Keyword research goes wrong at the point of commitment, not the point of discovery. Finding keywords is easy and tools do it for you. Deciding which ones you can actually win, and killing the ones you cannot, is the part that determines whether the next six months of content produces traffic.
This is a verification checklist, not a tutorial. If you need the process itself — seeds, expansion, tools — start with our keyword research guide and come back here before you write anything.
Stage 1 — Is this a real keyword?
| # | Check | Kill it if |
|---|---|---|
| 1 | Volume is above your floor | Under ~50/mo with no strategic reason |
| 2 | Volume is annual, not a spike | All searches fall in one month and it is not seasonal for you |
| 3 | The term is not a brand | It is a competitor’s product name you cannot outrank |
| 4 | It is not a misspelling of a bigger brand | Google is serving the brand it thinks you meant |
Check 4 matters more than it sounds. This site earns 1,703 impressions a month for variants of “seo journal” — searches from people looking for a different, much larger publication. Those impressions convert at zero and always will. If a keyword’s SERP is dominated by a brand whose name resembles yours, the traffic is not addressable regardless of your position.
Stage 2 — Can this site rank for it?
The stage most checklists skip, and the one that actually decides the outcome.
| # | Check | Kill it if |
|---|---|---|
| 5 | KD is inside your ceiling | KD is far above what your domain has ever ranked for |
| 6 | At least one weak domain sits in the top 10 | Every result is DR 70+ |
| 7 | The parent topic is the keyword itself | The parent is a much broader head term |
| 8 | The top 10 contains pages, not just tools | Software dominates and you are publishing an article |
| 9 | Results are not all from one platform | Reddit, YouTube or Amazon own the SERP |
Check 6 is the single most useful test available. In Ahrefs it is the serp_domain_rating_top10_min filter; by hand it is opening the SERP and looking at who ranks. If a DR 20 site currently sits in the top 10, a DR 20 site can rank there. If the weakest result is DR 78, difficulty scores are irrelevant — you are not getting in.
Check 7 catches the most expensive mistake. Ahrefs reports a parent topic for every keyword: the term driving most traffic to the current #1 page. If you target “keyword research for beginners” (KD 72) its parent topic is “ahrefs keyword research” — meaning Google answers that query with tool pages from DR 90 domains. You are not competing with an article, you are competing with a product. Match the parent topic or accept you are targeting something you cannot have.
The keyword difficulty guide covers how to read these numbers against your own domain strength; domain authority covers where your ceiling comes from.
Stage 3 — Does the intent match what you plan to write?
| # | Check | Kill it if |
|---|---|---|
| 10 | Format matches the SERP | You planned an essay; the SERP returns templates and tools |
| 11 | Commercial intent matches the page type | You planned a landing page; the SERP returns tutorials |
| 12 | You can serve the intent honestly | Winning requires data, tooling or proof you do not have |
| 13 | SERP features do not eat the clicks | An AI overview or knowledge panel answers it completely |
Check 10 is where good writing loses to mediocre assets. If the top 10 for “seo audit template” is nine downloadable templates and one blog post, publishing a tenth blog post is not competing. Match the format or pick a different keyword.
Check 11 is the cannibalisation trap in disguise. A commercial landing page targeting a query that returns how-to guides will lose to your own blog post, and you will have built two pages that fight each other.
Stage 4 — Does it fit what you already have?
| # | Check | Kill it if |
|---|---|---|
| 14 | No existing page targets this parent topic | You already rank for it with another URL |
| 15 | It is not a near-duplicate of a planned page | Two planned pages share a parent topic |
| 16 | Search Console shows no existing impressions | You are already earning for it — improve that page instead |
| 17 | There is a commercial page to link to | Nothing on the site monetises this traffic |
| 18 | You can name the next step for the reader | The article ends nowhere |
Check 14 and 15 are the same test applied to your live site and your plan. Enforce one page per parent topic in both directions and internal cannibalisation becomes structurally impossible rather than a judgement call you revisit every quarter.
Check 16 is free traffic most sites walk past. Before writing anything new, filter Search Console for queries where you rank 11–30. Those are pages Google already considers relevant; improving one is cheaper and faster than earning a new position from nothing.
Check 17 is the one that turns rankings into revenue. An article with no path to a page that converts is a cost centre. Decide the destination before you write, not after — our internal linking guide covers placement, and if the article has no plausible destination, that is a reason to question the keyword.
Running the checklist in practice
Eighteen checks sounds heavy. In practice stages 1 and 2 kill most candidates in about two minutes each, and only survivors reach stages 3 and 4.
A realistic funnel from a research session: 200 keywords discovered, 60 clear the volume and brand checks, 15 clear the domain-strength checks, 9 survive intent matching, 7 get written. The 193 you killed are the value. Every one of them was a month of writing that would have produced nothing.
After the checklist
A verified keyword list is not a content plan. Clustering, mapping keywords to URLs and deciding what to write first is separate work — what to do after keyword research picks up from here.
If your list is for a specific context, the constraints change: B2B keyword research deals with volumes too low for these thresholds, international keyword research with the same term behaving differently per market, and SaaS keyword research with the alternatives-and-integrations patterns that only exist in software.
If you would rather have the research done and pressure-tested against your actual domain strength, that is part of what our SEO services cover — and for sites where the constraint is a technical ceiling rather than a keyword one, a technical SEO audit tells you whether the pages you already have can rank at all.
