Branded search: why demand for your name is an off-page signal

Branded search is the set of searches that contain a brand’s name. Google’s own definition, from the Search Central post that introduced the branded queries filter in November 2025, is a query that includes the brand name, “variations or misspellings of the brand name”, or “brand-related products or services”. A search for a company’s name, for its name plus “pricing”, or for a product only that company makes all count. The volume of those searches is often called brand demand.

SEOs treat brand demand as an off-page signal for a simple reason. People search for a name because they met it somewhere else, in an article, on a podcast, at an event or from a friend, so the number of searches is a rough record of how much attention the brand has earned away from its own site. The evidence that Google uses that record directly is thin. The evidence that Google uses what happens next, the clicks, is strong. This lesson keeps those two apart, then covers how to measure brand demand and what raises it.

Why brand searches are treated as a signal

Start with what Google has published. Its public explanation of how results are ranked, on the How Search Works site, says Google uses “aggregated and anonymized interaction data to assess whether search results are relevant to queries”, and that this data is turned into signals that help its systems estimate relevance. That is a general admission that the behaviour of searchers feeds ranking. It does not mention brands. The Search Console post goes the other way and is careful to say the branded filter is a reporting feature that “has no effect on how Google Search ranking works”. The same post does state that branded queries “typically lead to higher-ranking pages from your site and result in higher click-through rates”, which describes how such searches behave without saying what Google does with them.

The nearest thing to a design document is a patent. As the lesson on brand mentions and citations sets out, Google’s patent US 8,682,892, granted in March 2014, describes a calculation that uses a count of independent links together with a count of “reference queries”, which are searches that refer to a site. In effect those are branded searches, counted beside links as evidence of standing. A patent shows what Google thought of and not what it runs, and Google has not confirmed that anything like it is live. Its value is that it shows the idea of weighing brand searches against links came from inside Google and was not invented by marketers.

What the trial and the leak are reported to show

The firmest evidence concerns clicks. In the US antitrust case, Google’s Pandu Nayak testified on 18 October 2023 that a system called Navboost uses 13 months of click data and is one of the important signals. A court exhibit reported in May 2025 says clicks are one of three inputs to Google’s topicality score, with anchors and body text, and that Chrome data is used in a popularity signal. The site’s page on the leak and the antitrust documents grades these as confirmed, because they were said under oath or filed as exhibits.

The step from there to brand demand is an inference, and it helps to lay it out. A known brand gets more clicks than an unknown one at the same position, because people choose names they recognise. A brand that is searched for by name produces searches where almost every click goes to one site. A system that learns from clicks would read both as signs that the site satisfies searchers. So activity that makes a brand better known could lift its rankings through clicks, with no link involved. Each link in that chain is plausible, and the first two can be seen in any site’s own data, but no document shows the last step, and nobody outside Google knows how clicks on branded queries are weighed against clicks on others.

The documentation leaked in May 2024 was read by some analysts as support for a stronger claim. Rand Fishkin, who was among the first to publish it, on the SparkToro blog that month, wrote that “brand matters more than anything else” and that Google rewards established brands with forms of popularity it can measure in search. He also reported that the documents describe click measures that separate satisfied from unsatisfied clicks, and the use of Chrome data. That is one analyst’s reading of documentation which contains no weights, and the leak cannot show which of its measures are in use. The practical conclusion is modest: brand demand is very likely useful, the mechanism is clicks more than search counts, and any claim about its size is a guess.

For AI answers there is a number, with the usual caution attached. In Ahrefs’ December 2025 study of 75,000 brands, branded search volume correlated 0.35 to 0.47 with brand visibility across ChatGPT, AI Mode and AI Overviews, below branded web mentions at 0.66 to 0.71. Large brands have more of everything, so the figure shows that searched-for brands are also the ones assistants name, and does not show that one causes the other.

How to measure brand demand

Three sources are in common use, and they measure different things, so it helps to compare them on the same points.

Source	What it shows	Unit	Main limit
Search Console, branded filter	Impressions and clicks on your own site from queries Google classes as branded	Actual counts for your property	Only your site, and only searches where you appeared
Google Trends	Interest in a term or topic over time, and against other terms	Relative scale of 0 to 100	No absolute numbers, and low volumes show as zero
Keyword tools	An estimated monthly search volume for the brand name	Estimated searches per month	Modelled figures that lag and differ between vendors

Search Console is the best record of your own brand. According to the Search Central post, the filter sits in the Performance report and splits queries into branded and non-branded across web, image, video and news search. Classification is done by “an internal, AI-assisted system” and not by a keyword rule, it covers the brand name in all languages and typos, and Google warns that some queries “may occasionally be misidentified”. The filter became available to all eligible sites on 11 March 2026. It is limited to top-level properties and to sites with enough query volume, so a small or new site may not see it. In that case a query filter built by hand on the brand name and its common misspellings does the same job less neatly. Branded impressions are the figure to chart month by month, because they count searches where you appeared whether or not anyone clicked. Measuring digital PR uses this same series as its proxy for attention that never produced a visit.

Google Trends answers a different question, which is how interest in your name compares with competitors and with its own past. Google’s help pages explain that each data point is divided by the total searches for the place and time it covers, and the results are then scaled from 0 to 100. A flat line in Trends can therefore hide growth if search as a whole grew at the same rate. Trends uses a sample of searches and shows zero where few people searched, which makes it unreliable for small brands. It also separates search terms, which match the words typed, from topics, which group terms that share a concept in any language. For a brand whose name is also a common word, the topic is the cleaner measure where one exists. The help pages say several groups of terms can be compared at once, which is the feature that makes Trends useful for seeing a competitor’s campaign land.

Which off-page work raises it

Nothing in this section is measured for any single tactic. The reasoning is that a branded search needs two things, a person who has met the name and a reason to look it up, and off-page work supplies both. Press coverage is the clearest case. Digital PR puts the name into articles read by people who have never visited the site, and many national newsrooms do not link out, so a share of interested readers will search the name instead. That makes branded impressions a fair way to credit coverage that produced mentions without links. The rise tends to be short. In BuzzStream’s State of Digital PR 2026 survey, 81% of coverage landed within one week of the pitch, so the attention from one campaign arrives together and then fades.

Consider a company that makes accounting software for freelancers and sponsors a regional design conference. The sponsorship yields one link from the event site, which is a paid link in Google’s terms and should carry the sponsored attribute. Its larger effect is that several hundred people in the target market see the name on a lanyard and a slide, and some of them type it into a search bar that week. The link report records one referring domain. The Search Console series records the part of the result that the link report misses, and comparing branded impressions in the weeks before and after the event, against the same weeks in a region where nothing happened, gives a rough reading of it.

Audio and video work the same way with one difference. A listener cannot click a spoken name, so podcast guesting and work with video creators send people to search more reliably than a written article with a link would. Reviews act later in the journey. Someone who has heard of a brand often searches its name with the word “reviews” before buying, which is a branded search in itself, and what they find on third-party platforms decides whether they click through. Reviews as an off-page signal covers that side, and off-page E-E-A-T explains why Google’s guidance tells its raters to look at what independent sources say.

Faked demand and other limits

Because the idea that brand searches help rankings is widely believed, services have appeared that sell them. They use bots, residential proxy traffic or paid users to search for a brand name and click its result. The research behind this site lists the method among manipulation tactics with variable and short-lived results, and notes that poor-quality traffic can leave negative signals of its own. Google’s Trends help page says it filters out duplicate searches from the same person over a short period, so the crudest attempts do not even register in the public data. Simulated searches also do nothing for the business, since no real person has learned the name.

The honest limits apply to the real thing as well. Brand demand rises for reasons that have nothing to do with marketing, such as a news story about the company, a seasonal peak, a price change or an outage that sends customers looking for a status page. A rise in branded impressions after a campaign is therefore evidence and not proof, and it is stronger when the timing is tight, the rise shows in the region the campaign reached, and nothing else changed. Branded traffic also converts well for an obvious reason, because these visitors were already looking for the company, so it should be reported apart from non-branded growth and not used to flatter it.

A sensible use of the measure is as one line in a wider report. Track branded impressions monthly, mark the dates of campaigns, sponsorships and appearances on the chart, and check the pages that appear when someone searches the name, since that results page is what a newly interested person sees first. Set up brand mention monitoring alongside it, because mentions are the cause and searches the effect, and a campaign that produced neither has told you something useful about the idea behind it.