What is search volume?

The figure answers one very specific question: how many people type a query in a given period. It says nothing about who types it, with what intent, or whether that person will end up buying anything. A term with high demand can be irrelevant to a business, and another with minimal demand can carry half the revenue.

It is worth separating it from two other metrics it gets confused with daily. Search Console impressions count how many times a link to your site appeared in front of a user in the results, so they measure your presence and not total demand. Clicks count how many of those appearances ended in a visit. The same query generates a single search, but spreads impressions across every site that appears on the results page.

The difference in definition explains why the three figures never reconcile. Volume covers all users in a location, whereas impressions only record the cases where your page was in the game at all. Their counting also has its own rules: when one and the same result block links several times to your domain, Search Console records it as a single impression.

The original source for almost the whole industry is the Google Ads Keyword Planner. Its average monthly searches metric is not an exact count of last month, but the mean of the twelve preceding months for the location and network you selected. Google also states that these statistics are rounded, which is why volumes for several locations do not add up the way you would expect.

Two details completely change how the figure should be read. The first: the metric does not cover only the term you typed, but that term together with its close variants. Singular and plural, common misspellings and equivalent phrasings can share the same number, so comparing two variants against each other tells you nothing. The second: the level of detail depends on the account. Without campaigns with active spend, the tool returns very wide ranges instead of values. An analysis of sixty million keywords published in 2024 documented seven ranges for such accounts, from 0 to 10, from 10 to 100, from 100 to 1,000 and so on up to more than a million, against roughly sixty discrete values available once there is spend.

That also explains why two paid tools give different numbers for the same word. Each one draws on its own sources, on user panels or on the Google API, and applies its model to fill the gaps Google does not publish. The differences reflect different estimation methods, not that one of them is lying.

The decision that depends on this figure is one of allocation: where the writing and internal linking time goes when there are a hundred possible topics and budget for ten. Without an estimate of demand, that choice is made on instinct or on whatever the sales team is asking for.

It also serves to size expectations before publishing. If a query averages, for example, two hundred searches a month in your country, no position will bring you thousands of visits with it, however good the page is. Doing that arithmetic in advance prevents traffic targets that do not fit in the market.

And it orders the architecture of the site. When several terms describe the same thing with different popularity, the most searched one usually sets the title and URL of the main page, while the rest are distributed across sections and supporting content. It is the cheapest signal you have for deciding what deserves a page of its own and what is a paragraph inside another.

• Read the figure as an order of magnitude and not as a visit forecast: it serves to separate big topics from small ones, not to calculate traffic.
• Always query with the location and language of the real market, because the tool default settings rarely match the country you sell to.
• Look at the twelve-month curve before deciding, since a flat average can hide strong seasonality concentrated in a few weeks.
• Cross-check the estimate against Search Console for the terms where you already hold positions, because there you have your own measurement instead of a model.
• Compare terms always within the same tool and with the same settings, never mixing figures from different providers in one table.
• Record the date and origin of every figure in the working document, because the values change and without a date nobody can audit the decision months later.

• Presenting volume as forecast traffic in a proposal or a results report.
• Comparing singular, plural or spelling variants as if they were independent demands when they actually share one grouped figure.
• Discarding long-tail terms because they show low numbers or are flagged as lacking sufficient data.
• Adding up volumes from several locations or several variants and presenting the total as real market demand.
• Mixing volume and impressions in the same table without saying which is an estimate and which is a measurement.