What keyword difficulty actually is
Keyword difficulty (often shortened to KD) is a 0 to 100 score that tools assign to a keyword as an estimate of how hard it is to rank in the top ten of Google for that phrase. A score of 0 means the SERP is wide open. A score of 100 means you are competing against the most authoritative sites on the web.
The score is calculated, not measured. Nobody can directly observe how hard a keyword is to rank for. The tools build a model from observable signals: the backlink profiles of the pages currently ranking, the authority of the domains they sit on, the structural strength of the SERP, and (for some tools) historical click data. The output is a confident-looking single number, but the inputs vary and the output reflects whichever signals that tool's model weights heaviest.
How each tool calculates it
The three big tools each take a different angle. Public methodology pages and our own correlation work across hundreds of Australian SERPs over the last three years gives us this rough picture:
Ahrefs (KD)
Heavily weighted on the number of referring domains pointing at the top ten URLs. Ahrefs has historically said its KD is mostly a function of how many unique sites link to the currently ranking pages. The model is updated as their index updates, and the score tends to over-state difficulty in markets where backlink data is patchy, like Australia compared to the US.
SEMrush (KD %)
Blends backlink signals with on-page authority and a few SERP-feature factors (presence of featured snippets, AI Overviews, knowledge panels). Their score tends to come in slightly lower than Ahrefs for the same keyword. SEMrush also incorporates a recent search trend factor that the others do not.
Moz (Difficulty)
Built around Moz's own Domain Authority and Page Authority metrics. Tends to give the most conservative scores. Moz's link index is smaller than the other two, so the model under-weights some link-heavy SERPs.
Google itself
Does not publish a keyword difficulty score. Google has never told the world how it ranks pages. Anything called keyword difficulty is a third-party estimate.
Why the scores disagree
Take a typical Australian commercial keyword and you will routinely see splits like this:
- Ahrefs KD: 52
- SEMrush KD: 41
- Moz Difficulty: 36
The keyword has not changed. The SERP has not changed. The difference is in the model. Three reasons the numbers diverge:
- Different link indexes. Each tool crawls a different subset of the web. The same page can show 800 referring domains in one tool and 1,400 in another. The score is built on the index, so the score moves with it.
- Different model weightings. Ahrefs leans hard on referring domain count. SEMrush blends in SERP features. Moz uses its own DA scale. The same input data produces different outputs.
- Different calibration markets. All three tools were calibrated primarily against US SERPs. Australia has thinner backlink graphs and different SERP dynamics, so the scores skew differently in Google.com.au.
The directional signal almost always agrees. If Ahrefs says 78, SEMrush will say 65 and Moz will say 60. All three are signalling "hard". If Ahrefs says 18, the other two will be in the teens. Trust the direction. Be sceptical of the exact number.
How to use difficulty in real planning
Three rules we run for every Perth client retainer.
Rule 1: Pair difficulty with your domain's reach
What KD bracket a site can realistically attack depends on its age, ranking footprint and backlink profile. Rough bands we use:
| Site profile | Realistic 12-month KD ceiling |
|---|---|
| Brand new domain, no ranking pages | KD 0 to 20 |
| Under 12 months, fewer than 50 ranking keywords | KD 0 to 30 |
| 1 to 3 years, decent topical footprint | KD 0 to 50 |
| Established Australian site, hundreds of ranking keywords | KD 0 to 70 |
| Authority site, strong backlink profile | KD up to 90, with patience |
These are guidelines, not laws. Use them to demote unrealistic keywords from your near-term plan, not to ban them forever.
Rule 2: Always verify against the live SERP
Before committing to a keyword, open it in Google.com.au and read the top ten. If the SERP is dominated by domains with weaker authority than yours, the score is overestimating difficulty for your situation. If it is dominated by global giants like Ahrefs, support.google.com or Wikipedia, the score is underestimating. The SERP is the truth. The score is the model's guess.
Rule 3: Mix the difficulty bands in your plan
A healthy keyword plan has a deliberate spread:
- 60 to 70 percent low and medium difficulty (KD under 30): these are your near-term wins. Pages can reach the top ten inside three to six months. They build momentum and rank a real percentage of your content.
- 20 to 30 percent medium-high difficulty (KD 30 to 60): the medium-term plays. Twelve months to rank. These are usually your commercial money pages.
- 5 to 10 percent high difficulty (KD 60 plus): the long-game head terms. You build supporting content around them while the authority compounds.
A plan that is 100 percent low-KD is fine but you cap your traffic ceiling. A plan that is 100 percent high-KD is a vanity exercise. The mix is what compounds.
When the SERP overrides the score
The KD number is the model's guess. The live SERP is the actual evidence. Five situations where you should override the score:
- The top ten is full of weaker domains than the score implies. If KD says 60 but you are looking at a SERP of small Australian businesses none of whom have done any real SEO, the keyword is winnable.
- The top ten is mismatched on intent. Sometimes a SERP shows pages that do not actually answer the query well. That is a gap. A purpose-built page can leapfrog ten OK pages.
- The SERP has dramatic recency bias. Some queries pull pages updated in the last 30 days. A fresh, well-written page can outrank an older, technically stronger one if recency dominates.
- The SERP is local-pack heavy. If half the SERP is taken by a Google Business Profile pack, organic difficulty is lower than the score suggests because the organic competition is tighter. See local keyword research.
- The SERP has high churn. If you check the same keyword three weeks running and the top ten shuffles each time, the SERP is unstable and your fresh page can squeeze into the rotation.
Common mistakes
- Treating KD as a relative ranking inside one tool, not as a universal score.
- Verifying every shortlisted keyword's difficulty against the live Australian SERP.
- Setting tool location to Australia, not the default US.
- Pairing difficulty with the site's realistic 12-month ceiling.
- Building a deliberate mix of low, medium and high KD across the plan.
- Picking keywords purely by KD without checking intent or SERP composition.
- Comparing KD scores across tools as if they meant the same thing.
- Ignoring difficulty entirely and chasing whatever has the highest volume.
- Treating keyword difficulty scores like a ranking factor. They are a third-party estimate, not what Google uses.
- Building a plan with no low-KD content. You leave the near-term wins on the table.
Perth and WA context
Three observations from running difficulty assessments across Australian client work:
AU difficulty is usually lower than US difficulty for the same phrase. A keyword that scores 65 in Ahrefs default (US) might score 45 in Google.com.au. Always set tool location to Australia before reading scores. Most agencies do not, which is why their plans look more pessimistic than they need to be.
Local commercial keywords often score lower than the SERP suggests. "Plumber Fremantle" might score KD 25 in Ahrefs, but the actual top ten can be wide open if the local plumbers have not done any SEO. We have lifted clients to position 1 for keywords the tools rated as harder than they really were. Check the SERP. See SEO Fremantle and the Local SEO Perth service for the pattern.
Mining and resources B2B keywords often score artificially high because of the link mass of large industry publishers. "FIFO recruitment Karratha" might look hard on paper because the Australian Mining publication ranks for it. The actual organic competition between service providers is often light. A focused service page can rank well in spite of a 50-plus KD score. See SEO Karratha and mining SEO.
For the broader research workflow that uses difficulty as one of four filters, see how to do keyword research. For the search intent side that pairs with difficulty, see understanding search intent. For what to do once you have your KD-filtered keyword map, the On-Page SEO pillar covers the page craft and internal linking is how you build the topical authority that, over time, lifts your realistic KD ceiling.