Open any keyword tool and every term carries a search volume number. 2,400 searches a month. The precision is an artefact of the interface, not of the data — and plans get built on it as though someone counted.
Where does the number come from?
From modelling, not counting. Tools build estimates from clickstream panels, third-party data and Google's own advertising ranges, then smooth the result into a single figure. Google's keyword planner has historically reported in broad buckets rather than exact counts, and a tool showing you 2,400 has usually turned a bucket into a point estimate for display.
Several things are folded together in that one number. Close variants are frequently grouped, so the figure may cover a family of phrasings rather than the exact string you typed. And the number is a monthly average, which erases seasonality completely. That is worth knowing before keyword research turns into a production schedule.
Volume estimates are directionally useful and precisely wrong. Ranking two terms against each other is reasonable. Forecasting traffic from the absolute figure is not.
How wrong, in practice?
Wrong enough that two reputable tools routinely disagree by a multiple on the same term, which is the easiest test to run yourself. Check five terms across two tools and the spread will tell you more about the reliability of both than any methodology page will.
Error is not evenly distributed either. High-volume head terms are estimated relatively well, because the panel data is dense. Long-tail terms are where estimates fall apart — a phrase reported at zero can be genuinely searched, and Google itself has said a substantial share of daily queries have never been seen before. Those queries cannot appear in anyone's database, and collectively they are a large slice of the total.
Discarding a term because a tool reports zero volume. Zero frequently means "below the reporting threshold" or "not in the panel", and the specific, high-intent questions your customers ask are exactly the ones that land there.
What can the number safely decide?
Relative priority within a set, and rough order of magnitude. Those are real decisions and volume estimates support them adequately.
| Decision | Volume estimates are |
|---|---|
| Which of two similar topics to write first | Adequate |
| Whether a topic has meaningful demand at all | Adequate as a rough signal |
| How much traffic a new page will earn | Not reliable |
| Whether a specific long-tail question is worth answering | Actively misleading |
| Building a revenue forecast for a client | Not defensible |
What is better than an estimate?
Your own impressions data, which is a count rather than a model. Search Console reports the queries where your pages actually appeared, on your site, for your audience. It is limited to terms you already rank for somewhere, and within that limit it beats every third-party estimate available.
Three ways to use it:
- Find queries with impressions and poor positions. Demand is proven and you are already almost visible.
- Find queries you rank for accidentally on a page not written for them. That is a page waiting to be written.
- Sanity-check a tool's estimate against your own impressions for the same term, adjusted for your position. When they disagree wildly, believe your own data.
How should this change a content plan?
Stop sorting the spreadsheet by volume and calling the top rows the plan. A term with a tenth of the volume and a clear commercial motive behind it is usually worth more than a broad informational term with a big number attached.
The practical substitution: plan around questions your customers actually ask — from sales calls, support tickets and your own impressions data — and use volume estimates only to break ties. That reverses the usual order, and it is the order that survives contact with the results.
Pull the last twenty questions your team was asked by a prospect. Check how many appear anywhere in your keyword tool. The gap between those two lists is where the volume-first approach has been costing you.
Building a content plan on volume estimates?
We will validate the terms that matter against your own impression data before you commit a quarter of production to them.
Get in Touch →Does this apply to your site?
Reading about it is one thing. Point the scan at your own site and see whether this applies to you, and what it is worth fixing.
Free and unlimited. No account, no card, and you get every finding rather than a teaser.