Competitor Keyword Analysis: A Step-by-Step Method for Building a Work Queue You'll Actually Ship
Turn a competitor's keyword footprint into a short, defensible queue of winnable terms - and learn which gaps to deliberately ignore.

Inheriting a competitor list from a previous agency is one of the more uncomfortable moments in search work: forty domains, six thousand rows, no explanation. Your job is not to defend the list but to re-establish, quickly and defensibly, which competitor terms your client can actually win before you pitch another rand of content budget.
Competitor keyword analysis ends with a work queue, not a data dump. If nobody can name the first ten pages to write and who owns each, the analysis has not finished.
The method below is deliberately narrow: seven steps, a short mistake list, and a pre-flight check you can hand to writers. Agencies running this across multiple clients can see how the workflow maps to Straider for agencies.
Competitor Keyword Analysis Is About Credit, Not Equivalence
A competitor ranking for a term proves that term has interest - not that it has value to you. Your competitor may rank third on a 50,000-searches-per-month head term and collect very few clicks, because an AI Overview answers the query in place, a featured snippet absorbs it, or the SERP is structurally zero-click. Copy the term into a content plan and you inherit the illusion of traffic without the clicks.
Treat your competitor export as a hypothesis list. Every row must earn its place against three questions: does this SERP actually send clicks, does my own site already have impression evidence for it, and does an AI answer cite someone already (in which case the competition is on a different axis).
What You Need Before You Run a Single Export
- Search Console access with at least 16 months of query-level history (see Google's Performance report documentation).
- One paid keyword database - not three open in different tabs.
- The client's commercial context: average order value or lead value, target margin, and which products or services actually pay the bills.
- A geolocation decision. A .co.za ecommerce brand must pull a ZA-geolocated database; default US volumes routinely overstate or invent terms that South African searchers phrase differently.
Step 1 - Screen Your Competitor List Down to Three to Five
Forty competitors dilute everything. Cap the list at three to five and write one sentence of justification per domain: overlaps the same query landscape, ranks in the same commercial cluster, has publishable content worth studying. Drop directories, marketplaces and publishers - they do not compete for the same pages you will build.
Step 2 - Extract Once, Reconcile to One Documented Source
Ahrefs, Semrush and Moz build their databases differently - crawl scope, index freshness, panel and clickstream inputs all vary - so the same competitor returns materially different keyword counts and volumes between tools. Our work with clients since 2017 shows the same domain can swing by several thousand keywords between providers depending on match type and location. Pick one tool as primary, record the pull date and filters, and stop exporting diffs.
| Source | What it is good for | Known limitation |
|---|---|---|
| One paid SEO database (primary) | Competitor keyword footprint, rough volume ranges | Modelled estimates, rounded; unreliable below ~100/mo |
| Your own Google Search Console | Query-level impressions for positions 8-20 | No competitor data; thresholded samples on larger sites |
| SERP feature check (manual) | Click-iness, AI Overview presence, snippet ownership | Point-in-time; record the date |
Label every volume you pass to a client as a modelled estimate. Presenting a rounded 2,400 as fact is how reporting relationships break down in month three.
Step 3 - Subtract Branded Volume and Terms You Already Own
Strip any query containing a competitor's brand name, and any query where your client's own brand dominates. A competitor's brand-name searches often account for the largest single block of its keyword count while being entirely unavailable to you. Then cross-reference your own Search Console: if your site already holds 40,000 impressions for a term at position 14, that is a gap you can close with an existing URL - not a new page. This overlap check is the single biggest quality filter in the whole process.
Step 4 - Read a Competitor's Sitemap Before You Read Its Keyword Export
A dated XML sitemap index tells you what a competitor bet on and when. A burst of [brand] vs [brand] slugs published inside three weeks is a clearer statement of intent than any volume export. Pair sitemap diffs with slug patterns - comparison, use-case, location, price - to infer the page archetypes you will need. Google's large sitemap guidance explains how sitemap indexes and segmented files are structured, which helps when reasoning about publish cadence.

