Keyword difficulty
KD
A third-party estimate of how hard it is to rank for a query.
Scored out of a hundred by SEO tools, based mostly on the backlink profiles of the pages currently ranking. Each vendor computes it differently, so a score of 45 in one tool and 45 in another are not the same claim.
Key takeaways
- The model has never seen the client site. It describes the competition, not the distance between you and it.
- Mostly link-based, so it underrates queries that are hard for other reasons: brand preference, marketplaces, intent no commercial page can serve.
- Not comparable across tools. A 45 in one vendor and a 45 in another are different claims.
- Genuinely useful for sorting a list of hundreds. Close to useless as a forecast for one term.
- Never put it next to a timeline in a proposal. It reads as a commitment.
What the number is actually built from
Every vendor keeps its formula private, but the shape is common knowledge and consistent across them: look at the pages currently ranking on page one, assess the strength of the links pointing at those pages and their domains, and compress that into a score out of a hundred.
Some vendors add search volume, some add the number of referring domains rather than their strength, some fold in how often the results change. The differences are enough that the scores are not interchangeable, which is why a client comparing two agency reports on the same keyword can see two different numbers and reasonably ask which is right. Neither is.
The important structural point is what is absent. The model knows the competition and knows nothing about the site being proposed. It cannot tell you whether this client can reach that page, only that those pages are where they are.
The half of the question it cannot answer
Difficulty is not a property of a keyword. It is a relationship between a keyword and a specific site, and the second half is the half that decides whether the work is worth commissioning.
A score of 60 is a different proposition for an established domain in the category than for a site launched last quarter. For the first it may be a month of work; for the second it may be unreachable this year. The number is identical in both proposals.
The way to close that gap is not a better tool, it is looking at the results. Open the first page and ask whether these are sites like this client or sites unlike this client. Are they national publishers or ordinary businesses. Do any of them have less standing than the client already has. That takes two minutes and answers the question the score was standing in for.
This is also the honest thing to put in a proposal. "Three of the ten results are businesses comparable to yours" is a defensible statement about a specific client. "Difficulty 42" is a vendor model output presented as analysis.
The queries it systematically gets wrong
Because the models are mostly link-based, they underrate every kind of difficulty that does not come from links, and those are common.
Entrenched brand preference is the clearest case. A query where people are looking for one specific company, or a category where one brand is the default answer, can have an unremarkable difficulty score and be effectively unwinnable, because the thing standing in the way is not link equity.
Results dominated by marketplaces behave the same way. A commercial query returning Amazon, a comparison site and two directories is not a page an independent business displaces by publishing something better, whatever the score says.
And intent, which is the one that costs most. A query whose first page is entirely guides is not difficult for a product page, it is impossible for one, and no difficulty score encodes that. Checking intent before checking difficulty is the right order, because intent can rule a term out in a way difficulty never does.
They also overrate in one direction: a term can carry a high score because the ranking pages sit on strong domains while none of them actually answers the query well. Those are among the best opportunities available, and the score points away from them.
Where it earns its place
None of this makes the metric useless. Used as a sorting device it is genuinely good: given a list of eight hundred candidate terms, it orders them well enough to decide which fifty deserve a human looking at the results page.
That is the job it should keep. Triage at scale, followed by judgement on the shortlist, is a workflow that uses the number for what it can do and does not ask it for what it cannot.
What it should not do is appear in client-facing documents. A score next to a keyword in a proposal invites two conversations you cannot win: why the number differs from the one their other agency quoted, and why the term has not moved when the difficulty was only 30. Both are arguments about a vendor model rather than about the work.
If a proposal needs to convey how hard something is, describe the results page. It is more useful, it is specific to the client, and nobody can produce a different number for it.
What it means for an agency
Never put a difficulty score in a client proposal next to a timeline. It reads as a commitment, and it came from a vendor model that has never seen the client domain.
Where it goes wrong
- Treating difficulty as a property of the keyword rather than a relationship between the keyword and one site.
- Comparing scores across tools, which are built on different indexes and different formulas.
- Checking difficulty before checking intent, when intent is the thing that can rule a term out entirely.
- Skipping the high-scoring terms where none of the ranking pages actually answers the query.
In short
Keyword difficulty is a sorting tool that gets used as a forecast. It orders eight hundred terms well and predicts one term badly, and the two minutes spent looking at the actual results page will tell you more than the score ever does.
Frequently asked questions
Is keyword difficulty accurate?
It is a reasonable description of the competition and says nothing about your site, which is the half of the question that decides whether the work is worth doing. It is also mostly link-based, so it underrates queries made hard by brand preference, marketplace dominance or an intent no commercial page can serve. Useful for sorting a long list; unreliable as a forecast for one term.
Why do different tools give different difficulty scores?
Because each vendor uses its own link index and its own formula, and none of them publishes it. Some weigh referring domain counts, some weigh their own authority score, some fold in volatility or search volume. The consequence is that scores are not interchangeable: a 45 in one tool and a 45 in another are two different claims, and comparing them tells you nothing.
What difficulty score should I target?
There is no threshold that means anything, because the score describes the competition rather than the gap between you and it. A 60 can be a month of work for an established site and out of reach for a new one. Open the first page instead and ask how many of the results are sites comparable to the client. That answer is specific and the score is not.
Should I show keyword difficulty to clients?
Better not to. A number next to a keyword in a proposal reads as a commitment, and it sets up two arguments you cannot win: why it differs from the figure another agency quoted, and why the term has not moved when the difficulty was only 30. Describing the results page does the same job, is specific to that client, and nobody can produce a rival number for it.
Related terms
Last reviewed
Keplent keeps track of these for you.
It holds what each client agreed to, watches what their campaigns actually did, and tells you in plain sentences what needs you this week.
Join the waitlist