Nobody has to ask people what they search for. That makes share of search cheap to read, and easy to over-read.
What does share of search mean?
Share of search means your brand's portion of all searches for brands in your category. The IPA reports Les Binet's definition: a brand's searches divided by the total for all brands in the category. It counts organic Google queries, not paid search.
The appeal is that nobody is prompted. Someone typed a brand name because it came to mind, which makes the metric a behavioural reading of interest. It says nothing about what they thought of the brand.
What did Les Binet actually find?
Binet found that share of search correlated with market share in three categories and tended to move first. He presented this at the IPA's EffWorks Global 2020 conference. The categories were automotive, gas and electricity, and mobile phone handsets.
The IPA reports that the lead can be substantial, up to a year for cars. When share of search rises, market share tends to rise, and the reverse. The gap between the two is called Extra Share of Search. The IPA calls it a particularly solid indicator of movements.
The IPA summary gives no correlation coefficient. It also covers three categories, so it is not a law for every market.
Binet himself calls the metric "by no means a silver bullet". He says the data "needs to be interpreted with care", because conversion depends on other factors, particularly price.
Not all search is positive either. A product recall can send people to a brand name too.
Later work has tried to widen this base. Treat that work as ongoing, and not as settled proof for your own category.
James Hankins writes that the IPA set up a Share of Search Think Tank. Its aim is an evidence base and benchmarks.
It set out seven hypotheses. They cover the link with market share, the response to advertising, and comparison with existing models.
How do you calculate share of search?
Divide the searches for your brand by the searches for all brands in your set, then multiply by 100. The arithmetic is simple. The quality depends on which brands and terms you include, and on using one consistent source.
- List the brands. Name those a buyer really chooses between, in one country.
- Choose the terms. Start with brand names, then add products people search for without the brand.
- Pull the data. Use the same place, period and source for every brand.
- Divide. Sum all brands, then divide each brand by that total.
- Smooth it. Read a rolling average, so one busy week does not steer the conclusion.
- Compare. Set the result beside market share and watch the gap over time.
The IPA notes that the data can come from Google Trends. It is free, generally goes back to 2004 and is available weekly. Google scales a sample of searches from 0 to 100, so compare brands inside one query only.
A 2025 industry analysis points out that Google Trends handles five terms at once. Swapping terms to fit more brands can cause inaccuracies.
One workaround is to keep the same reference brand in every batch, then rescale the other batches to it. Treat small brands with care, because rounding weighs more heavily on them.
Larger sets usually need a data source that reports across all terms at once.
Invented illustration. The numbers below are made up to show the arithmetic. They describe no real brand, category or result.
| Fictional brand | Average index | Share of search | Market share | Gap in points |
|---|---|---|---|---|
| Alder | 36 | 45% | 40% | +5 |
| Birch | 24 | 30% | 33% | -3 |
| Cedar | 12 | 15% | 16% | -1 |
| Dune | 8 | 10% | 11% | -1 |
| Total | 80 | 100% | 100% | 0 |
Alder's 36 divided by the total of 80 gives 45%. Its market share in this example is 40%, so the gap is 5 points. In Binet's framework, that gap is the number to watch. For a real brand, it would be a hypothesis to test across many periods.
What are the limits of share of search?
It measures interest in a name, not preference and not sales. That leaves several ways to misread it. Most become visible once you read the search terms themselves.
- Ambiguous names. That analysis uses Dodge as an example: the brand shares its name with a verb. Qualifiers remove noise but can also drop genuine brand searches.
- Products without the brand. The same analysis notes that people often search for a model such as the RAV4 without naming Toyota.
- Low-interest brands. Google says a 0 does not always mean nobody searched. Trends only shows data for popular terms, and noise shows most on low-interest queries.
- Negative attention. Google is explicit: Trends is not a scientific poll, and a spike does not mean something is popular.
- Price and conversion. Interest can rise while a higher price holds back sales.
- Other search surfaces. The IPA definition rests on Google. Interest on marketplaces, social platforms or AI assistants falls outside it.
Keep the term list under review too. A launch or renamed range can change what people type.
The same analysis reports that its set of about 30,000 keywords added roughly 36% more search interest on average. The comparison is with topline brand terms.
This is the vendor's own figure, and the vendor sells this kind of analysis. The direction is the useful part: a handful of terms sees only part of the interest.
How is share of search different from share of voice?
Share of voice measures what a brand puts in, and share of search measures part of what comes back. A media measurement explainer defines share of voice as a brand's percentage of all media spend in its category.
| Point | Share of voice | Share of search |
|---|---|---|
| What it counts | Media spend, or sometimes mentions and visibility | Searches for brand terms |
| Type of metric | Input the brand largely controls | Response from people in the market |
| Typical source | Media monitoring and spend data | Google Trends or other search data |
| Main question | Are we loud enough for our size? | Are we coming to mind more or less? |
| Blind spot | Says nothing about the response | Says nothing about the reason |
Check the definition before comparing numbers. One social media glossary uses a wider meaning, covering conversation, advertising and visibility, with social mentions as one version. A figure based on mentions cannot be set beside one based on media spend.
The IPA reports that share of voice has a large short-term effect that fades fast. A smaller long-term effect accumulates. In this analysis, 60% of searches came from long-term effects and 40% from short-term effects.
Each brand also has an equilibrium share of voice. Below it, share of search tends to fall over the next two years. Above it, share of search tends to rise.
The same explainer cites the older Binet and Field finding: across 171 campaigns, market share typically gained 0.5% for every 10% of extra share of voice.
Share of voice is what you put in. Share of search is part of what comes back.
How should you use share of search?
Use it as an early question, not a verdict. A rising or falling line tells you where to look. It does not tell you what happened, which is why it works best beside sales data and research.
First test it on your own history. Take several years of both series and shift one by a few months. See where the fit is closest. Our guide on how to measure market demand describes that check.
Trendata publishes this guide, and you can run everything above in Google Trends. Keyword and SEO tools show how often terms are searched, for content and findability. Trendata does a different job.
Trendata translates demand into market, customer needs and brand position, for decisions beyond SEO. It measures aggregated orientation moments: moments when people look into products, needs and brands, explore alternatives and prepare choices.
Its demand forecasts run up to 12 months ahead, at category, type and specification level. That scope does not automatically cover brand position, so read share of search as a current indicator.
Then agree in advance what a movement means. Write down how large a shift must be, how long it must last, and who looks into it.
Pair it with questions a search bar cannot answer. Our guide to brand tracking sets search beside surveys, and consumer insights covers the why behind the movement.
Frequently asked questions
How do you calculate the share of search?
Divide searches for your brand by searches for all brands in your category set, then multiply by 100. Use the same source, place and period for every brand.
What is a good share of search?
No general benchmark applies. Compare it with your own market share and with its direction over time. A figure above market share is the more interesting signal.
Does share of search predict market share?
It did in the categories Binet tested, with leads of up to a year for cars. The IPA stresses it is not a perfect predictor. Test it on your own history.
Can you measure share of search for free?
Yes. Google Trends is free and goes back to 2004. It compares five terms at a time and reports a relative index, not absolute counts.
What is Google's share of search?
That phrase usually means Google's portion of all searches across engines, which StatCounter publishes as search engine market share. It compares engines, while this metric compares brands.
Sources
- Binet presents fast, cheap, predictive Share of Search metric (IPA, 2020)
- In search of benchmarks for Share of Search (James Hankins, IPA, last updated 6 Mar 2025)
- Demystifying share of search (Achaibar and Kodu, Kantar, 11 Aug 2025)
- FAQ about Google Trends data (Google)
- Need to know: What is share of voice? (Nielsen, Mar 2025)
- Share of voice, social media glossary (Brandwatch)
- Search engine market share (StatCounter)