White-label links & AI search for agencies
Get a tailored proposal in 1 business day — no contracts.
James AllsoppYes — on the four websites we tested, Ahrefs' Organic Keywords count was dramatically lower than what Google Search Console showed was actually happening. In one case Ahrefs reported just 1 US keyword while Semrush reported 2,192, and GSC confirmed broad real-world visibility. The most likely explanation is coverage, not accuracy: Ahrefs' own documentation admits its keyword count is "almost always smaller" than reality — the scale we saw in 2026 was simply far larger than normal.
There has always been disagreement between SEO tools.
Put the same website into Ahrefs, Semrush or another platform and you rarely get identical keyword counts, traffic estimates or ranking positions. Anyone who regularly uses multiple SEO tools will be familiar with that.
A small discrepancy is normal.
What caught our attention in August 2026 was something very different.
On several websites we monitor, Ahrefs appeared to be reporting extraordinarily few organic keywords — sometimes dozens where Semrush reported hundreds, and in one country database just one keyword where Semrush reported more than 2,000.
These weren't dead websites.
They had tens of thousands of impressions recorded directly by Google Search Console.
So instead of assuming Semrush was inflating its figures or Ahrefs was undercounting them, we decided to test the numbers.
We compared current Ahrefs and Semrush data against first-party Google Search Console data from four websites to which we have verified access.
The results were striking.
“We didn't start this research trying to prove that Ahrefs was wrong. The keyword numbers simply became so low across several unrelated sites that they stopped matching the search visibility we could see ourselves inside Google Search Console. That was the point where we decided it needed testing.”
— James Allsopp, iNet VenturesThe study was carried out on 20 August 2026.
We used four unrelated websites under our control but have anonymised them as Sites A–D. We are deliberately not publishing their individual ranking keywords, URLs or commercially useful search opportunities.
For each website, we exported 28 days of Google Search Console Performance data, using Web search with no keyword filter.
We then compared that visibility with current Ahrefs and Semrush organic-ranking data.
An important part of the methodology was avoiding misleading global comparisons.
Ahrefs and Semrush don't necessarily aggregate worldwide keyword counts in precisely the same way, so the most useful comparisons in this study use the same website and the same country database.
This is also why anyone comparing competing SEO analysis tools should be cautious about treating two headline dashboard numbers as automatically equivalent.
Our Search Console data was not treated as a conventional rank tracker either. GSC position is an average affected by dates, devices, countries and increasingly different types of search result.
More on that later.
Across the four websites, Google recorded 134,338 impressions during the 28-day test periods.
The four Search Console exports contained 4,000 query rows.
Their average-position distribution was:
| Google Search Console position | Query rows |
|---|---|
| Positions 1–3 | 183 |
| Positions 4–10 | 543 |
| Positions 11–20 | 379 |
| Positions 21–50 | 1,660 |
| Positions 51–100 | 1,233 |
| Position 100+ | 2 |
| Total | 4,000 |
In other words:
Yet the keyword universes being reported by Ahrefs could be dramatically smaller.
Here are some of the country-matched comparisons from the four domains.
| Test site | Country | Ahrefs organic keywords | Semrush organic keywords | Semrush/Ahrefs |
|---|---|---|---|---|
| Site A | US | 3 | 1,160 | 386.7× |
| Site A | UK | 0 | 503 | — |
| Site B | US | 31 | 192 | 6.2× |
| Site B | UK | 32 | 270 | 8.4× |
| Site C | US | 1 | 2,192 | 2,192× |
| Site C | UK | 2 | 438 | 219× |
| Site D | US | 14 | 418 | 29.9× |
| Site D | UK | 1 | 50 | 50× |
Across the seven comparisons in which Ahrefs returned at least one keyword, Semrush reported between 6.2 and 2,192 times more organic keywords.
The median difference was 50×.
That is far beyond the ordinary variation we'd expect between competing databases.
We dug into exactly how those country-by-country gaps break down in a dedicated Ahrefs vs Semrush accuracy comparison, which looks purely at organic-keyword coverage across the same four sites.
Flexible link building plans with allocation across clients. Track everything via your live dashboard.
Site C produced the most dramatic result.
Ahrefs reported:
1 organic keyword in the United States.
Semrush reported:
2,192 organic keywords in the United States.
If we only had those two figures, we couldn't determine which one was closer to reality.
Perhaps Semrush was filled with stale rankings.
Perhaps Ahrefs had intentionally become much more conservative.
But we also had Google Search Console.
Over the previous 28 days, Google exposed 1,000 queries in our export for that website, with:
It is therefore extremely difficult to regard one US organic keyword as a representative description of this site's actual Google search footprint.
It doesn't prove all 2,192 Semrush keywords were currently accurate.
It does provide strong evidence that Ahrefs was missing a substantial amount of genuine search visibility.
Site A generated 28,206 Google impressions during the 28-day period.
Its Search Console data contained:
Ahrefs nevertheless reported only three US keywords, while Semrush reported 1,160.
Site D generated 37,502 impressions.
Its Search Console data contained an extraordinary:
Semrush found 418 US keywords.
Ahrefs found 14.
Site B was considerably less extreme, which is actually useful for the credibility of the study.
Semrush found 270 UK keywords compared with 32 in Ahrefs — still an 8.4× difference, but nowhere near the 2,192× outlier.
The pattern nevertheless remained the same on all four domains.
For keyword coverage, our results strongly favour Semrush.
But there is an important distinction between precision and recall.
Suppose a website genuinely ranks for 1,000 queries.
Tool A identifies 50 of those queries and all 50 are correct.
Tool B identifies 700, but perhaps 30 are stale or incorrect.
Tool A may technically have better precision.
Tool B still provides a much more complete representation of the website's search footprint.
That is important for competitor research.
If you're performing keyword research or analysing a competing site's content strategy, missing hundreds of long-tail rankings can lead you to underestimate entire pages, topics and content clusters.
It can also undermine competitor research because the site may look almost invisible in one platform despite clearly appearing across a broad range of Google searches.
“For competitor research, coverage matters enormously. If a tool shows almost no keyword footprint while Google is recording hundreds of real search queries, you can draw completely the wrong conclusion about how strong that competitor actually is.”
— James Allsopp, iNet VenturesOur evidence therefore does not suggest that the individual keywords Ahrefs does report are generally fictitious or bad.
Quite the opposite.
On Site A, all five of the worldwide Ahrefs keywords we examined were also present in its recent Search Console data.
The problem looked much more like missing rankings than incorrect rankings.
This isn't entirely controversial.
Ahrefs' own current documentation explicitly warns users that its Organic Keywords number doesn't represent every keyword a website actually ranks for.
It states that the reported number is:
“almost always smaller than the actual number of keywords your website is ranking for.”
— Ahrefs documentationAhrefs explains that even its extensive database does not contain every possible Google query and notes that unpopular keywords can be removed during database updates. Ahrefs' Organic Keywords documentation
That's an important disclaimer.
The question raised by our study isn't whether Ahrefs misses some queries.
We already know it does.
The interesting finding is the scale of the difference we're seeing on certain websites in 2026.
Semrush describes its Organic Positions report as showing keywords for which a domain has pages appearing within Google's top 100 results in the selected database. Semrush's Organic Rankings documentation
So conceptually both tools are trying to answer a similar question:
For the keywords in our database, where does this domain appear within Google's organic results?
The enormous difference therefore appears to stem not from one tool measuring Top 10 and the other Top 100, but from their underlying keyword universes, collection infrastructure, refresh methods, thresholds and potentially SERP depth.
Those details matter considerably more than the headline number.
Organic keywords aren't unique in this regard.
Metrics such as Domain Authority, Domain Rating and Authority Score are likewise third-party measurements produced by different providers rather than Google metrics.
Our Domain Authority Checker makes the same distinction: DA, DR and Authority Score use different datasets and methodologies, so they're most useful when compared consistently rather than treated as interchangeable facts.
The same philosophy is behind our newer Bulk Domain Intelligence tool, which combines Ahrefs, Moz and Majestic data rather than relying on one provider alone. It checks more than 20 domain metrics and surfaces them together because a single number rarely tells the complete story.
Our Ahrefs/Semrush experiment demonstrates why that matters.
An “organic keywords” metric isn't Google's count.
It's a model's view of Google's search landscape.
There is another important piece of context.
In September 2025, Google stopped allowing the &num=100 parameter to behave as it previously had.
That parameter was widely used by SEO data providers because it made retrieving approximately 100 Google results in a single request far easier.
Ahrefs publicly acknowledged the impact.
In its September 18 update, the company said Google's change would:
“reduce the number of results we can check for each term beyond Google's first page.”
— Ahrefs, 18 September 2025 updateAhrefs' official SERP monitoring update documents the issue and subsequent recovery.
This was an industry-wide disruption rather than an Ahrefs-specific problem.
Semrush issued its own update on September 15, emphasising that its Top 10 and Top 20 reporting remained intact while its infrastructure adapted to Google's change. Semrush's response to the SERP collection change
For anyone using rank tracker tools, the economics of retrieving positions 21–100 suddenly became much more important.
No — and this needs to be made clear.
Ahrefs continued updating its announcement after September.
On October 29, 2025, it said it was gradually restoring the ability to show up to 100 results per query.
Then on November 14:
“We now request the top 100 results for all keywords in Rank Tracker.”
— Ahrefs, 14 November 2025 updateAhrefs noted that a small percentage of queries could still fall back without a complete Top 100 result set.
So it would be incorrect to claim that Ahrefs simply stopped tracking beyond page one and never recovered.
There is another important nuance, though.
That announcement specifically refers to Rank Tracker.
Our present investigation concerns the keyword universe presented through Site Explorer's organic keyword estimates.
The internal collection and sampling decisions behind those different datasets aren't fully public.
We can therefore say that Google's num=100 change created a significant collection problem and that Ahrefs later restored Top-100 Rank Tracker functionality.
We cannot prove that the current Site Explorer discrepancies are directly caused by the September 2025 change.
Links placed into aged, indexed content with ~5 day turnaround. No new posts needed.
Some people may reasonably ask whether deep rankings matter at all.
If a site is position 72, who cares?
Independent SEO consultant Brodie Clark made a good argument during the num=100 disruption:
“Personally, I'd ideally like to have the top 100 still.”
— Brodie Clark, independent SEO consultantHe explained that movement below position 20 can be useful when diagnosing SEO issues and identifying emerging performance. Brodie Clark's analysis of the change
We agree.
A page moving from position 84 to 43 to 18 isn't producing meaningful traffic at the beginning of that journey, but the progression can tell an SEO that something is starting to work.
Deep ranking data also helps with:
Losing that data makes a competitor's actual search footprint look smaller.
The num=100 story had another surprising consequence.
SEO Tyler Gargula analysed 319 Google Search Console properties after the change.
Search Engine Land reported that:
The Search Engine Land analysis connected the change to reduced visibility of deeper ranking terms.
One leading theory was that rank trackers retrieving 100-result SERPs had themselves been creating impressions recorded by Search Console.
Clark reported examples where impressions declined by as much as 47% after the parameter stopped working.
That means historical GSC impression data around September 2025 needs careful interpretation too.
Our study avoids that comparison problem because all four Search Console datasets come from August 2026, long after the initial disruption.
This is perhaps the most interesting historical comparison we found.
In 2022, Search Engine Land published a study comparing Google Search Console, Ahrefs, Semrush and Moz across 51 websites.
Among the three third-party tools, Ahrefs had the largest keyword dataset for:
50 of the 51 domains.
Semrush ranked second on 50 of 51.
On average, Ahrefs reported roughly twice as many ranking keywords as Semrush. Read the 2022 keyword database comparison
That is almost the complete opposite of what we're seeing now.
Our study isn't directly comparable: different websites, industries, dates and methodologies are involved.
But the contrast is fascinating.
In 2022:
In our 2026 country-matched sample:
Something about the relative coverage of these platforms appears to have changed significantly.
Whether that is primarily due to keyword database strategy, SERP collection economics, Google changes, data-refresh policies or other internal decisions isn't something we can establish externally.
But it deserves further investigation.
This isn't only an SEO-blog debate.
In May 2026, researchers Cristina I. Font-Julian, Carlos Lopezosa, Enrique Orduña-Malea and Isidro F. Aguillo published a peer-reviewed study in the Journal of Information Science comparing Ahrefs, Semrush, Serpstat and Ubersuggest.
They examined organic keyword counts, keyword volumes and URL-level web traffic for two large academic publishers.
Their conclusion was clear: there was significant variation and low overlap between the SEO platforms, to the extent that the choice of tool could materially influence the conclusions drawn from the data.
The researchers concluded:
“The differences between the tools are great enough for us to draw different conclusions in comparative studies.”
— Font-Julián et al., 2026 peer-reviewed studyThat's highly relevant to anyone benchmarking competitors.
If Tool A says a competitor ranks for 50 terms while Tool B says 2,000, those aren't simply different-looking dashboards.
They can produce entirely different strategic decisions.
We also found another extremely recent independent experiment.
GEO Toolbox compared Ahrefs and Semrush on three sites it manages in July 2026, then checked their estimates against Search Console.
| Site size | Ahrefs keywords | Semrush keywords |
|---|---|---|
| Small / new | 35 | ~1,100 |
| Medium | 367 | 2,089 |
| Large | 26,835 | 80,700 |
Semrush found more ranking keywords on all three. Read the independent Semrush vs Ahrefs 2026 test
The author appropriately warns that some geographic scopes weren't perfectly matched, so the exact multipliers shouldn't be overinterpreted.
But the direction is remarkably similar to our findings.
There's an important twist in that independent study.
When estimated organic traffic was compared with actual Search Console clicks, Ahrefs was closer on the two larger websites, while Semrush was closer on the smallest site.
That's perfectly possible.
Keyword coverage and traffic modelling are two different problems.
A platform can miss thousands of long-tail keywords but still model traffic reasonably well if those missing terms contribute very few clicks.
Conversely, a platform can identify substantially more genuine rankings but overestimate how much traffic they generate.
This is why our verdict is specifically about:
organic keyword coverage
rather than declaring one platform universally more accurate.
For backlinks, for example, Ahrefs remains a particularly useful platform, and cross-checking backlink metrics through tools such as our Backlink Analyzer remains part of our workflow.
There is another major reason why Google Search Console cannot simply be treated as a conventional rank tracker in 2026.
Google says appearances within AI Overviews and AI Mode are included in the normal Web performance data in Search Console. Google's official AI features documentation
That means a Search Console query isn't necessarily recording only a classic blue-link ranking.
Google's AI systems can also use query fan-out, issuing multiple related searches across subtopics and data sources before constructing an answer.
That creates new routes through which pages can appear in a search experience.
This is one reason we've increasingly been writing about broader AI search visibility rather than treating traditional rankings as the only visibility metric that matters.
Google's John Mueller clarified how Search Console handles links within AI Overviews:
“Basically an AIO counts as a block, so it's all one position.”
— Google's John MuellerSo if an AI Overview occupies the first position on a page, links inside that block can inherit the position associated with that block.
That is fundamentally different from saying:
this URL was the traditional number-one organic blue link.
This is why throughout our study we say:
“queries averaging a GSC position of 10 or better”
rather than claiming every one of the 726 queries represented a conventional Top-10 organic listing.
It's an important methodological limitation.
There's good news.
On June 3, 2026, Google launched dedicated Search Generative AI performance reports inside Search Console.
The reports provide separate views of impressions generated by features such as AI Overviews and AI Mode, while Google confirmed that those impressions continue to be included in the normal overall Performance data. Google's Generative AI Search Console announcement
Google initially rolled the feature out to a subset of websites.
The launch documentation lists reporting for impressions, pages, countries, devices and dates.
That should make it easier to separate AI visibility from traditional search visibility over time, although it still doesn't turn the ordinary Search Console query report into a conventional rank tracker.
For brands trying to understand this new layer, brand mentions and AI visibility increasingly need to sit alongside classic keyword tracking.
Discover our suite of free SEO tools to enhance your rankings, track performance, and uncover opportunities.
AI-generated results aren't confined to obscure informational queries.
In July 2026, Semrush published an analysis of more than 600,000 keywords across 10 industries.
It found that the share of commercial-intent SERPs triggering AI Overviews increased by an average of 71% during its six-month study period. Read Semrush's AI Overview study
That makes the traditional concept of “ranking keywords” increasingly incomplete.
A brand might now have:
That's a much richer visibility landscape than a single keyword count can express.
This may be the most useful lesson from the entire study.
If Ahrefs says:
28 organic keywords
that does not mean:
Google ranks this domain for exactly 28 queries.
It means something much closer to:
Ahrefs has detected this domain ranking for 28 unique keywords represented within the relevant Ahrefs dataset and methodology.
The same applies to Semrush.
A Semrush count of 4,000 isn't Google's official figure either.
The platform's own documentation defines it in terms of the keywords represented in its selected database where the domain has been observed within Google's Top 100.
These numbers are useful.
They simply need the correct label:
Based on the evidence in this test, Semrush is currently our preference when the objective is discovering the breadth of keywords for which a competitor may be ranking.
Ahrefs remains extremely valuable when working with backlink profiles and other SEO research.
Google Search Console remains the primary source when we actually control the site.
And increasingly, we'd cross-check several data providers rather than depending on one.
That approach is also reflected in the growing number of free SEO tools we're building internally. Different datasets answer different questions.
There is no single magic number.
This is original research, but it isn't a census of the web.
⚠️ Before you rely on these numbers, keep the following in mind:
Our conclusion is therefore deliberately narrower:
Across all four websites tested, Semrush showed substantially broader organic keyword coverage than Ahrefs, and first-party Google Search Console data consistently suggested that the sites' real search footprints were much broader than Ahrefs' keyword counts implied.
That's what our data demonstrates.
And in some cases, the gap is substantial.
Across four unrelated sites, we analysed:
In every major country comparison we performed, Semrush identified more ranking keywords than Ahrefs.
Where both returned at least one keyword, Semrush reported between:
6.2× and 2,192× more keywords.
The median difference was:
50×.
Those findings are directionally consistent with another independent 2026 comparison, while peer-reviewed academic research also confirms that SEO tools can produce dramatically different organic-keyword universes.
Perhaps most interestingly, historical research suggests the relationship used to be reversed: in a 2022 study Ahrefs beat Semrush for keyword breadth on 50 of 51 sites.
That's why we don't think this should be dismissed as:
“SEO tools always disagree.”
Of course they disagree.
The interesting question is why the direction and magnitude of the disagreement appear to have changed so dramatically.
Google's num=100 change is clearly part of the industry's recent data-collection history, although the evidence does not allow us to blame it directly for Ahrefs' current Site Explorer coverage.
The rise of AI Overviews complicates measurement further.
And as Google search becomes more fragmented, relying on one third-party metric is becoming increasingly risky.
Our biggest takeaway isn't:
Semrush good. Ahrefs bad.
It's this:
An SEO platform's “organic keywords” number isn't a measurement of Google. It's a measurement of that platform's view of Google.
And in 2026, those views can be wildly different.
Ahrefs itself says the Organic Keywords figure is almost always smaller than the true number of queries a site ranks for because its database does not contain every Google search query. Keywords can also be added and removed as its database changes.
Our testing suggests the scale of that under-reporting can be particularly large on some sites.
On every one of the four websites tested in August 2026, yes.
In the country-matched comparisons where Ahrefs returned at least one keyword, Semrush reported between 6.2× and 2,192× more keywords.
That doesn't prove every additional Semrush keyword is current, but Google Search Console showed substantially broader visibility than Ahrefs' smallest counts implied.
For keyword coverage in our sample, Semrush was the clear winner.
That doesn't make it universally more accurate. Traffic estimation, backlinks, ranking freshness and other metrics are separate questions.
For a website you own, Search Console is the most important first-party dataset because the impressions and clicks originate from Google.
However, Search Console isn't a traditional rank tracker. Its average position combines multiple searches and can include AI Overviews and AI Mode visibility.
It caused an industry-wide disruption to deep SERP collection, and Ahrefs initially said it would reduce how many results it could check beyond page one.
Ahrefs subsequently restored Top-100 retrieval in Rank Tracker. Therefore, we don't have enough evidence to say the current Site Explorer discrepancy is directly caused by num=100.
No.
Ahrefs remains an extremely useful SEO platform, particularly for backlink analysis.
The lesson from this study is simply that its Organic Keywords figure should not be treated as a complete count of everything a domain ranks for.
Use several sources.
Start by comparing the competitor across more than one SEO database, investigate important pages manually, and cross-reference authority and backlink signals rather than trusting a single metric.
For link-prospect and competitor-domain research specifically, our Bulk Domain Intelligence tool can compare DR, DA, Trust Flow, Citation Flow, traffic and other signals in one place.
Original studies like this have another advantage: they give publishers something worth citing.
Rather than emailing bloggers with a generic guest-post pitch, useful proprietary data creates an actual editorial reason for someone to reference your website.
Our complete blogger outreach guide covers the research, prospecting and outreach process in more detail.
And for agencies or brands that want the outreach handled for them, our Blogger Outreach service focuses on contextual editorial placements on vetted publishers.
Research gives you the asset.
Good outreach gets that asset seen.
Find answers to common questions about this topic
Explore services, view pricing and manage campaigns in one place.
No subscription required • 100% white-label
James is the founder and CEO of iNet Ventures, bringing over two decades of experience in SEO, link building, and online growth. He has helped online agencies scale through white-label solutions and has overseen the delivery of 25,000+ links, combining proven strategies with AI-driven workflows to drive efficient, long-term results.
Want Link Building Without The Hassle?
Our team handles outreach, placements, and reporting — 100% white-label, fully managed.
Start A CampaignRelated Articles
Continue exploring our latest insights and strategies for digital marketing success.
