set a canonical on every content page – no query-string variants.
4. Verify your brand entity in Google Knowledge Graph.

Pass 2 – Prompt-shape content

LLMs cite claims that answer a question in isolation. Rewrite your top 20 pages so the answer sits above the fold, followed by evidence and a link to your source. Bullet lists, tight tables and numbered steps all lift citation rates in our tests.

Pass 3 – Measurement

You cannot optimise what you cannot see. Track your citation share across the four surfaces weekly. When a competitor overtakes you on a prompt, note the shape of their cited page – that is your next rewrite brief.

A worked example

# /llms.txt
User-agent: *
Allow: /

# Full content index
Full-content: /llms-full.txt

# Brand and author
Brand: SEMOptimiser
Author: Sujan Studio

The teams winning at AI visibility in 2026 are the ones treating every top article as a prompt-answer pair, not a keyword-density exercise.

The 47-point checklist

Below is the full checklist condensed into a table you can copy into your project tracker. Each row is one shipping decision, not one rewrite – expect a good team to close 6-10 items in a sprint.

What to do next

Run a citation audit on your top 20 pages this week. If we detect fewer than 3 citations across the four surfaces, you have a foundations problem – start with Pass 1. If you are cited on 3-10 pages, you have a shape problem – Pass 2. Anything above 10 is a measurement problem – start reporting weekly and defend your ground.