Keyword difficulty and search volume in 2026: why the numbers disagree

Keyword difficulty (KD) and search volume are modeled estimates, not measurements, and different tools produce very different numbers for the same keyword. Treat each score as a relative ranking inside one tool, never as an absolute fact about Google. Then validate the shortlist against your own Search Console data and the live results page before you commit a piece of content to it.

This is the companion to our keyword research guide for queries with no volume yet. That post covers demand the tools cannot see. This one covers the demand they can see, and how far to trust the numbers they give you.

How much do keyword difficulty scores disagree?

A lot. Ahrefs’ own keyword difficulty guide compares four tools on three keywords, and its own conclusion is that “the numbers will vary quite substantially”:

Same keywords, same day, and “bike tire pump” is 13 in one tool and 56 in another. A team that filters on “KD under 30” in one tool is selecting a different set of keywords than a team doing the same in another. (This table is vendor-published and was a snapshot when written; the point is the spread, not the specific values.)

Why the scores differ

Each vendor builds KD from different inputs and none of them publishes a full formula.

• Ahrefs looks at the top 10 ranking pages and counts the websites linking to each, on a 0 to 100 scale. Per Ahrefs, a KD near 0 to 5 means top pages have barely any backlinks, around 50 means a couple of hundred, and 90+ means thousands.
• Semrush blends in more variables. Public descriptions of its score cite referring domains, link quality and Authority Score, and SERP features, which is why its numbers often run higher than Ahrefs’ for the same term.
• Moz leans on the domain and page authority of the pages that rank, so it is more sensitive to how strong the competing sites are overall than to individual page backlink counts.

The common weakness: all of them are mostly link-based proxies. They say little about whether the top results match the search intent, how fresh they are, or whether an AI answer now sits above the organic listings. A KD of 20 on a query whose top results are all genuinely excellent is harder than the number says. A KD of 60 on a query where the top results are thin forum threads can be easier.

Is search volume accurate?

Not exactly, and the vendors say so. Ahrefs’ help center states that search volume in any SEO tool, including its own, is an estimation. Three known distortions:

• Keyword Planner groups and rounds. Google’s Keyword Planner merges keywords with similar meaning, and without an active Google Ads spend it often shows ranges such as 1K to 10K instead of a number. Ahrefs’ GSC vs. Keyword Planner study of 72,000 keywords found Keyword Planner tends to overestimate, in part because grouped variants and misspellings are added to the head term.
• Tools disagree. One practitioner comparison of “invoice software for small business” reported 900 in Ahrefs, 720 in Semrush, a 1,000 to 10,000 range in Keyword Planner and 480 in Ubersuggest. That is an anecdote, not a benchmark, but it matches what you see in practice.
• Averages hide seasonality and spikes. A monthly average smooths over a keyword that peaks for six weeks a year, and lags a topic that has just started growing.

Volume also stops being the right target as AI answers take clicks. If you want that side of the picture, see our AI Overviews CTR statistics and zero-click SEO guides: a keyword with healthy volume can still deliver few visits.

A shortlisting method that does not trust one number

We use the tools to build a candidate list and a second pass to decide.

1. Pull candidates, then normalize across two sources. Take volume and KD from one tool, and sanity-check the top candidates in a second. If the two disagree by more than a factor of two on volume, treat the keyword as “unknown volume” rather than averaging.

2. Use your own data first. In Search Console, find queries where you already get impressions at positions 8 to 30. These are the cheapest wins, because Google has already decided your site is relevant. Our Search Console playbook for AI-era queries covers the pull. KD is least useful exactly here, since you have direct evidence of how you rank.

3. Read the actual results page. For each finalist, open the SERP and check:

• Do the top results match the intent you plan to serve, or are they a different format (product pages, tools, video)?
• How old are they? Results dated two or more years ago on a fast-moving topic are an opening for a fresher, better page.
• Is an AI Overview or featured snippet answering the question outright?
• Are the ranking sites ones you could plausibly outrank on authority and on first-hand material, such as data or examples they cannot copy?

4. Weigh business value, not just traffic. A keyword with 50 monthly searches from buyers who match your customer profile beats one with 5,000 searches from students. Score each finalist by how close the searcher is to a purchase conversation. Our internal linking strategy guide covers how to point that traffic toward service pages once it arrives.

5. Set a threshold relative to your site, not an industry rule. Ahrefs itself gives no universal KD cutoff, and says the right target depends on your authority, your ability to earn links, and your resources to satisfy the intent. Benchmark against keywords you already rank for: if you hold position 5 on KD 25 terms, KD 25 to 35 is a reasonable reach target, whichever tool you use.

Common mistakes

• Filtering on a single KD number. It hides high-value terms that sit just above your cutoff and surfaces low-KD terms nobody searches for.
• Comparing KD across tools. A “KD 30” from two vendors is not the same difficulty.
• Ignoring intent. A low-KD keyword that wants a calculator or a product page will not rank for a blog post.
• Chasing volume alone. Volume does not account for click loss to answer boxes, or for whether the visitors are buyers.
• Never revisiting. Check rankings 60 to 90 days after publishing and compare with the estimate. Over time that gives you your own calibration for how a tool’s KD maps onto your site.

FAQ

What is a good keyword difficulty score? There is no universal number. Because each tool calculates KD differently, a good target depends on your site’s authority and the tool you use. A practical test is to compare against keywords you already rank on page one for, and aim a little above them.

Why is keyword difficulty different in Ahrefs and Semrush? They use different inputs. Ahrefs scores mostly on backlinks to the top 10 pages, while Semrush blends in additional factors such as authority and SERP features. Neither publishes its exact formula, so scores are not interchangeable.

How accurate is search volume? It is an estimate. Ahrefs states that search volume in any SEO tool is an estimation, and Keyword Planner groups similar keywords and rounds into ranges. Use volume to rank keywords against each other, not to forecast traffic to a precise number.

Should I ignore keyword difficulty? No. It is a quick filter for finding candidates. Just do not let it make the final decision: check the live results and your own Search Console data.

Can I rank for a high-difficulty keyword? Yes, if your content is better matched to the intent, you have links or first-hand material competitors lack, or the top results are weaker than their link profiles suggest. Check the SERP before dismissing a term on score alone.

Further reading:

• Keyword research in 2026: finding demand that has no volume number yet
• How to write a blog post for SEO: turning a chosen keyword into a page
• Content decay audit: re-checking what you already rank for
• Technical SEO and content services: how we run keyword research for clients
• Talk to us: we build keyword strategies from Search Console data, not tool scores alone