Keyword difficulty: what it is, how it is calculated and how far to trust it

Keyword difficulty is a score from 0 to 100 that SEO tools use to estimate how hard it is to get onto the first page of Google for a keyword.

It is not a Google metric. Each tool calculates it with its own formula, nearly always starting from the backlinks of the pages already ranking, and for the same keyword the numbers disagree. It is useful for discarding hundreds of terms quickly during keyword research, not for deciding which one to work on.

How the tools calculate it

The formulas are proprietary, but almost all of them start from the same place: who already holds the top 10 and how strong they are.

Ahrefs looks at one thing only, the number of referring domains pointing to the ten ranking pages, and its documentation states that the metric takes no on-page factors into account. Semrush uses a wider set: the median number of referring domains to the ranking URLs, the median follow/nofollow ratio, the median Authority Score of the ranking domains, plus data specific to the query. Mangools (KWFinder) combines Domain Authority and Page Authority from Moz with Citation Flow and Trust Flow from Majestic, and gives more weight to the weakest pages on the first page. Moz works on the Page Authority and Domain Authority of the first-page results, two metrics Google does not use.

Tool	Scale	Based on	What it estimates
Ahrefs	0-100, non-linear	Referring domains to the top 10 pages	Likelihood of reaching the top 10
Semrush	0-100 (KD%)	Referring domains, follow/nofollow ratio and Authority Score of the ranking pages, SERP data	Effort needed to reach the first page
Mangools	0-100	DA, PA, Citation Flow and Trust Flow of the first-page URLs, weighted towards the weakest	Difficulty of reaching the first page
Moz	1-100	Page Authority and Domain Authority of the first-page results	Effort needed to overtake the pages already ranking

Two practical consequences follow. First, comparing KD 35 in one keyword research tool with KD 35 in another means nothing: they are different scales built on different data. Second, since nearly all of them look at links, keyword difficulty measures how well linked the top 10 pages are, and little else. On a SERP made of product pages with no backlinks, the number has very little to work with.

Going from 40 to 80 is not twice the effort

Ahrefs publishes the conversion between its score and the number of referring domains needed to reach the first page. It is the most useful document on the subject, because it takes apart the idea that the scale is linear.

Keyword difficulty	Estimated referring domains to reach the top 10
10	10
20	22
30	36
40	56
50	84
60	129
70	202
80	353
90	756

Moving from KD 10 to KD 20 costs twelve more domains. Moving from 80 to 90 costs over four hundred. That is why a new site targeting a KD 70 keyword is not slightly behind: it is behind by around two hundred referring domains, which in a narrow B2B sector can mean years of link building.

On the same page Ahrefs points out a limit that hardly anyone repeats: the score estimates the likelihood of reaching the top ten results, not of ranking first. Between tenth position and first there are factors the metric does not look at at all.

Keyword Planner competition is something else

This is the most common mix-up, so much so that Ahrefs flags it at the top of its own documentation: people who come to SEO from paid search read the Competition column in Keyword Planner as if it were a ranking difficulty.

The two measures have nothing to do with each other. The Google Ads documentation defines Competition as "the number of advertisers that showed on each keyword relative to all keywords across Google", shown as low, medium or high. It tells you how many advertisers are buying that query and how expensive the cost per click is, not how hard it is to rank organically.

Keyword	Monthly searches in the UK	Google Ads competition	Keyword difficulty	Average referring domains in the top 10
electric forklift	1,000	high (1.0)	7	18.5
seo agency	6,600	low (0.32)	42	124.4
what is seo	5,400	low (0.08)	63	4,734.5
car insurance	450,000	high (0.83)	50	195.4

"Electric forklift" has ad competition at the maximum and an organic difficulty of 7. Neither number knows anything about the other.

The same table says something else. "Car insurance" gets 450,000 searches a month and its top 10 pages have 195.4 referring domains on average; "what is seo" gets 5,400 and faces pages with 4,734.5. The keyword with about eighty times less traffic has far more heavily linked competitors. Volume does not predict difficulty, and sorting a list by descending volume is still the fastest way to work on the wrong things.

A low score is not low for everyone

Keyword difficulty is an absolute number answering a relative question. The tool estimates how strong the top 10 is; how strong you are does not enter the calculation.

Semrush says so in its documentation, almost as an aside: KD should always be read against the authority of the domain you are working on. In practice it is the opposite of an aside, it is the first thing to check. A site with ten years of history and a settled link profile gets into the SERP at KD 50 where a six-month-old domain does not get in at KD 20. Same score, two different projects, two opposite answers.

Semrush also publishes the bands it uses to read its own scale: 0-14 very easy, 15-29 easy, 30-49 possible, 50-69 difficult, 70-84 hard, 85-100 very hard. They are reasonable as a first orientation, as long as you remember they apply to the Semrush scale and to no other.

Where the number goes wrong most often

There are SERPs where a metric built on links has nothing to measure. For "safety boots", 18,100 searches a month in the UK, the first-page results have 1.9 referring domains on average. Ranking there is decided by other things: which product pages are there, how the listing is built, what search intent Google is serving. A backlink-based difficulty score, in that context, is thin air.

The same happens when there is little data. On low-volume B2B niches the tools have few sampled SERPs and the estimates wobble: "warehouse management software", 590 searches a month and an average CPC of £51.79 (converted from USD at £1 = $1.3242), has a KD of 24 and top 10 pages with 25.1 referring domains on average. The cost per click says the query is worth money, the links say nobody has taken it with real authority. Neither piece of information sits inside a difficulty score.

One limit of what I have just shown: I picked these keywords because they illustrate the point well, not at random. They show that the two numbers diverge, not how often they do.

How to use it without getting hurt

As a first filter, never as the final criterion. Ahrefs recommends this in its documentation and it works: take a list of three hundred terms, cut what is out of reach for the domain you are working on, and from there open the SERPs by hand.

What the score does not see, you see in thirty seconds of SERP: what kind of pages are there, whether the intent is what you expected, whether you are up against real competitors or directories and forums, how much room is left under the ads and the features. It is also when you work out whether two terms belong on the same page or on two, which is a keyword clustering problem, not a difficulty one.

Technical long-tail terms are where this changes the result most. On a manufacturing project the work focused on low-volume product terms, the ones a list sorted by volume would have thrown out: the main keyword moved from position 88 to 2, with a 25.55% share of voice ahead of Amazon. If you need help with this kind of shortlisting in a specific sector, it is the work every one of our SEO projects starts with.

One thing the industry rarely says: keyword difficulty answers the wrong question. It tells you how hard it is to get into the top 10, not how much being there is worth. On a query where the first organic result sits below an AI Overview, three ads and a questions box, a KD 15 can bring in less than a KD 60 on a narrow commercial search made by someone about to buy. Before difficulty, look at the Google results page and ask where the click would end up.

Keyword difficulty FAQs