We typed one keyword into four tools for keyword research on the same day, with every tool set to the US database. We got four search volumes, four difficulty scores and three different CPCs. This article explains where those gaps come from, which numbers you can lean on, and what a paid plan changes.
Why Keyword Data, Not Features, Is the Real Comparison
Most keyword tool comparisons line up feature lists. That misses the part you actually use every day: the numbers. Our October 21, 2026 test log shows US search volumes from 3.6K to 9.9K for a single keyword, so the tool you pick can change which topics look worth writing about.
Why does that matter so much? Because a keyword plan is a ranking of numbers. If one tool says a term gets 9.9K searches a month and another says 3.6K, the term moves up or down your list depending on the screen you happen to open.
We picked Semrush, SE Ranking, Mangools and Ubersuggest because they are the four suites we had already put through the same task list in our side-by-side test of four SEO suites on one website. That article covers audits, backlinks and rank tracking. This one stays on keyword data and goes deeper than the one-line summary table there.
Key Takeaways
No keyword tool counts real searches; every figure is an estimate. Ahrefs, a tool we did not test, says its own volume estimates were “roughly accurate” for 60% of the keywords it studied when checked against Google Search Console impressions, and Google Keyword Planner for 45% (Ahrefs Help Center). In our own test, three of the four tools agreed on the rough size of our keyword and one stood well apart, while difficulty scores barely agreed at all. So should you trust any single number? Compare keywords inside one tool, and use a second tool only to sanity-check size.
Quick Verdict: Which Keyword Tool to Trust for What
Each tool earned a different job in our test: Semrush gave the richest single report (seven metrics plus a country split), SE Ranking gave the most cautious volume (3.6K), Mangools was the only one to show a trend figure in our lookups (-42%), and Ubersuggest gave the lowest difficulty (10) by a wide margin, per our October 21, 2026 test log.
| If you want… | Lean on | Why, from our test |
|---|---|---|
| The most data per lookup | Semrush | Volume, global volume, country split, KD, intent, CPC, competitive density and ads count on one screen |
| A conservative volume to plan around | SE Ranking | Lowest US volume of the four (3.6K) and a country table to match |
| A quick trend check | Mangools | KWFinder showed a -42% trend next to the volume |
| A second opinion on CPC | Ubersuggest | Highest CPC of the four ($28.56, labeled “HIGH”) |
| A difficulty score you can act on | Semrush | Its KD came with a concrete target: “124 high-authority referring domains” |
So which keyword research tool is best? There is no single winner on data quality, because nobody outside Google can check the true count. The better question is which tool’s numbers you will read the same way every week. Pick that one as your main tool.
How We Tested: One Keyword, Four Tools, One Day
We ran one lookup per tool on October 21, 2026, using the same work account, the same English keyword, “generative engine optimization,” and the US database in every tool. The term names a young topic (what generative engine optimization means), so no tool could lean on years of stable history.
Here is exactly what we did in each tool:
- Semrush: Keyword Overview, database United States.
- SE Ranking: Keyword Research, database switched from Taiwan (the default for us) to the US.
- Mangools: KWFinder, location United States, language English.
- Ubersuggest: AI Keyword Overview, language switched from Chinese (Traditional Han) to English and location from Taiwan to United States.
Two tools defaulted to our physical location, which is Taiwan. That detail turned out to matter, and it gets its own section further down.
We copied every number from the screen into our test log the same day and kept masked screenshots. We did not open paid plans, and we did not run the same keyword twice. Is one lookup enough to judge a tool? No. A single lookup is a snapshot, not a study. What it does show well is how far apart four tools can land on the same question.
The Numbers Side by Side
For one keyword on one day, the highest US volume (9.9K in Semrush) was about 2.75 times the lowest (3.6K in SE Ranking), and difficulty ran from 10 to 78, per our test log. Here is every number the four tools showed us, not just volume and difficulty.
| Metric (US database) | Semrush | SE Ranking | Mangools KWFinder | Ubersuggest |
|---|---|---|---|---|
| Search volume | 9.9K | 3.6K | 4.1k | 4.4K |
| Global volume | 22.9K | 13.8K | Not shown | 22.2K |
| Difficulty | 72% (“Hard”) | 78/100 (“NOT EASY”) | 51 | 10 (“LOW”) |
| Search intent | Informational | Informational | Not shown | Informational |
| CPC | $14.49 | $11.78 | N/A | $28.56 (“HIGH”) |
| Paid competition | 0.60 (Competitive Density) | 0.54 | N/A (PPC) | Not shown |
| Other fields | Ads 12 | Country table | Trend -42% | SERP Features 3, Top Pages Backlinks 4.8K |
What jumps out once the full set is on one screen? Three patterns.
Volume clusters, except for one tool. Three of the four estimates sit within 800 searches of each other, from 3.6K to 4.4K. Semrush is the outlier at 9.9K. The average of all four is 5.5K, but the median is about 4.25K, which describes the group better.
Global volume disagrees just as much. Semrush and Ubersuggest were close on worldwide searches (22.9K and 22.2K). SE Ranking said 13.8K. So the US share of global demand came out at 43% in Semrush, 26% in SE Ranking and 20% in Ubersuggest, using each tool’s own two numbers.
Difficulty does not cluster at all. Three tools put the keyword between 51 and 78. Ubersuggest put it at 10. That is not a rounding difference; it is a different opinion about the same results page.
For a metric-by-metric reading of the Semrush report alone, including what Competitive Density and the ads count mean, see our walkthrough of what each of Semrush’s seven keyword metrics means. The rest of this article asks a different question: why do the tools disagree?
Why Search Volume Differs Between Tools
Search volume differs because each tool starts from different raw data and applies a different recipe: Semrush, for example, overlays third-party data with historical clickstream data and runs it through a machine-learning model (Semrush Knowledge Base). No tool here showed us a raw count from Google. Here is what each vendor documents about its own number, from pages we read on October 23 and 24, 2026.
| Tool | What its volume is, per its own documentation |
|---|---|
| Semrush | Average monthly searches over the past 12 months, updated monthly, mobile and desktop combined; modeled from third-party data plus clickstream data |
| SE Ranking | “Ungrouped search volume”: plurals, superlatives and possessives are separate keywords with exact volumes, across 188 country databases |
| Mangools KWFinder | Google is “one of our main data sources”; close variants are not grouped, so you see the volume for the exact phrase |
| Ubersuggest | Data gathered “from Google’s API, with updates occurring monthly” |
| Google Keyword Planner | Average searches for a keyword “and its close variants,” averaged over 12 months by default, and rounded |
Sources: Semrush KB, SE Ranking Keyword Tool, Mangools, Ubersuggest Help Center, Google Ads Help.
Grouping is the biggest hidden variable
Google Keyword Planner adds close variants into one number. Mangools and SE Ranking both say they split those variants apart. Mangools gives a plain example: Keyword Planner may fold “apple iphone 7,” “iphone 7 apple” and “apple iphone 7 price” into one phrase, while KWFinder shows each one on its own (Mangools).
What does that do to a young keyword like ours? A tool that groups variants will report a bigger number for the head term. A tool that splits them reports a smaller one and spreads the rest across the variants. Neither is wrong; they answer slightly different questions.
Averages hide a moving topic
Semrush’s figure is a 12-month average: it divides the past year’s total by 12, and says the number can be higher or lower than the average in a given month for seasonal and trending keywords (Semrush KB). Mangools showed our keyword trending down 42%. Put those together and the gap makes more sense. A topic that peaked earlier in the year can carry a high yearly average while recent months run lower. That is our reading of the two numbers, not something either vendor told us.
Every vendor’s study crowns its own tool
Can a study settle which number is right? Semrush ran one in 2021 and 2022, using anonymized Search Console impressions as the “true” volume for 10,000 US keywords, and concluded Semrush was closest most often (Semrush blog). Ahrefs ran its own and reports that Keyword Planner “drastically overestimated search volumes 54% of the time” in 2021 (Ahrefs Help Center). Mangools quotes an outside review that ranked KWFinder’s difficulty first.
Here’s the catch: each study was run or quoted by the vendor that wins it. Treat them as evidence of method, not as a verdict. If you want a real benchmark for your own site, use your own Search Console impressions for keywords you already rank on page one for. Our guide to using Google Search Console for SEO shows where those impressions live.
Keyword Difficulty Is Not One Scale
Keyword difficulty is four different formulas that share a name and a 0 to 100 range: our keyword scored 72% in Semrush, 78 in SE Ranking, 51 in Mangools and 10 in Ubersuggest on the same day, per our test log. Reading one score with another tool’s yardstick is the most common mistake we see.
So what goes into each score? Here is what each vendor documents:
| Tool | Main inputs | How the vendor reads its own score |
|---|---|---|
| Semrush KD% | Median referring domains of the ranking URLs, median dofollow/nofollow ratio, median Authority Score of the ranking domains, SERP qualities | Six bands; 70–84 is “Hard,” which needs “a lot of effort” on better backlinks |
| SE Ranking | How difficult it is to get a page into the top 10 organic results | Shown with a word label; ours read “NOT EASY” |
| Mangools KD | Link Profile Strength of each first-page site, built from Moz Domain Authority and Page Authority and Majestic Citation Flow and Trust Flow | Weighs low-strength sites “more than ever,” because weak pages often still rank |
| Ubersuggest SD | Domain authority, backlinks and content of the pages ranking now | “10% might be relatively easy,” 70% or higher “is likely to be more competitive” |
Sources: Semrush KB, SE Ranking Keyword Tool, Mangools, Ubersuggest Help Center.
Translate each score into its own vendor’s words
Read through each vendor’s own guidance, the four results stop looking random:
- Semrush 72% lands in its Hard band, two bands above “Very easy” (0–14), which Semrush calls the best place to start new pages. The note on screen made it concrete: the keyword “will demand 124 high-authority referring domains.”
- SE Ranking 78 came with the label “NOT EASY.” Its public keyword tool page describes the score as how hard it is to reach the top 10; ask SE Ranking support if you need the exact weights.
- Mangools 51 sits near the middle of its scale. Because Mangools gives extra weight to weaker first-page sites, a mid score can still hide a few very strong ones.
- Ubersuggest 10 falls in the range its own guide calls “relatively easy.” The same guide tells you to “manually review the current top 10 results in Google,” and for this keyword that advice outweighs the score.
What surprised us was how the outlier changed. On volume, Semrush stood apart. On difficulty, Ubersuggest did. So no single tool is “high” or “low” across the board; each one is off from the group on a different metric.
Mangools makes the practical point plainly: compare difficulty “within one tool,” because the order of keywords inside one tool matters more than the gap between tools (Mangools). Our look at how KWFinder scores difficulty with Moz and Majestic data goes through the Link Profile Strength idea in more detail.
CPC and Commercial Signals
CPC was the one metric where two tools landed far apart in the other direction: Ubersuggest showed $28.56 per click, about 2.4 times SE Ranking’s $11.78, with Semrush in between at $14.49, per our test log. Mangools showed no CPC or PPC figure at all for this keyword (both N/A).
Why would ad prices differ when they all describe Google Ads? CPC is a market estimate, not a list price. Google’s own Keyword Planner gives a range, not one number: a “top of page bid (low range)” near the 20th percentile of what advertisers paid, and a high range near the 80th percentile (Google Ads Help). A tool that reports one CPC has picked a point somewhere in that spread, or modeled its own.
The paid-competition figures lined up better. Semrush showed a Competitive Density of 0.60 and 12 ads; SE Ranking showed competition of 0.54. Keyword Planner’s own Competition column instead reads low, medium or high, based on “the number of advertisers that showed on each keyword relative to all keywords across Google” (Google Ads Help).
One warning from Mangools is worth repeating. Keyword Planner’s Competition “estimates the competition for paid keywords in PPC campaigns and has nothing to do with organic results and SEO” (Mangools). Semrush draws the same line: CPC and Competition level “relate to Google Ads and Keyword Difficulty relates to SEO” (Semrush KB).
So how should a blogger use CPC? As a signal of commercial value, not as a forecast. Our keyword is informational in all three tools that showed intent, yet every CPC we saw was above $11. That usually means a buying decision sits a few searches later, and it is a reason to write the guide with a clear next step for readers.
Extra Data Each Tool Showed: Intent, Trends, Countries
Beyond volume and difficulty, each tool added one or two fields the others did not, and the country split was where they disagreed most: Semrush put Germany at 2.4K searches, while SE Ranking put Germany at 590, per our test log.
| Country | Semrush | SE Ranking |
|---|---|---|
| United States | 9.9K | 3.6K |
| India | 2.4K | 1.9K |
| Germany | 2.4K | 590 |
| Netherlands | 1.9K | 590 |
| United Kingdom | 1.6K | 500 |
| France | 1.3K | Not listed |
| Canada | Not listed | 590 |
Why does this matter if you only write for US readers? Because the ratio between countries is a quick way to check a tool’s model. Both tools put the US first and India in the top three. After that they part ways: Semrush rated Germany about four times higher than SE Ranking did, and the UK about three times higher. If you plan a translated version of a page, those gaps decide whether it looks worth the effort.
The other extra fields, tool by tool:
- Semrush showed an ads count of 12 next to Competitive Density, and a line under the report: “Go beyond basic volume and discover advanced metrics with Semrush Pro.”
- SE Ranking paired its global volume with the country table above. The counter after our lookup read “Account limit: 3/10.”
- Mangools KWFinder showed a Trend of -42%. Its related keyword list drifted toward broad agency terms such as “marketing agency” (60.2k) and “seo services” (51.7k), so plan to filter the list by hand.
- Ubersuggest showed 3 SERP features and 4.8K backlinks across the top-ranking pages. Its volume trend chart and AI summary were locked on the free plan (“Start free trial to unlock data”).
Working on a channel where Google’s numbers do not apply, such as video? The same estimate problem shows up there with different tools; our guide to YouTube SEO tools covers YouTube Studio and the browser extensions.
Country and Database Settings That Change the Answer
The database setting can change the answer more than the choice of tool: in Ubersuggest, a project left on the Taiwan market showed our keyword at 140 searches with SEO Difficulty 27, against 4.4K and 10 in the US database, per our test log. That is roughly a 31-fold gap from one setting.
Two of our four tools picked a market for us before we typed anything:
- Ubersuggest set the Keyword Overview form to Chinese (Traditional Han) and Taiwan, and its project setup did the same for rank tracking. We switched the lookup to English and the US. The project’s tracking stayed on Taiwan.
- SE Ranking set both the project location and the keyword database to Taiwan. We switched both to the US.
Our Semrush and Mangools logs record only the US settings we used. Whatever tool you use, read these settings before pressing search: a lookup on the wrong market still uses up one of your daily searches.
The vendors document the same effect. Semrush says volume “can change when you change a regional or a local database” and offers local data down to a city or region (Semrush KB). Mangools’ advice for comparing any two tools is to “set the same location and language” first (Mangools). Google’s Keyword Planner notes its volumes are rounded, so totals across several locations “might not add up as you’d expect” (Google Ads Help).
Is the difficulty change a bug? No. A smaller market has a different first page, with different sites on it, so a different difficulty is what the formula should produce. Our account of how Ubersuggest’s project setup locked us into the Taiwan market shows the five setup steps where that happens, and our SE Ranking trial notes on location defaults cover the same issue there. For the other settings worth fixing before a first search, see our beginner’s guide to using SEO tools effectively.
Lookup Limits and What Paid Plans Add
Free plans cap how many keywords you can check, from 3 searches a day in Ubersuggest to 10 reports a day in Semrush, and paid plans mostly buy more lookups rather than better numbers, per our test log and each vendor’s pricing page. Here are the caps we saw while testing.
| Tool | Free option we used | Keyword lookup cap we saw |
|---|---|---|
| Semrush | Free account | 10 reports per day across tools (Semrush KB) |
| SE Ranking | 14-day trial, no card | “Account limit: 3/10” after one lookup |
| Mangools | Free plan | 5 keyword lookups per 24 hours |
| Ubersuggest | Free plan | 3 free searches per day, shared with backlink lookups |
Those limits deserve their own comparison, and we wrote one: our free keyword research tools compared by daily limits covers what each free account shows and how far it stretches.
What a paid keyword research tool adds
On the paid side, here is what each public pricing page listed for keyword research on October 23, 2026, in US dollars:
| Tool | Entry plan | Price | Keyword research on that plan |
|---|---|---|---|
| Semrush | SEO | $139/month, or $117.33/month billed annually | “Basic keyword research and competitor analysis”; Starter ($199/month) adds “Keyword research and optimization tools” |
| SE Ranking | Core | $129/month, or $103.20/month billed annually | “Unlimited keyword, competitor, backlink research” |
| Mangools | Basic | $37.70/month billed annually ($452.40/year) | 100 keyword research requests per 24 hours; Premium raises it to 500 |
| Ubersuggest | Individual | Not quoted (lifetime countdown offer on the page) | 150 searches per day; Business 300, Enterprise 900 |
Sources: Semrush pricing, SE Ranking pricing, Mangools pricing, Ubersuggest pricing.
Notice what the right column does not promise: different search volumes. None of the four plan descriptions promises more accurate volume or difficulty. You pay for more lookups, more keyword suggestions per lookup, history and exports. Semrush, for example, lists “Historical SEO data” from Pro+ ($299/month) upward, and Ubersuggest’s free plan locked its volume trend chart.
Is the cheapest keyword research tool the best deal? If you check a handful of keywords a day, Mangools Basic is the lowest list price of the three we can quote, on annual billing. If you run bulk lists all day, SE Ranking’s “unlimited” research on Core is the cleaner fit. The breakdowns of Semrush’s plan names and what each one costs, SE Ranking’s Core and Growth limits and Ubersuggest’s free, monthly and lifetime options go further than this table.
How to Use Two Keyword Tools Together
Two tools work better than one when they have different jobs: one tool to rank your keyword list, and a second to check the size of the few keywords you will actually write for. In our test, three of four tools agreed on size within 800 searches, so a second opinion catches the outlier. This is our judgment from one test, not a vendor rule.
A simple routine that fits free plans:
- Pick a home tool and stay with it. Build and rank your whole list there. Its difficulty scores only mean something next to each other.
- Check the shortlist in a second tool. Look only at size and direction. If the second tool lands within the same rough range, keep the keyword. If it is several times higher or lower, look at the results page before you commit.
- Fix the database before every lookup. Set the country and language first. Two of our four tools did not start in the US.
- Read difficulty against the live results page. Open Google, look at who ranks, and compare those sites with yours. Ubersuggest’s own guide says the same.
- Check your own data once you rank. After a page reaches the first page of results, compare your Search Console impressions with the tool’s volume. That tells you which tool runs high or low for your niche. Our walkthrough of checking keyword rankings and impressions in Search Console shows where those figures sit.
Which pairing makes sense? From our test, Semrush plus Mangools is a natural one on free plans: Semrush for the full report and country split, Mangools for a quick trend and an exact-phrase volume. If you are on SE Ranking’s trial, its conservative volume is a useful brake on Semrush’s higher figure. That is a judgment call, not a ranking.
Honestly, the biggest gain is not the second tool. It is the habit of reading every number as a range. When we put our four volumes next to each other, the honest summary was “a few thousand US searches a month, trending down,” and that sentence is more useful than any single figure on the screen.
References
- Ahrefs Help Center: How accurate is keyword search volume in Ahrefs?
- Semrush Knowledge Base: What is Search Volume in Semrush?
- Semrush Knowledge Base: What is KD%?
- Semrush Knowledge Base: Subscriptions and Toolkits
- Semrush: Volume Update! Why Semrush Search Volume Is the Most Accurate on the Market (Study)
- Semrush: SEO + AI Search pricing
- SE Ranking: Keyword Tool
- SE Ranking: Pricing and plans
- Mangools: Top 5 questions about KWFinder answered in detail
- Mangools: Keyword Difficulty Checker
- Mangools: Plans and Pricing
- Ubersuggest Help Center: How Does Ubersuggest Gather Data?
- Ubersuggest Help Center: How SEO Difficulty Works
- Ubersuggest: Pricing
- Google Ads Help: About Keyword Planner forecasts