Where Do Your Competitors' Best Keywords Hide, and How Do You Pull Them?

By the end of this guide you will have a working list of the keywords your real competitors win, the subset they win and you do not, and a ranked plan for which of those to publish first. I will also show you how to check the second layer most teams skip: which brands ChatGPT, Claude, Perplexity and Gemini name when a buyer asks a question your keywords should own.

Competitor keywords are the search queries a rival domain earns organic or paid traffic for. Finding them is one part of a competitive analysis, which is the broader practice of studying how rivals position, price, sell and acquire customers. Here we stay narrow: the keyword layer, done properly, in an order that saves you from the usual mistakes.

We run this process for our own site and for the topic clusters Gofylo builds, so the steps below come from repetition, not theory. Plan for an afternoon the first time and less on each refresh. What you need:

A keyword database tool with domain-level organic and paid reports (Ahrefs, Similarweb or a comparable index)
Your own Google Search Console property, with enough history to show stable query data
A spreadsheet or database where you can merge, dedupe and tag exports
Access to ChatGPT, Claude, Perplexity and Gemini for manual prompt checks
A list of the product categories and use cases you actually sell into

Competitor keyword work starts with picking the right rivals, not with the export button.

Step 1: Separate search competitors from sales competitors

Your sales team's competitor list is rarely your search competitor list. The vendor you lose deals to may publish almost nothing, while a review site, a niche blog or an adjacent tool owns every query your buyers type. If you analyze the wrong domains, every later step inherits the error.

Write down the five to ten head terms that describe your category and core use cases
Search each one in an incognito window and record every domain on page one
Count how often each domain appears across your head terms
Keep the domains that appear repeatedly, whether or not they sell a competing product
Split them into two tags: direct product rivals and content rivals (publishers, directories, communities)

Find competitor websites you did not know about

The fastest way to find competitor websites is to let the data surface them. Most keyword tools have a competing domains report that lists sites sharing the most organic keywords with yours. Run it on your domain, then run it again on your strongest known rival. The overlap between the two lists is usually your true competitive set.

Two more sources catch domains the tools miss. Ask ChatGPT and Claude to list the leading tools for your use case and note every brand they name. Then scan the review platforms and comparison pages your buyers read. A new entrant with a fast-growing blog often shows up in AI answers before it shows up in a keyword tool's competitor report.

Cap your set at three to five domains for the first pass. Pulling ten competitors produces an export too large to judge by hand, and you will end up filtering on volume alone, which is the mistake this whole process exists to avoid.

Step 2: Export each competitor's organic keyword set

Open the organic keywords report for each competitor domain and export the full list with position, estimated traffic, keyword difficulty, ranking URL and search intent where the tool provides it. Do this per domain, not per page, so you see the whole footprint.

Filter to positions 1–20 so you capture both current wins and pages close to winning
Remove branded queries that contain the competitor's name or product names
Keep the ranking URL column: it tells you which page format earns each keyword
Add a column with the competitor's domain so you can merge files later

Then sort by ranking URL and read the top pages. This is where the export becomes intelligence. You will often find that one competitor's traffic sits on a small set of templates: integration pages, alternative pages, glossary entries or calculators. That pattern matters more than any single keyword, because it tells you which page type the search engine rewards for this audience.

In practice, a project management tool might discover its rival earns most non-branded traffic from hundreds of near-identical integration pages, each targeting a phrase like 'connect X with Y'. The insight is the template, and the keyword list is the evidence.

Step 3: Pull their paid keywords to see what converts

Paid keywords show you which queries a competitor will spend money on, which is the closest public signal of what converts for them. Organic data shows reach. Paid data shows revenue intent.

Open the paid or advertising keywords report for each competitor domain
Note which terms they bid on persistently across months, since short tests mean less
Record the landing page for each ad: a demo page, a comparison page or a free tool
Flag any paid term where the competitor has no organic ranking, since that is a page they have not built yet

The last item is the richest. When a rival pays for a bottom-funnel term and has no organic page for it, the query is valuable to them and the organic slot is open to you. Those terms go straight to the top of your shortlist.

Step 4: Run the gap filter against your own domain

Now merge the competitor exports and compare them with your own rankings. Most keyword tools have a gap or intersect report that does this in one view: enter your domain and up to several competitors, then filter for keywords where at least two competitors rank in the top 20 and you do not rank at all.

Pull your Search Console queries too. The tool's view of your rankings is an estimate, while Search Console shows the queries you already get impressions for. A keyword that looks missing in the tool may sit at position 30 in your own data, which makes it a refresh job and not a new article.

Sort the merged list into the buckets below. The table shows the conclusion: missing keywords that several competitors share are your strongest new-content opportunities, while weak keywords are often cheaper wins through updates.

Keyword gap buckets and what to do with each
Bucket	Definition	Action
Missing	Competitors rank, you have no page	Brief a new article
Weak	You rank lower than every competitor	Rewrite and deepen the existing page
Untapped	Only one competitor ranks	Validate intent before committing
Shared	You and competitors all rank well	Defend with updates, low priority
Paid-only	Competitor bids, nobody ranks organically	Fast-track a new page

Keep difficulty in view but do not let it decide alone. The difference between a keyword you can realistically win and one that absorbs months of effort depends on who ranks and why, which is the subject of What Separates Winnable Competitive Keywords from traps. Use that lens before you discard a high-difficulty term that only weak pages rank for.

Step 5: Check which brands AI assistants name for those queries

Keyword tools only see Google. A growing share of B2B buyers in 2026 ask ChatGPT, Claude, Perplexity or Gemini first, and those assistants cite or recommend a handful of brands per answer. A competitor can lose a keyword in Google and still own the AI answer for it.

Take your top 20 gap keywords and rewrite each as the question a buyer would ask an assistant
Run each question in ChatGPT, Claude, Perplexity and Gemini, ideally in a fresh session
Record which brands are named and which pages are cited, where the assistant shows sources
Mark keywords where a competitor appears in AI answers but not in Google, and the reverse

The pages that get cited usually share a shape: a direct answer near the top, headings that read as standalone questions, specific definitions and clear comparisons. When you see a rival cited, open the cited page and note its structure. That structure is part of what you are competing against, alongside the keyword.

Assistant answers vary between sessions and over time. Treat a single run as a sample, repeat important prompts on different days, and look for brands that appear consistently before you conclude a competitor owns an AI answer.

Step 6: Use AI to expand and cluster the raw list

AI is useful for finding keywords once you give it real data to work from. Paste a batch of gap keywords into ChatGPT or Claude and ask it to group them by search intent, suggest the question variants a buyer would use, and identify which terms belong on the same page.

Avoid asking a model to invent keywords from nothing. It has no search volume data and will produce plausible phrases nobody types. The reliable pattern is data first, AI second: the keyword tool supplies demand, the model supplies grouping and phrasing, and you verify any new suggestion against the tool before it enters the plan.

keyword,intent,competitors_ranking,our_position,cluster
saas onboarding checklist,informational,3,none,onboarding
user onboarding software,commercial,4,28,onboarding
onboarding email examples,informational,2,none,onboarding
reduce churn first 30 days,informational,2,none,retention

The sample above shows the shape your merged sheet should take. The first line holds the column names. Each following line is one keyword with its intent, the number of competitors ranking in the top 20, your current position from Search Console, and the cluster an AI pass assigned. The second row is a weak keyword to refresh. The other three are missing keywords to brief.

Step 7: Score, cluster and schedule the winners

Give each cluster a simple score based on business relevance, the number of competitors ranking, AI answer presence and the effort to win. Relevance comes first. A high-traffic glossary term your buyers never search before purchase is worth less than a modest query that leads to a demo request.

Assign one primary keyword per page and fold close variants into the same article
Publish the pillar topic of each cluster before its supporting articles, so internal links have somewhere to point
Schedule weak-bucket refreshes alongside new pages, since they often move faster
Re-run the gap report each quarter, because competitors publish too

If you want the reasoning behind how these signals compound over time, What Keyword Research Actually Does covers that mechanism. For the full workflow beyond competitor data, the same scoring logic slots into broader keyword research and analysis. Competitor domains are also a strong starting point when you need to find backlinks, since the pages ranking for your gap keywords show which sites link in your niche.

This is the step where most small teams stall. The analysis takes an afternoon. Writing and publishing a cluster by hand takes weeks. That gap is why we built Gofylo to research competitor keywords from the live results, group them into topic clusters and publish 30 articles a month to your CMS, each written in a few minutes with internal links, schema and FAQ blocks in place.

Troubleshooting

Most problems with competitor keyword research come from noisy inputs or misread data. These are the ones I see most often and the fix for each.

The export has tens of thousands of irrelevant keywords. You probably included a large publisher or a multi-product company as a competitor. Filter that domain's export to the subfolder that matches your category, such as its blog section on your topic, or drop it and replace it with a narrower rival.

The gap report says you rank for nothing. New or small domains often fall below a tool's tracking threshold. Use Search Console as your source of truth for your own positions and use the tool only for competitor data.

Traffic estimates look wildly wrong. Third-party traffic numbers are modeled, not measured. Use them to compare keywords relative to each other and never as a forecast for your own results.

Competitors rank with thin pages you could easily beat, yet you still do not rank. Check whether you have a page that matches the intent. A feature page will not outrank a how-to guide for an informational query, however strong your domain. Match the page format the ranking results use.

AI assistants never mention your brand even where you rank in Google. Restructure your ranking pages with answer-first openings, question headings and specific definitions, and get your brand named on third-party pages assistants cite. Then measure again with a visibility check before and after.

Frequently Asked Questions

How do I find the best keywords?

The best keywords sit where buyer relevance, real demand and a realistic chance of winning overlap. Start from competitor gaps several rivals share, confirm the intent matches a page you can build, check your Search Console data for queries already close to page one, and prioritize terms a competitor pays for but does not rank for organically.

What is a competitive analysis?

A competitive analysis is a structured review of how your rivals position, price, sell and acquire customers. In SEO and GEO it narrows to which queries competitors win in Google, which pages earn that traffic, and which brands AI assistants name for the questions your buyers ask.

Want to see where competitors already beat you in AI answers before you build the plan? Run the free AI Search Grader, then start a 3-day Gofylo trial to turn your competitor gaps into published, AI-ready articles at $79/month, cancel any time.