Judging whether you can realistically win a keyword

Most keyword decisions are made by looking at two numbers: volume and a difficulty score. Both come from a company selling a subscription, and neither knows anything about your site.

Difficulty scores are not fraudulent — they are models, usually built on sampled link data, and they are reasonable for sorting a thousand keywords into rough tiers. They are unreliable for the decision that actually matters: *can this particular site win this particular query?* That question depends on your starting position, which the score has never seen, and on how good the incumbent pages are, which the score does not read.

What follows takes about twenty minutes per keyword and answers the real question.

The three things that decide it

How strong are the pages currently there. Not the sites — the pages.

How strong are you, relative to them. A score of "40" means nothing without this half.

Whether the intent lets you compete at all. If the query wants product pages and you have an article, strength is irrelevant.

Difficulty scores address the first, partially. The other two are yours to assess, and they are usually decisive.

Step 1: is the intent even available to you?

Do this first because it is the cheapest disqualifier.

Search the query. If the results are entirely a page type you do not have and will not build — shop category pages, marketplaces, directories, video — the question is settled regardless of authority. You are not competing with those pages; you are absent from that competition.

Half an hour spent on the wrong page type is the most common wasted effort in this whole subject. The method is in how to read a search results page before you write anything.

Step 2: read the weakest page in the top five

The relevant comparison is not the best result. It is the weakest of the ones you would need to pass.

Open positions three, four and five. Actually read them. For each:

• Does it answer the query completely, or partially?
• Is it specific, or generic?
• Is it current?
• Does it show any evidence of first-hand knowledge?
• Is it on a site that is genuinely about this subject?

Then the question that decides it: can you produce something clearly better than this? Not marginally. Clearly, and in a way a reader would notice within fifteen seconds.

If three of the five are thin and out of date — which happens far more often than difficulty scores suggest — a small site can take that page. If all five are excellent, specific and current, no amount of effort makes this a good target this year.

Step 3: assess your own position honestly

Four questions. There is no scoring system; the pattern is what matters.

Have you ranked for anything adjacent? If you already appear for related queries, you have some standing in the subject. If you appear for nothing, you are starting from zero and should say so.

How old is the site, and has it been consistently published? A site with three years of steady work is in a different position from one launched in March.

Does the subject sit inside what you genuinely know? This is not sentiment. First-hand knowledge produces specifics — real numbers, real cases, real constraints — that generic competitors cannot manufacture, and that is often the only advantage available to a small site.

Do you have anything the incumbents structurally lack? Your own data, your own cases, access to practitioners, a genuine speciality.

Step 4: put the two halves together

| Incumbents | You have standing | You are starting from zero | |---|---|---| | Weak, generic, dated | Target it. Best available opportunity | Target it. Slower, still winnable | | Mixed — two weak, three strong | Target it, expect months | Target a narrower version of it | | All strong and specific | Possible with real depth | Avoid. Choose a longer-tail query | | All marketplaces or directories | Not available as a page | Not available. Consider being on them |

The bottom-right cell is the one people ignore. If the query is structurally held by platforms, the realistic options are a different query or appearing on those platforms — the trade-off is in when renting attention beats owning it.

The narrowing move

When a query is out of reach, the answer is almost never "try harder". It is to find the version of the question you *can* own.

Narrow along one of these axes, and be specific rather than merely longer:

| Axis | "Project management software" becomes | |---|---| | Audience | …for architecture practices | | Constraint | …for teams that cannot use cloud storage | | Situation | …when you are migrating off spreadsheets | | Scale | …for a team of three | | Geography | …with UK data residency |

Each is a smaller query with far weaker competition and, crucially, a searcher who is *more* likely to buy, not less. A page that wins a narrow query outperforms a page that ranks fourteenth for a broad one, and it is achievable this year rather than in three.

The narrow versions also accumulate. Ten of them covering one subject is how a site eventually becomes credible enough to attempt the broad term. That sequence is the whole argument in topic clusters and pillar pages.

The method run end to end

Worth seeing once as a whole, on an illustrative case: a four-person firm that installs and maintains commercial coffee machines, wanting "commercial coffee machines".

Step 1 — intent. The results are almost entirely product listings from manufacturers and large suppliers, plus two marketplaces. The searcher wants to buy a machine. Our firm does not sell machines; it services them. The intent is not available to us at all, and everything after this point would have been wasted effort.

Reframe. What does someone need who would eventually become a customer? They have a machine that is broken, or they are deciding whether to lease or buy, or they have been quoted for a service contract and do not know if it is reasonable. Those are three different queries and none of them is the one we started with.

Step 2 — take "commercial coffee machine service contract cost". The results are two supplier pages that avoid giving figures, a generic blog post that mentions no numbers at all, a forum thread from four years ago, and a manufacturer's page about their own plan. The weakest of the five answers the question with "it depends on your requirements".

Step 3 — our position. Site is three years old, ranks for the firm's name and a couple of local terms, has never published anything substantial. Starting close to zero on authority. But the firm quotes this exact figure every week, knows what drives it up and down, and knows what a bad contract looks like. That is first-hand knowledge the incumbents visibly lack.

Step 4 — the verdict. Mixed-to-weak incumbents, near-zero standing, genuine unique knowledge. Row two, right-hand column: target a narrower version. So not the national question but "what a commercial coffee machine service contract should cost, and what changes the price" — with real ranges, the four factors that move them, and the two clauses worth refusing.

That page has a plausible route to the top five within a year, because it is the only one that answers the question. The original query never did.

Note what did most of the work: not a difficulty score, but reading five pages and noticing that all of them dodged the question.

When to conclude you got it wrong

Choosing a keyword is a prediction, and some predictions are wrong. The failure mode is not making a bad call — it is continuing to invest in one for two years because abandoning feels like waste.

Reasonable review points for a page targeting a query you judged winnable:

Three months. Are there impressions for the query at all? If Google is not showing the page for it even occasionally, it has not been understood as an answer to it. Usually an intent or format mismatch, not a patience problem.

Six months. Is the average position improving, even from position 40 to 25? Direction matters more than absolute position this early. Flat at nothing after six months, on a query you judged weak, means the judgement was wrong.

Twelve months. Is it in the top twenty? If not, and the page is genuinely good, the honest conclusion is that the incumbents were stronger than they looked — usually because their authority, not their content, is what holds the position.

At that point the options are narrowing further, folding the page into a stronger one, or accepting that the query is not available and buying the position instead. What is not an option is another year of the same.

What difficulty scores are actually good for

Not useless — just misused.

Good for: sorting a long list into rough tiers before you read anything; noticing that one keyword is wildly harder than its neighbours; comparing within a single tool consistently.

Bad for: deciding a single keyword; comparing across tools, which use different scales; anything involving your own site, which they cannot see.

Use them to shortlist twenty candidates. Use the results page to choose three.

An honest note on timelines

Even a well-chosen, winnable keyword takes months. Choosing correctly changes whether it works at all; it does not make it fast. If the business needs results sooner than that, keyword selection is not the lever — see how long SEO takes for the actual sequence, and paid versus organic for the routes that are quicker.

If you would rather see what a position costs directly rather than estimating what it costs to earn, the rankings board publishes it.

Common questions

Is there a difficulty number below which I am safe? No. The same number means different things in different subjects, and none of them account for your site.

Should I only target easy keywords? Early on, mostly yes. A page ranking for a narrow query beats a page ranking nowhere for a broad one.

How do I know if I am improving? Watch whether the queries you appear for are getting broader over time. That is the signal that standing is accumulating.

What if my whole market is dominated? Then organic is a long game or the wrong game, and it is better to know that in month one than month ten.