SEO Competitor Keywords: How to Reverse-Engineer Any Rival's Footprint Straight From the SERPs (No Subscription)

Search Google for “seo competitor keywords” and look at who ranks. Every single page-one result is a paid tool or a guide that funnels you toward one — SpyFu at the top, then SE Ranking, SEO Sherpa, Mangools, Semrush, and Moz. The message is consistent: to see the keywords your competitors rank for, open your wallet. Even every step of Semrush’s own guide runs through a paid tool — Domain Overview, Organic Rankings, Keyword Gap, and Position Tracking, one subscription feature per step.

Here is the thing those pages would rather you not dwell on. The data they sell is already public. A competitor’s keyword footprint is scattered across the search results themselves: the pages they rank on, the People Also Ask boxes, the related searches, the autocomplete suggestions, and the AI Overview citations. You can reconstruct it with nothing but SERP scraping and a spreadsheet. No seat license required.

This guide is the method. We reverse-engineer a real domain end to end, publish the full recovered-keyword table, and then spot-check it against live SERPs to show how precise a subscription-free approach actually is. Fair warning on the bias: cloro sells a SERP API, so when we get to automating this at scale we are describing our own product. The manual method below works with zero budget, and every number here is reproducible — that is the point.

What SEO competitor keywords actually are (and which competitors to reverse-engineer)

SEO competitor keywords are the search terms a rival ranks for that you want to rank for too. They are the raw material of a content plan: every term a competitor already earns traffic on is a term with proven demand and a proven path to the top of the page.

The trap is analysing the wrong competitors. SEO competitors are not always your business competitors. An SEO competitor is any site that shows up in search results for the terms you care about. A business competitor is a company that sells what you sell. A running-shoe brand competes commercially with other shoe brands, but in the SERPs it competes with Runner’s World and a dozen review blogs. Reverse-engineer the sites that actually occupy the results, not just the ones on your sales team’s whiteboard.

Once you have picked a domain, the footprint you are trying to recover is everything that domain touches on a results page: the organic terms it ranks for, the questions it appears under in People Also Ask, the phrases it shows up in under related searches, and — increasingly — the queries where it gets cited in the AI Overview. If you are fuzzy on what those SERP elements are, our explainer on what a SERP is maps every one of them. Each element is a leak. Put enough leaks together and you have the competitor’s keyword strategy.

The no-subscription method: query fan-out, People Also Ask mining, and autocomplete expansion

The whole method rests on one idea: the SERP is a database, and every query you run returns a few more rows of a competitor’s footprint. You expand outward from a seed until the new queries stop returning new keywords. Three engines drive that expansion.

Query fan-out starts with a seed — the competitor’s brand name and its two or three core topics — and branches into the natural variations searchers actually use: [brand], [brand] alternatives, [brand] vs, [topic] tool, best [topic] for [audience]. Each branch is a live search. The domains and titles that come back tell you which terms the competitor competes on. They also tell you who else is in the fight. This is exactly the “fan-out” that large language models perform when they answer a broad question: they silently expand it into many sub-queries. It is the single highest-leverage step in the whole process.

People Also Ask mining turns the question boxes into a keyword tree. Run a seed query, record every PAA question, then expand the promising ones (clicking a PAA question loads more questions beneath it). Two or three levels deep, you have dozens of the exact questions your competitor’s content is built to answer. They arrive phrased in searcher language, not marketer language.

Autocomplete expansion completes the long tail. Type the competitor’s brand or a core topic into the search box and record the suggestions; then append each letter of the alphabet to surface more. Autocomplete is Google telling you, for free, which completions have real search demand.

Do all three by hand and it costs nothing but time. Automate them and you are making one API call per query. cloro is built for exactly this: cloro’s fan-out engine returns PAA and related searches as JSON on every Google Search call, and its ChatGPT full-response endpoint runs the query fan-out for you. And do not skip the AI Overview — the AI Overview renders on about 40% of commercial queries, so the domains it cites are part of the modern footprint whether or not they hold organic position one.

Keyword research and competitor analysis, done from the SERP

This is where keyword research and competitor analysis stop being two separate jobs. The paid workflow calls the output a “keyword gap” and hides it behind a dashboard; you are going to build the same output from the rows you just scraped.

Take every keyword you recovered and do three passes. First, cluster by intent — group the terms into topics (informational “how to”, commercial “best tool”, navigational “brand + feature”). Second, filter to what you can realistically win — drop the terms owned entirely by domains with far more authority than yours and keep the ones where the page-one results are beatable. Third, map each surviving cluster to a page you will write or improve.

That is the entire “keyword research competitor analysis” loop that subscription tools charge monthly for: recover terms a competitor ranks for, subtract the ones you already cover, prioritise the remainder. The only difference is that your source rows came from the live SERP instead of a vendor’s index — which means they are current, not a snapshot from the last crawl, and they cost nothing to refresh.

Worked example: reverse-engineering a keyword tool’s own keywords

For the worked example we picked a deliberately ironic subject: SpyFu, the paid competitor-keyword tool sitting at position one for our head term. We reverse-engineered its footprint using only the free method above — no SpyFu account, no Ahrefs, no Semrush. Everything below was captured from live Google SERPs on 9 July 2026 and is reproducible.

We seeded the fan-out with the brand (spyfu), the brand-plus-modifier (spyfu alternatives), and two category topics (seo competitor analysis tool, ppc competitor research). Here is the recovered footprint — every row traceable to a real SERP signal we observed:

Recovered keyword	Recovered via	Observed intent
seo competitor keywords	Head-term fan-out — SpyFu ranks #1	Commercial
seo competitor analysis tool	Category fan-out — SpyFu ranks #2	Commercial
ppc competitor research	Category fan-out — SpyFu ranks #2	Commercial
competitor keyword research tools	Brand SERP title (“Competitor Keyword Research Tools for Google Ads PPC & SEO”)	Commercial
google ads competitor keywords	Brand title fan-out (“…for Google Ads PPC”)	Commercial
download competitor keywords	Brand snippet (“Search for any competitor. Download their keywords.”)	Transactional
competitor ppc & seo	Brand snippet (“Learn competitors’ PPC & SEO tricks”)	Commercial
spyfu alternatives	Brand + “alternatives” fan-out — SpyFu’s own page ranks #3	Commercial / navigational

The fan-out also recovered SpyFu’s SEO-competitor set — the domains that repeatedly appear alongside it: Semrush, SE Ranking, Ahrefs, Moz, Mangools, and Similarweb. That single list, pulled from the “spyfu alternatives” results, is a ready-made target list for the next round of reverse-engineering. Nothing in this table required a login; it was assembled from result titles, snippets, and rankings that anyone can read.

Validation: spot-checking the recovered keywords against live SERPs

A recovered keyword is a hypothesis until the live SERP confirms it. Validation is one search per keyword: run it, and check whether the subject domain actually appears on page one. That is the same precision test a paid tool runs internally — you are just doing it in the open.

We spot-checked the three non-brand category keywords from the table against live results:

• seo competitor keywords → spyfu.com at position 1. Hit.
• seo competitor analysis tool → spyfu.com at position 2. Hit.
• ppc competitor research → spyfu.com at position 2. Hit.

Three for three on the category terms, plus the brand terms that trivially resolve to SpyFu’s own pages. That is a small sample, and we are not going to dress it up as a precision percentage across the whole footprint — the honest read is that fan-out around a domain’s brand and core topics recovers terms it genuinely ranks for, and each one is cheap to confirm before you commit content to it. Terms that fail the spot-check are not wasted, either: a miss usually means the competitor is targeting the term without ranking yet, which is its own signal. To turn a one-time spot-check into ongoing monitoring, the same call feeds rank tracking, where you watch a competitor’s positions change over time.

When to reach for an API: keyword research competitor analysis at scale

The manual method has one real limit: it does not scale. Reverse-engineering one competitor across a handful of seeds is an afternoon. Doing keyword research competitor analysis across 50 competitors and 500 keywords is 25,000 searches, and no one is running those by hand.

That is the point where you automate — and where our bias is on the table, because automating it is what cloro sells. The economics are the argument. cloro’s Google Search endpoint is 3 credits per call, from $0.40 per 1,000 credits. A full 25,000-query sweep costs a few dollars of SERP data, returned as JSON you drop straight into a spreadsheet or warehouse. Compare that to the seat-based model of the tools ranking for this keyword. SpyFu gates its API behind a $119/month plan, and most competitors price by the seat and the lookup. With the fan-out and PAA data arriving as structured JSON per call, the three engines from earlier become one loop your code runs on a schedule.

Whether you stay manual or automate, the strategy is identical — this piece sits between the Competitor Analysis use case and keyword research precisely because the two are the same motion. Start free, prove the method on one competitor, and reach for an API only when the spreadsheet gets too big to refresh by hand.