Search Volume

Vendors define the term in their own words. Semrush calls it the average number of times a specific search query is entered on a search engine per month, in the main database from national-level data, and works out the average monthly figure by dividing the total searches for the past 12 months by 12; it notes that for seasonal and trending keywords the number could be higher or lower than that average in certain months^[1]. Ahrefs describes how often a target keyword is searched per month in a given country, as an average for the last 12 months^[2].

None of them counts searches directly. Semrush says it takes sets of data from third-party providers, overlays them with historical clickstream data, builds a computational model from the result and applies it to get volumes for billions of keywords^[1]. Ahrefs says it calculates its figure by processing large amounts of clickstream data, and describes the accuracy as exceptional, which is the vendor’s own claim^[2]. The figures are model outputs, and two models give two numbers.

Google’s own tool for advertisers defines it differently again. Keyword Planner shows average monthly searches as the number of times people have searched for a keyword and its close variants, averaged over 12 months by default. Google notes that these statistics are rounded, so figures for several locations might not add up as expected, and that forecasts take seasonality into account while historical metrics, search volume among them, do not^[3].

Google also says where tools stop. Its guidance on third-party SEO tools says that some services provide data that people misinterpret as being from Google, that third-party tools don’t have access to its internal ranking data and can’t guarantee performance, and that Search Console provides key information and data directly from Google Search itself^[4]. Volume tells you about demand in general; impressions in Search Console tell you what happened to your own pages.

Ahrefs draws one more line worth keeping: the volume of a keyword differs from the traffic potential of its topic, which it estimates from the organic traffic of the page that currently ranks first for the topic’s parent keyword^[2]. A list of ranked sites is a different thing again. QueryRev’s study draws its panel from the first 10,000 places of Tranco list N26XW, which was built from rankings between 15 September and 14 October 2026 and ranks domains, not searches^[5].

Record every volume with its tool, country, date and period. A figure without them cannot be compared next quarter, and a figure for one country is not one for another.

Measure the spread. Look up the same keyword in two tools and keep both numbers: the gap between them is a rough size of the uncertainty. If one tool shows a range or a rounded figure, record it as given.

Read the twelve months, not the average. For a seasonal keyword a flat average hides the peak and the trough, and the trend line says more than the figure. Put it beside seasonality before you plan around a month.

Check against Search Console for keywords you already rank for. Impressions and clicks for the query show what your pages received, at the positions they held, and depend on those positions. They do not replace volume for keywords you do not rank for, and volume does not replace them for those you do.

Illustrative: an invented site, to show the method.

A bike shop looks up one keyword in four places. Tool A says 2,400 searches a month, Tool B 1,900. Google’s Keyword Planner shows a range of 1K–10K. Search Console shows 1,150 impressions last month for the same query, at an average position of 3.2.

None of the four is wrong in the way a wrong sum is. The two tools disagree because they model different data; the range is a rounded band; and the impressions count only the times the shop’s page appeared. The shop plans on the order of magnitude, about a thousand to two thousand searches a month, watches the trend line, and does not forecast revenue from a single figure.