Four segments · one pass mark
How to Choose SEO Keywords: Score Every Word on Four Wheels Before You Write
To choose SEO keywords, score each candidate 1 to 5 on four things — search intent, commercial value, competition, and resource fit — then keep only the words that clear a fixed total. That single rule separates a list you can execute from a wish list you cannot. Most content plans do not fail because the words were wrong. They fail because nobody agreed on a pass mark before the writing started, so every brief became an argument.
Below: the wheel we run in client workshops, the 20-point scorecard, and the words we deliberately throw away.
What Is the Fastest Way to Choose SEO Keywords?
Score each candidate from 1 to 5 on intent fit, commercial value, competition and resource fit, then add the four numbers. On a 20-point scale, 14 or more means build the page now, 10 to 13 means park it for a later quarter, and 9 or below means drop it. The fixed pass mark takes opinion out of the room.
Why a number beats an argument
The Problem With “This Keyword Feels Big”
Keyword lists usually get built in one of two ways. Either somebody pastes a tool export into a spreadsheet and sorts by volume, or the loudest person in the meeting picks eight words they like. Both methods produce the same outcome: a list that mixes queries needing a calculator tool, queries needing a twelve-page guide, and queries that only convert in Germany. Three months later, half the list is untouched.
A scoring wheel forces the same four questions onto every phrase, in the same order, with the same scale. It also creates a written reason for rejection, which matters more than it sounds. When a sales colleague asks why their favourite phrase was dropped, you can point at a row and a number instead of defending a hunch.
The wheel is deliberately blunt. It has four segments because four is the most a small team can hold in their head during a two-hour workshop. Anything more granular gets ignored after the second session.
The four segments
What Each Segment of the Wheel Is Measuring
Each segment scores one risk. A low score on any single segment is a warning even when the total looks healthy, which is why we flag any segment scoring 2 or lower for a manual review before the page is briefed.
1. Intent Match
A 5 means the phrase describes exactly the page you already have or can publish within a week: same stage, same format, same promise. A 1 means the phrase is asking for something you do not make, such as a free calculator, a job listing, or a definition you would only cover in one paragraph. Intent mismatch is the most expensive mistake on the wheel because it wastes the whole production budget.
2. Commercial Value
A 5 means the query sits one step from a purchase decision and the searcher knows what they want to buy. A 1 means the query is curiosity with no buying stage attached, such as a general history question. Value is not the same as volume: a phrase with 200 monthly searches from procurement managers can be worth more than one with 20,000 from students.
3. Competition
Open the results page and look at what actually ranks, not what a difficulty score claims. A 5 means the top ten is thin: old pages, thin pages, or pages that answer a different question. A 1 means the top ten is owned by marketplaces, forums, or government sites that will not move for a new page. Read the results before you trust any metric.
4. Resource Fit
A 5 means you can produce the asset that wins this phrase within ninety days: the photos, the test data, the interview, the comparison table. A 1 means winning requires something you cannot supply this year, such as lab testing or a proprietary dataset. Resource fit is the segment teams skip, and it is the reason half of all planned pages never ship.
Scorecard
The 1–5 Scoring Table
Print this or paste it into the sheet. The middle column is what a top score looks like in practice; the right column is what should drag a phrase down. Score every phrase on all four rows before you allow yourself to look at search volume, because volume quietly rewrites every other judgement.
| Segment | What a 5 looks like | What earns a 1 or 2 |
|---|---|---|
| Intent | The phrase names the exact job your page performs, and the reader arrives ready for that format. | The phrase wants a tool, a template, a video or a local branch you cannot provide. |
| Value | One step from a quote, a sample request or a specification comparison. | Pure curiosity, student research, or a term that only appears in academic papers. |
| Competition | Top ten includes out-of-date pages, thin pages, or pages with the wrong angle. | Top ten is marketplaces, forums, or official bodies with unbeatable authority. |
| Resources | You have the photos, data, samples or engineer time to build it this quarter. | Winning needs testing, certification or a dataset you do not own. |
| Total | 14–20 build now · 10–13 park · ≤9 drop | Any single segment at 2 or below goes to manual review regardless of the total. |
Running the wheel
How to Run the Wheel in a Single Afternoon
The whole exercise should take one working session for a list of sixty phrases. If it stretches past that, the team is debating strategy instead of scoring, and the scores will drift.
- Collect raw phrases from humans, not just tools.Pull forty to sixty phrases from sales calls, support tickets, quote requests and the questions buyers ask before they trust you. Tool exports fill gaps; they should not set the agenda.
- Cut obvious mismatches before scoring.Delete anything that names a product you do not sell or a format you cannot publish. This usually removes a third of the list and saves an hour of pointless scoring.
- Score all four segments in one pass per phrase.Do not score intent for the whole list, then value for the whole list. Phrase by phrase keeps the judgement consistent and stops one segment from dominating.
- Sort by total, then break ties toward intent.When two phrases land on the same total, the one with the higher intent score gets built first. Intent errors are harder to fix than weak copy.
- Write the pass mark into the brief.Every approved phrase gets one page or one cluster, a stated reader, and the exact asset it needs. If the asset is unclear, the phrase was not ready.
Walk-away rules
Keywords You Should Deliberately Drop
Editing is the part nobody enjoys, and it is the part that protects the calendar. These are the word types we remove from client lists even when the volume column looks attractive.
| Word type | Why it fails the wheel |
|---|---|
| Tool-seekers | “Free generator” style queries want software, not a service page. The visit ends in two seconds and the bounce teaches you nothing. |
| Student queries | Definitions and homework questions pull traffic that will never request a quote, and they distort every downstream metric. |
| Forum-owned terms | If the top ten is all discussion threads, a commercial page will not displace them without years of brand building. |
| Brand terms you do not own | Competitor names generate clicks and complaints. They belong in paid search, not in your content plan. |
| Mixed-intent phrases | One phrase hiding two jobs, such as audit meaning a free checklist and a paid engagement, needs two pages and usually gets neither. |
| Seasonal spikes | A phrase that only breathes for three weeks a year rarely repays the production cost unless you already own the page. |
Dropping a phrase is not a verdict on the topic. It is a scheduling decision: the idea stays in the parking file with its score attached, so a later quarter can pick it up when the resources catch up.
Common misreads
Four Ways Teams Score the Wheel Wrong
Scoring volume instead of intent
A high volume number feels like validation, so the intent score quietly gets raised to match. The fix is mechanical: hide the volume column until all four scores are locked.
Treating difficulty scores as competition
Third-party difficulty numbers are estimates built from link counts. Ten minutes reading the actual results page tells you more than any single metric on the market.
Ignoring the resource segment
Teams score intent, value and competition, then forget that nobody can photograph the product or interview the engineer. The page never ships and the plan loses credibility.
Scoring phrases that are really clusters
Five near-identical phrases are one page, not five. Score the cluster once using its strongest member, otherwise the same job gets built three times.
Where we help
What Our Keyword Work Actually Produces
We run the wheel with your sales team in the room, then hand back a scored sheet, a page map and a build order. That work sits inside our Keyword Research Services, which covers discovery, clustering, scoring and the pass-mark workshop in one engagement.
Once the list is settled, the pages themselves need structure: one clear intent per URL, headings that answer the question in order, and internal links that push authority toward the money pages. That is the job of On-Page SEO Services, and it is where a good keyword list either compounds or stalls.
We do not promise rankings, positions or traffic numbers. What we can promise is that every phrase on the sheet has a reason, a score and an owner.
Process
From Raw List to Build Order
- Discovery call and raw exportWe collect your existing queries, support questions and competitor gaps, then merge them into one deduplicated list.
- Scoring workshopNinety minutes with your team, four segments, one pass mark. Disagreements get resolved on the spot with evidence from the results page.
- Page map and cluster planEach cluster gets one URL, one reader, one format and a short brief that states what the page must contain to win.
- Build, measure, re-scoreSixty days after publishing we re-score the phrase with real data and either expand the page, merge it, or retire it.
Bring Us Your List — We Will Score It With You
Send a spreadsheet of the phrases you are considering and we will run the four segments on a call, including the ones you should drop. No cost for the first session.
Evidence
One Small Example
An industrial parts exporter came to us with a list of ninety phrases sorted by volume. The top phrase scored 2 on intent — searchers wanted a free sizing calculator — and 1 on resources. We dropped it and kept a lower-volume phrase that scored 5, 4, 4 and 5: buyers comparing two mounting standards, with the specification sheet already sitting in their shared drive.
That page took eleven days to write because the engineer time existed. It now sits in the first page of results for its comparison query and routes readers to the quote form. The dropped phrase stayed on the parking list, and eighteen months later it became a small calculator tool on the same domain. Nothing was wasted; it was only re-ordered.
FAQ
Questions We Get About Keyword Selection
How many keywords should one page target?
One intent, one page. A page can rank for dozens of phrasings, but it should be written for a single job. If two phrases need different formats or different answers, they need different URLs.
Do I need a paid tool to score the wheel?
No. The four segments can be scored with a browser, a spreadsheet and your own product knowledge. Paid tools speed up collection and tracking, but they do not decide intent or resource fit for you.
What if a phrase scores 13 and I really want it?
Park it and note what would raise the score. Usually one segment is dragging: the page needs an asset you do not have yet, or the results page needs six months of link building first. Re-score when that changes.
Should long-tail phrases always win?
Long-tail phrases are easier to match on intent, which is why they often score well. But a short phrase with clear buying signals and a thin results page can beat a long-tail one. Score the four segments, not the length.
How often should the scores be revisited?
Every quarter for parked phrases, and sixty days after publishing for live ones. Competition moves fastest, resource fit changes when a new batch of photos arrives, and intent shifts when a product line changes.
Can a low-value phrase still be worth building?
Yes, when it supports a cluster. A definition page that scores 2 on value can still feed links and context to a product page. Just be honest that it is support content and keep it out of the revenue forecast.
Stop Guessing Which Words Deserve a Page
One call, four segments, a written build order. If the list is weak we will say so before you spend a month writing.