Find competitors of a website: a domain-first walkthrough

To find competitors of a website, treat it as a data problem: collect the domains that rank for the same searches, count them, and filter out the noise. You can do it with a browser and a CSV, or with one domain lookup. This walkthrough does both, so you can see what the tool is doing for you and where its output needs a second look.

The manual pass

Start with a short query file. Ten to twenty lines is enough:

query,type
invoice api for saas,product
stripe invoice alternative,product
how to generate pdf invoices,problem
send invoice from node,problem

Mix product queries with problem queries. Developers search problems far more than product names, and the sites answering those problem queries are competitors for your docs and blog even when they ship nothing like your product.

For each line, open a private window, set the region to your main market, and record the top ten domains. One row per domain per query gives you a long table. Group by domain, count rows, sort descending.

Two filters before you trust the result:

• Drop the giants. GitHub, Stack Overflow, Wikipedia, Reddit and YouTube rank for almost everything. Keep one only if it holds first place on most of your queries.
• Keep the publishers. A tutorial site or a "best invoice APIs" roundup that appears on eight of your queries is a real search competitor. It takes the click before your page gets a chance.

This scales to a few dozen queries. After that, maintaining the query file becomes its own project.

The domain-first pass

A keyword-overlap lookup flips the process. Instead of starting from queries you guessed, it starts from every Google keyword the domain already ranks for and finds the sites that share the most of them.

seogeoaeo.ai's Competitor Finder is the tool we use for this. Input is a domain and a country. A full URL works too; it gets reduced to the domain. You can point it at your own site or at someone else's.

The output is 30 sites. For each one you get:

The share column does the filtering you did by hand. A site where at least 5% of its keywords are also yours is a close competitor. A site that shares many keywords with you but ranks for millions of others gets put with the large general sites. That is the same "drop the giants" rule, applied with numbers.

A run costs 38 credits, and the price is shown before it executes. There is also an API and MCP server, so the same lookup can run from a script or an AI coding agent.

Reading the result critically

Three things to check:

1. Coverage. The lookup reads Google rankings in one country for the whole domain. If you sell in several markets, run it per market. If your docs live on a subdomain or a folder you care about most, the result still reflects the whole domain.
2. Missing rivals. Companies that win through sales, partnerships or ads alone will not show up. Cross-check against ad libraries (Google Ads Transparency Center, Meta Ad Library, LinkedIn Ad Library) and against what your users say they evaluated.
3. Small sites. A new project that ranks for a handful of terms shares them mostly with huge sites, so it may get no close competitors. The tool page says so, and suggests running Competitor Keywords on a rival you already know instead.

Then what

Take the three closest rivals into Competitor Keywords (35 credits) to see the top 100 keywords bringing each one traffic, or into Keyword Gap (85 credits) to list only the top-20 keywords they have and you do not. That is where the list starts turning into a backlog of pages to write.

If you want the broader method, including AI assistant answers as a third source of rivals, the original guide on seogeoaeo.ai covers it. The manual pass is still worth doing once: it teaches you which of the tool's rows to trust.