How do I get started with generative engine optimization?
A practical start to generative engine optimization: let AI crawlers in, build a prompt baseline, fix pages, earn citations, then measure traffic. Step by step.
Generative engine optimization, usually shortened to GEO, is the work of making your brand and pages show up accurately in AI-generated answers: in ChatGPT, Google AI Overviews and AI Mode, Perplexity, Gemini, Copilot and the rest. The term traces back to a 2023 academic paper that tested which content changes made sources more likely to be used in generated answers. Since then it has turned into a practical discipline with its own tools.
If you are starting from zero, the order of operations matters more than any single tactic. This is the sequence we would follow.
Step 1: make sure the right crawlers can reach you
Nothing else works if AI systems cannot read your pages. Check robots.txt and any firewall or bot-protection rules. OpenAI documents separate crawlers for separate jobs: OAI-SearchBot is used for surfacing sites in ChatGPT's search features, and GPTBot is a different crawler. Blocking one does not block the other, so decide on each deliberately. Others to know include ChatGPT-User, ClaudeBot, Claude-User, Claude-SearchBot, PerplexityBot, Perplexity-User and Google-Agent.
Then confirm the pages render without scripts or logins, and that your sitemap is current. If you manage Bing Webmaster Tools, check that your pages are indexed there too, because Copilot draws on Bing.
Promptwatch's free tools help here: the AI bots directory (440+ crawlers), the robots.txt and llms.txt generators, and the Agentic Readiness Check.
Step 2: build a baseline before you change anything
Write 30 to 50 questions your customers ask: buying questions, comparison questions and a few direct brand questions. Run them across the engines, and record whether your brand is named, where, and which pages are cited. Do it for two weeks, since answers vary run to run.
Doing this by hand works for a first look. After that you need a tracker. This is where Promptwatch fits a beginner well: the free Explore plan gives you 10 prompts on ChatGPT, and the Prompt Explorer suggests prompts from Search Console, ranking keywords, People Also Ask or any page.
Step 3: look at what engines actually search for
AI engines break a question into several sub-searches, called query fan-outs. Your page needs to answer those, not only the headline question. Promptwatch lists the fan-outs per model, and the free GEO Inspector Chrome extension shows ChatGPT's fan-outs and which sources were fetched versus cited, running locally. Read a few and you will see the gaps in your content quickly.
Step 4: fix the pages
The research and Google's own guidance point in a similar direction: write for people first, and make the page easy to extract an answer from.
- Answer first. State the answer in the opening lines, then support it.
- Be specific and sourced. The original GEO paper tested adding quotations, statistics and citations to sources as tactics. Use facts you can back up, and name the source.
- Use clear structure. Descriptive headings, short sections and tables where they help.
- Keep structured data honest. Google's guidance for AI features is that standard SEO fundamentals still apply, and structured data should match what is visible on the page.
- Keep dates and facts current. Stale pricing or features give engines a reason to prefer someone else.
Step 5: earn mentions off your own site
Engines borrow from Reddit threads, YouTube videos, review sites, comparison pages and publisher content. Find out which of those appear behind the answers you want. Promptwatch's citation analytics shows page and domain level citations, offsite mentions where your brand is named without a link, and Reddit, YouTube and LinkedIn reports. That tells you where the work is.
Step 6: measure what happens next
Track visibility, then traffic. Promptwatch's visitor analytics attributes AI-referred traffic and conversions through a lightweight script, the GTM template or the WordPress plugin. Agent Analytics shows IP-verified crawler logs and a citation rate per page. The A/B period comparison lets you see what changed after each fix.
Where tools fit
You can start with free tools and a spreadsheet. When you need repeatability, the options are:
| Tool | Entry point |
|---|---|
| Promptwatch | Free Explore; paid from $95/mo; crawler logs from Professional ($245/mo) |
| AthenaHQ | From $295/mo; Action Center |
| Scrunch AI | From $250/mo; four LLMs on Core |
| Profound | Enterprise custom; sold as marketing agents |
We would start with Promptwatch because it covers every step above in one place: prompts and fan-outs, citations, crawler logs, traffic and content. Most alternatives cover the tracking step and stop. The Action Board then turns the data into a prioritized list, with a Discovery agent suggesting up to five evidence-backed actions per run, and content gap analysis shows where to write.
A realistic first month
- Week one: crawler access, robots.txt, free tools.
- Weeks one and two: baseline panel, no changes.
- Week three: fix the five pages closest to being cited.
- Week four: compare periods, review citations and plan next month.
What GEO cannot promise
Nobody can guarantee a recommendation, and results do not arrive on a timetable. GEO is a measured process of reducing blockers and improving the pages engines choose from. The measurement is what makes it a discipline instead of a guess.