Home · Methodology

Methodology

This page explains how we research, test and recommend — so you can judge our guides on their evidence rather than taking them on faith. GEO is a young field full of confident claims; here is exactly where ours come from and where they stop.

Which AI engines we check

When we talk about “getting cited in AI answers,” we mean the four surfaces most brands actually need to show up in today:

• ChatGPT — including its browsing and search modes, where answers cite live sources.
• Perplexity — which footnotes the pages behind each answer.
• Gemini — Google's assistant and its grounded responses.
• Google AI Overviews — the AI summaries that now sit above many search results.

These engines change their behaviour often, and an answer can differ by account, region and phrasing. So we describe patterns we can observe and reproduce — not guarantees.

What our recommendations rest on

Every ranking, comparison and “do this” in our guides rests on three things:

• Public pricing and product facts. When we compare tools, we use the vendors' own published prices, plans and feature pages — and we link to them so you can check.
• Public citation research. For claims about how AI engines pick and cite sources, we lean on published third-party studies — from analysts such as Ahrefs and BrightEdge, among others — and we attribute each one. We don't invent statistics, sample sizes or datasets of our own.
• Our own hands-on page tests. The founder applies the moves we write about to real pages — adding structure, an llms.txt file, clearer entities — then watches whether the engines' answers and citations actually change. This is qualitative, first-hand observation, and we present it as exactly that.

What this is not: there is no paid placement anywhere on this site. No vendor can buy a higher ranking, a kinder review or a spot in a comparison. If we recommend something, it earned the spot.

How often we update

AI search moves fast, and advice that was right last quarter can be wrong this one. So our default is to revise guides as the evidence changes — when a cited study is updated or superseded, when a vendor changes pricing, or when our own testing stops matching what a guide says. Where a page carries a published or modified date, that date reflects the last substantive review.

Affiliate and disclosure

Today we don't earn commissions from the tools we mention. In the future, some outbound links may become affiliate links — meaning we could earn a commission if you buy through them. If and when that happens, three rules hold:

• It will be disclosed clearly, on the pages where it applies.
• It will never change a recommendation. Rankings stay merit-based; an affiliate relationship is never pay-for-placement.
• We will keep recommending tools we don't earn anything from when they're the right call.

Found something wrong? Tell us

If a price is out of date, a study has been superseded, or a claim doesn't hold up, we want to fix it. Send corrections through our contact page and we'll review them — getting it right matters more than being right the first time.