glossary · keyword difficulty
What keyword difficulty scores measure, and what they miss
A number between 0 and 100 that no search engine publishes, calculated differently by every tool that sells it. What KD is built from, where it holds up, and the checks that beat it.
Keyword difficulty is a third-party score, usually 0–100, estimating how hard it would be to rank on page one for a term, derived mainly from the backlink strength of the pages already ranking.
reviewed 2026-09-02 · by the IT Master editorial team · how we check facts
What keyword difficulty measures in practice
No search engine publishes a difficulty score. Every KD number you have seen comes from a tool's own crawl and its own index, which is why two reputable tools can rate the same term differently and both be internally consistent.
Most KD formulas start from one thing: the backlink profiles of the URLs currently in the top ten. Referring domains, a domain-strength estimate, sometimes the age of the ranking pages. Some vendors blend in SERP features or estimated click share, but link data does the heavy lifting.
So the score answers a narrow question. It tells you how much link equity sits in front of this query today. It does not answer the question people ask of it, which is whether your site, with the content you can realistically produce, could take a position here.
- It is relative to a vendor's index, never to your domain.
- It reads the incumbents, not the query or the intent behind it.
- It moves when the SERP moves, so any score is a snapshot with a shelf life.
A keyword difficulty score describes the sites already ranking, not your site, which is why two tools can rate the same term differently and both be right about their own index.
Why it matters in 2026
Two forces pull the number in opposite directions.
Links still gate the head terms. For a commercial query with ten established retailers in front of it, a high KD is telling you something real: no amount of writing quality moves a young domain into that set quickly. Reading that early saves months of wasted publishing.
But the share of queries where links decide the outcome keeps shrinking. Long-tail and question-shaped searches turn over on relevance and specificity, and those are exactly the terms KD tools score lowest, or decline to score at all for want of data. A low score on a term almost nobody searches is not an opportunity signal. It is an absence of measurement.
The assistant layer changes the arithmetic again. When an AI assistant answers a question it runs several narrow searches, fetches a handful of pages and quotes passages from them, and that fetch turns on passage relevance rather than domain strength. A term you would never take position one for can still supply the sentence an assistant lifts. LLM SEO covers how that retrieval works.
How IT Master treats difficulty
IT Master discovers topics from real search demand: keyword and SERP data from DataForSEO, plus the site's own Search Console. Difficulty estimates arrive with that keyword data, so they inform the shortlist rather than settle it.
Search Console is the stronger signal. A term a site already collects impressions for, sitting on page two with no clicks, is a term Google has already judged that site relevant to, whatever a third-party score says about the link gap. That evidence is first-party and current, which a vendor score is not.
The rest is execution. Articles are grounded in the customer's own first-party data, such as product records, datasheets and support questions, which is what makes them hard to copy.
Claude models write; Gemini and GPT models fact-check, score novelty against competitor pages and hunt AI tells, so the model that writes is never the model that checks. Across 186 Standard-tier runs, 51% of drafts passed every check first time; the rest were revised or rejected. No scoring input predicts a ranking, and nothing here promises one. How it works sets out the stages.
A term you already collect impressions for is better evidence than any third-party difficulty score, because it is your own site being judged on the live SERP.
Common misunderstandings
"Low KD means easy." It means little link competition, measured by one index. A term can score in single digits and still be locked up by forum threads Google prefers over any commercial page, or by a search intent your site cannot serve.
"KD is comparable across tools." It is not. Each vendor has its own crawl, its own authority estimate and its own curve. Compare within one tool over time, and never mix two.
"Difficulty is a property of the keyword." It is a property of the current SERP and of who is asking. A mid-range score is routine for the site that owns the subject and out of reach for a domain published last month. Relative difficulty, this term against this site, is the only version worth acting on.
"Filter to the lowest scores and publish." That filter selects for terms nobody has bothered with, which usually means nobody searches them with buying intent. Volume, intent and whether you can say something the top ten did not are better filters. The site check reads where your domain currently stands.
questions people ask
How is keyword difficulty calculated?
Every tool uses its own formula, but nearly all of them start from the backlink profiles of the URLs ranking in the top ten: how many referring domains point at them, and how strong those domains score in that vendor's index. Some blend in SERP features, the age of the ranking pages or estimated click share. The output is then normalised onto a 0 to 100 curve specific to that vendor's data, which is why the same term returns different numbers in different tools. No search engine publishes or endorses any of these scores.
What is a good keyword difficulty score to target?
There is no universal threshold, because difficulty is relative to your site rather than absolute. A domain with established rankings in a subject can take mid-range scores; a site three months old usually cannot, whatever the number says. A more useful filter is your own Search Console data, because terms where you already appear on page two or three are proven relevant and cheap to improve. When you do use KD, compare it against the scores of terms you already rank for, rather than against a fixed cut-off someone published in a blog post.
Is keyword difficulty still useful in 2026?
Yes, for one job: sizing the link gap on competitive commercial terms before you commit months to them. That is what it measures, and it does it reasonably well. It is much weaker on long-tail and question-shaped queries, where relevance and specificity decide the outcome and the underlying link sample is thin or missing. Treat KD as a filter for the head of your list and largely ignore it for the tail, where the real test is whether you can say something the existing results did not.
Does keyword difficulty affect whether AI assistants cite you?
Not directly. An assistant answering a question issues several narrow searches, fetches a small set of pages and quotes passages from them, and that fetch turns on how well a passage matches the sub-question rather than on domain strength. Pages that never reach position one for a difficult term still get quoted when they carry a specific, checkable fact. The practical consequence is that a high difficulty score is a reason to expect slow ranking progress, not a reason to skip the topic.
See what search engines and AI assistants find on your site
Free, no account. Type your address and we show you what is missing and what we would write first.