How to Find Low-Competition Keywords for a New Affiliate Site
A new affiliate site can't win "best running shoes". It can win "best running shoes for flat feet wide toe box", and the whole first year of a site is finding fifty keywords like that and writing them before anyone else bothers. The skill is reading a results page and knowing, in thirty seconds, whether a page from a site with no authority can land on it.
This post gives you a scoring method for that judgement, the modifiers that reliably lower competition, what to do when the results are mixed, and how to turn the list into a first plan. It goes a level deeper than the difficulty section in the keyword research post, which covers where the keyword ideas come from.
The short answer
Search the keyword. Score the top ten on five signals: how many results are small sites, how many are forums or Q&A, how many titles don't contain the phrase, how many pages are thin or old, and whether the intent matches your page.
Eight or more points out of fifteen is a target. Add modifiers (for a user, a constraint, a situation) to turn hard keywords into easy ones, and stop when you have thirty.
What "competition" actually is
Google ranks pages, not keywords. A keyword is hard when the pages already ranking for it are strong: authoritative sites, exact-match titles, long current content, many links pointing at them. It's easy when they're weak.
A paid tool's difficulty score is a guess at the strength of those pages from their backlinks. Looking at the pages directly is the same judgement with more information.
So competition is something you read off the results page, and the question is always the same: could a well-made page from a site with no history displace one of these ten?
The five-signal score
Search the keyword in a private window, look at the ten organic results, and score each signal from 0 to 3.
| Signal | 0 | 3 |
|---|---|---|
| Small sites | All ten are brands, retailers or national publications | Four or more are sites you've never heard of |
| Forums and Q&A | None | Two or more Reddit, Quora or forum threads on page one |
| Title mismatch | Every title contains the phrase or a close variant | Half or more don't; Google is ranking approximate answers |
| Thin or old pages | All long, current, structured guides | Three or more are short, undated, or years old |
| Intent match | Results are a different page type from yours (how-to when you'd write a guide) | Results are exactly the page type you'd write |
Add them up. Twelve to fifteen: write it this week. Eight to eleven: a good target for a site with a few months of history. Below eight: not yet. The point isn't precision; it's making the same judgement the same way for every keyword so the list is comparable.
Modifiers that lower competition
The head term is the hardest version of every keyword. Each modifier you add narrows the search and, usually, removes the big sites, which don't bother with pages that specific. The ones that work most reliably for affiliate keywords:
- For a user: for beginners, for seniors, for kids, for small hands, for tall riders.
- For a situation: for apartments, for travel, for winter, for a small car, for two children.
- With a constraint: under $100, under 5 lbs, without a hitch, no subscription, quiet.
- With a feature: with a gauge, with a built-in light, with a long battery.
- A pairing: for [product] and [other product] together, for [brand] bikes.
Two modifiers together ("best bike rack for small cars without a hitch") are usually where a new site's first rankings come from. The search volume is small; the buying intent is total, because someone that specific is ready to buy.
Checking the keyword is real
Low competition sometimes means nobody searches it. Before scoring, check the phrase exists: type it into Google and see whether autocomplete offers it or a close version, and look at the "related searches" at the bottom of the results.
If neither shows anything like it, pick a different modifier. A keyword that passes both is one real people type, whatever a volume tool would estimate.
When the results page is mixed
Most pages aren't clearly easy or hard. Three big sites, two forums, five mid-sized blogs. The way to read a mixed page is to look at positions four to ten, not one to three. The top three may be unassailable for now; if positions four to ten contain a forum and two thin blogs, there's a spot a new site can take, and a page that holds position five on a buying keyword earns.
The other mixed case is intent: half guides, half single reviews. That means Google hasn't settled what the query wants (the intent post covers reading a results page), and a page that does both, a guide with a clear top pick and a proper section on it, often wins.
From a list to a plan
Score thirty to fifty keywords in one topic, sort by score, and take the top thirty. Then shape them into a cluster: the hub term (probably hard, write it anyway as the hub page), ten to twelve buyer-guide keywords, reviews of the products those guides will pick, and the question keywords around them.
Publish the highest-scoring guides first, because they're the ones that can rank while the site is new.
Re-score anything that doesn't rank after six months. Results pages change, and a keyword that scored six last year may score ten now.
Doing it with AffSERP
Three AffSERP tools cover this. The Keyword Explorer runs a seed term through autocomplete with the modifier expansions above, so the long-tail list builds itself. The Competitor Sitemap Explorer shows which of those keywords a ranking competitor has already covered, which is a fast proxy for competition: what they skipped is often what's open.
Keyword Grouping clusters the survivors by intent into the hub-guide-review shape, and each keyword has a Write button that sends it to the article tool. The scoring itself is still a human judgement, and it takes about thirty seconds per keyword once you've done ten.
FAQ
What counts as a low-competition keyword?
One where the current top ten has weak pages: small sites, forum threads, titles that don't match the query, thin or old articles. It isn't about search volume. A keyword with 200 searches a month and a beatable results page is worth more to a new site than one with 20,000 and a page one of national brands.
Do I need a paid tool for keyword difficulty?
No. Paid difficulty scores are estimates built from the backlinks of the ranking pages. Reading the results page directly tells you the same thing more precisely, and it shows you things a score can't, like whether the results match the intent of your planned page.
How many low-competition keywords does a new site need?
Enough to fill the first cluster: roughly ten to twelve buyer-guide keywords in one topic, plus the reviews and questions around them. Thirty keywords that all pass the check is a good first plan. More than that and you're researching instead of publishing.
Can a low-competition keyword still be worth nothing?
Yes, if nobody searches it or if it has no buying intent. Check that it appears in autocomplete or related searches, which means real people type it, and check that the results page shows product pages or guides rather than only how-to articles. Easy and worthless is a common combination.
Keep reading
