A market says what it wants every day. Share of search turns that into a number.

Share of search is the portion of all searches in a category that name one brand. If a category produces 100,000 brand-name searches a month and 20,000 of them name you, your share of search is 20 per cent. It is a measure of demand expressed by people, rather than of presence bought by a budget.

Take the brand names competing in the category, collect the search interest for each over the same period and the same market, add them to a total, and divide your own by that total. Google Trends gives relative interest rather than absolute counts, which is enough for a share, and the series is only comparable within one chart. The period matters more than the day: most people work with a twelve-week or thirteen-week moving average.

Share of voice counts money in the market. Share of search counts attention in the market. The first is an input you control, the second is an outcome you observe, and the second is the harder one to buy.

“Share of voice measures the percentage of media spending by a company compared to the total media expenditure”

Reference: Wikipedia, Share of voice. https://en.wikipedia.org/wiki/Share_of_voice

Excess share of voice is a brand's share of voice minus its share of market. Advertising more than your size predicts growth, advertising less predicts decline, and the effectiveness studies of Les Binet and Peter Field put the rule of thumb at roughly ten points of excess share of voice for about half a point of market share growth in a year. The figure is a planning heuristic and it varies by category.

“Many factors besides SOV contribute to increased market share”

Reference: Nielsen, Budgeting for the Upturn: Does Share of Voice Matter, 2009. https://www.nielsen.com/insights/2009/budgeting-for-the-upturn-does-share-of-voice-matter/

Because people look before they buy, and the looking shows up in query volume weeks or months before the purchase shows up in sales. The general mechanism is well documented outside marketing too: search query volume has been used to estimate the present state of things that official statistics report later, from unemployment claims to influenza rates.

“the original model significantly mispredicted influenza-like illness rates in the US during the 2012”

Study: Lampos, Miller, Crossan and Stefansen, Scientific Reports 5, 12760, 2015. https://www.nature.com/articles/srep12760

From Google Trends in most cases, sometimes from a paid tool that models absolute volumes. Trends reports a normalised index rather than counts, it samples, it rounds small values to zero, and it covers one search engine. A figure built on it is a ratio between brands measured the same way, and it holds as long as every brand in the set is measured in the same chart, market and period.

The ones that name the brand: the name alone, the name with a product, the name misspelled, the name with a complaint attached. Generic category searches belong in a separate series, because they measure the size of the category rather than any brand's place in it. Mixing the two produces a number that rises when the category grows and says nothing about you.

A crisis. A recall, a lawsuit, a viral complaint or a public dispute sends brand searches up sharply, and a share of search calculated that week reads like growth. The searches are real, the interest is real, and the intent behind them is the opposite of demand. Any spike wants a look at what people searched alongside the name before it is reported as a result.

Mental availability is the likelihood that a brand comes to mind in a buying situation, and a search is one of the few moments where that coming-to-mind becomes visible. A brand that is searched by name in its category has been recalled, unprompted, by a person with a reason. That is what makes the measure interesting, and it is also why a pure volume reading misses the point: the question is which situations bring the name to mind, not how often the name appears.

The measure counts queries, and queries are still made, so the series continues. What changes is what a query leads to and where it is typed. Questions that once produced ten visits now produce one answer and no visit, and a growing part of the asking happens inside assistants that report nothing to the public tools. Share of search will keep producing a number while the behaviour underneath it moves, which is an argument for reading it next to something that counts people rather than queries.

The difference between a brand's pull and its size. Share of voice has a companion figure for that, excess share of voice, and share of search has none in common use. The same subtraction is available, share of search minus share of market, and a brand whose searches run ahead of its sales is in a different position from one whose searches trail them. Nothing is stopping a marketing team from calculating it.

Track it weekly against the same competitor set, report it as a trend rather than a level, put it beside share of voice to see whether spending and attention are moving together, and treat every sharp move as a question rather than an answer. Used that way it is an early warning system that costs nothing but the time to build the series.