In short
No, not as a count. Keyword tools derive volumes from Google Ads data, which is rounded into bands and averaged over twelve months, then extend it with clickstream samples and models; comparisons against real Search Console impressions find tool estimates frequently off by 50% or more in either direction, with long-tail and new queries worst. Volumes are useful for ranking keywords relative to each other and for ruling out terms nobody searches. For any term you already rank near, your own Search Console impressions are the accurate number.
Where the numbers come from
Almost every keyword tool's volume figure traces back to Google Keyword Planner, which exists to sell ads. Keyword Planner reports volumes rounded into broad bands, averaged over the previous twelve months, and, for accounts without significant spend, as ranges such as 1K–10K. It also groups close variants together, so "web design agency" and "web design agencies" may share one figure.
Third-party tools take that data and refine it with clickstream panels, samples of real users' browsing bought from browser extensions and apps, and with statistical models that spread the rounded figures across variants and extrapolate to queries Google never reported. The output looks precise, 1,300 searches a month, and is a model of a rounding of an average.
Six reasons the number misleads
Rounding and banding. The source is coarse; the precision is added afterwards.
Twelve-month averaging. A seasonal term shows the same volume in its peak and its trough. A term that doubled last quarter shows half its current volume. A term that died shows its ghost for a year.
Clickstream bias. Panels over-represent certain countries, browsers and demographics, and under-represent mobile and private browsing.
Grouping. Variants merged in the source and split by the model produce figures that may or may not reflect how people actually phrase the query.
Long tail and new terms. Anything with low volume is below the resolution of the source data and is largely modelled. New terms have no history and are guessed.
AI and zero-click. Volume counts searches, not clicks. On queries answered by an AI Overview, the searches happen and the visits do not; the zero-click share varies enormously by query and no volume figure reflects it.
What the comparison studies found
Several practitioners have compared tool volumes against Search Console impressions for pages ranking in position one, where impressions approximate real search volume. The consistent finding is that tools are often wrong by a factor of two, sometimes far more, with no consistent direction: some terms are heavily overstated, others understated. Tools agree with each other more than they agree with reality, because they share a source.
The practical conclusion is not that the numbers are useless. It is that they are ordinal, not cardinal. A term showing 5,000 is very probably searched more than one showing 50. Whether it is searched 3,000 or 9,000 times is unknowable from the tool.
What volumes are good for
Ranking terms against each other. Which of these twenty phrasings do people use most? The tool answers that reasonably well.
Ruling out ghosts. A term with no measurable volume in any tool is probably not searched enough to build a page around, though the long tail is real and a specific question can be worth answering at low volume.
Sizing a topic. Add up the cluster and you have a rough sense of whether a subject is a niche or a market, within a factor of two.
Spotting seasonality, if the tool shows monthly trend rather than an average.
What they are not good for is forecasting traffic. "This term has 2,000 searches, position one gets 30%, so 600 visits" compounds three estimates, each of which can be off by half, into a number that will be wrong.
The data that is accurate
Search Console impressions. For any query where your site appears in the results, Search Console reports how many times it was shown. For a page in the top positions, impressions are close to actual search volume for that query in your markets. This is real data about real searches, free, and it is the number to use for any term you already rank near. The Performance report's query list is also the best source of new terms, because it shows what people actually typed.
Google Trends. Relative, not absolute, but real and current, and the best way to see whether a term is rising, falling or seasonal.
Autocomplete and People Also Ask. Not volumes, but evidence that Google sees the query often enough to suggest it, which for long-tail research is the signal that matters.
Your own site search and support inbox. The questions people already ask you, which are the questions people search.
How to use all of it
Use a tool to generate and rank candidate terms and to size topics. Use Search Console to see what you actually receive impressions for and to measure any term you rank for. Use Trends for direction. Then decide what to write by intent and fit, not by the volume column, because the volume column is the least reliable number on the screen. If you would like help building a keyword plan that treats the estimates with the right amount of respect, book a call.
Common questions
How accurate is Google Keyword Planner?
It reports twelve-month averages rounded into bands, with close variants grouped, and shows only ranges for accounts without ad spend. It is the source most other tools build on. Treat its figures as rough relative indicators; Search Console impressions for terms you rank for are far more accurate.
Why do different keyword tools show different search volumes?
Because each combines the same rounded Google Ads data with its own clickstream sample and its own model for splitting and extrapolating. The models differ, so the outputs differ, often by a factor of two. Agreement between tools does not mean accuracy, since they share the underlying source.
Can I get real search volume for a keyword?
For queries your site already appears for, Search Console impressions are real counts of how often your result was shown, and for top positions they approximate true search volume in your markets. For queries you do not rank for, no source provides real counts; Trends gives relative direction and tools give modelled estimates.
Should I target keywords with zero search volume?
Sometimes. Tools cannot resolve low-volume queries and report many real ones as zero. If a specific question appears in autocomplete, People Also Ask, or your own customers' emails, it is being asked, and a page answering it can rank quickly and convert well precisely because nobody else bothered.
How do I estimate traffic from a keyword?
Cautiously. Multiplying tool volume by an assumed click-through rate compounds two rough estimates and ignores zero-click behaviour. A better approach is to rank for a few terms in a cluster, read the actual impressions and clicks in Search Console, and extrapolate from your own data rather than the tool.
