What share of search measures, and how it relates to market share

What the metric counts

Share of search is a brand's searches divided by the total searches for all brands in its competitor set. It uses organic Google queries, not paid search advertising, according to the IPA's account of Binet's presentation.

The arithmetic is simple. Take a category with three brands. If Brand A draws 500 searches in a month, Brand B 300 and Brand C 200, the total is 1,000. Brand A's share of search is 500 divided by 1,000, or 50%.

The competitor set matters. Because each figure is divided by the total for the set, leaving out a rival or adding one changes every brand's figure.

Why search is treated as a proxy for demand

Binet's argument rests on a gap in conventional tracking. He said the relationship between tracking metrics and actual purchase behaviour is often surprisingly weak, and he described share of search as a cheap way to measure what people actually do online.

Binet's test links the two measures: when share of search goes up, share of market tends to go up, and when it goes down, share of market falls.

What the IPA research found

Binet's 2020 work tested three categories: automotive, energy (gas and electricity) and mobile phone handsets. Search share showed a link to market share in all three.

Two further findings gave the metric its forecasting role. Rises and falls in search share tend to precede similar moves in market share. The gap between a brand's search share and its market share, which Binet called Extra Share of Search, was described as a solid indicator of market share movements. For cars, the lead time can be up to a year.

The IPA's follow-up think tank drew on 30 case studies across 12 categories in seven countries. On average, share of search appears to represent about 83% of a brand's share of market, according to the IPA. The two measures move together, and the exact ratio varies by category and country.

The think tank also reported that brands outspending their market share, and so raising their share of voice, tend to see higher search share. Broadcast and mass-reach media appeared to have a larger effect on search share than targeted approaches.

How advertising feeds search

Binet's findings describe two effects of advertising on search. Share of voice (advertising) has a large short-term effect on search share that fades quickly. It also has a smaller long-term effect that decays slowly.

Under sustained advertising, Binet says 60% of searches come from the long-term effects and 40% from the short-term effects, which he links to growth in share of search. The long-term effects accumulate over time.

Binet also reported that each brand has an equilibrium share of voice. Falling below it tends to lower search share over the following two years, and rising above it tends to raise search share.

The think tank reports that when brands stop communicating, search share seems to decline in a consistent pattern. That decline is tied to the purchase-frequency half-lives of the category. Strong brands appear able to recover, though not fully to prior levels.

It is by no means a silver bullet

How to calculate it with free tools

Binet's data can come from Google Trends. The data is free, generally goes back to 2004 and can offer weekly frequency, according to the IPA.

Google's documentation explains why care is needed. Trends divides each data point by the total searches for its location and time range, then rescales results from 0 to 100. It uses a sample of searches rather than the full dataset, and adds statistical noise to protect privacy. The noise is most visible for low-interest queries, which can show as 0.

James Hankins of Vizer Consulting, who presented the think tank's data, recommends a rolling average of 6 to 12 months to smooth the data.

Google also says Trends reflects search interest, not polling, and should be one data point among others.

Where the metric stops working

The think tank reminds readers that its findings refer to correlations, not causal relationships. The ratio between share of search and share of market varies by category and country, so a single benchmark such as 83% may not transfer cleanly.

Binet's own caveats are blunt. Share of search is not a perfect predictor, because conversion is affected by other factors, particularly price. He called it by no means a silver bullet, and noted that not all search activity is positive.

Binet said researchers need more detail before the metric can be applied to the real world, and James Hankins of Vizer Consulting called the work 'only the beginning.'

In practice, the metric gives a free read on a brand's searches against named rivals, and Google says Trends data should be one data point among others.