Every marketing report we see has a rankings table, a traffic graph and a leads count, and almost none of them has a line for branded search.
That is odd, because the number of people typing your company name into Google each month is the most direct measure of whether anyone noticed your marketing.
It sits in Search Console already and costs nothing to pull. It moves when radio adverts run, when a trade magazine mentions you, when a LinkedIn post travels, and when ChatGPT recommends you to someone who then goes to Google to check you exist.
This article covers why branded query volume is the cleanest signal you have, the regex filter that splits it out, how to read the trend without fooling yourself, what a sensible branded share looks like by business type, and how to credit a rise to the campaign that caused it.
Why branded search is the cleanest signal that marketing is working
Branded search counts the people who decided to look for you by name, which is the one outcome every channel is trying to produce. A billboard, a podcast interview, a referral from a client and a mention in an AI answer all end in the same place: someone opens Google and types the company name.
Attribution tools cannot see most of that. GA4 sees the final click and a survey sees what the customer remembers. Branded search volume sees the intent at the moment it formed, before any landing page or ad gets involved.
The clearest recent example is AI referrals. A 12-month GA4 analysis by Visibility Labs across 94 ecommerce brands, reported by Search Engine Land in February 2026, found that many buyers get a recommendation from ChatGPT and then search the brand on Google before buying. Those sales land in the books as branded organic.
If you are not watching branded volume, the AI channel is invisible to you, and in AI search brand mentions can matter more than your own website.
The scale of branded search is also larger than most owners expect. Ahrefs looked at about 150 million US keywords in May 2025 and found that 45.7% of all Google searches are branded. Nearly half of Google is people looking for a company they already know.
There is an older strand of evidence too. Les Binet, then head of effectiveness at adam&eveDDB, presented work at the IPA's EffWorks 2020 showing that a brand's share of category search correlated with market share in cars, energy and mobile handsets, and that search moved first.
"What we found is that share of search can predict market share, sometimes up to a year ahead." Les Binet, WARC
That is big-brand research. The mechanism is the same for a solicitor in Milton Keynes or a kitchen fitter in Kettering: the number of people searching your name this quarter tells you something about next quarter's enquiries.
Google now has a branded filter, and why you still need the regex
Google added a branded queries filter to the Search Console Performance report in November 2025 and opened it to all eligible sites in March 2026. It splits queries into Branded and Non-branded, works across web, image, video and news, and adds a card to the Insights report showing the click split.
The classification is not a keyword list. Google says it is "determined by an internal, AI-assisted system" that covers the brand name in all languages, typos, and queries that name a product of yours without naming you. Google also says some queries "may occasionally be misidentified".
Three limits matter for a small business. The filter is only available on domain or top-level properties, so a property set up as example.co.uk/blog/ will not have it. It is not shown for sites "with a low number of impressions", and Google does not publish the threshold.
Its history is short too: Google's help page says the data starts from March 2025, so there is no 2024 to compare against.
A regex filter has none of those limits. It works on any property, at any volume, back to the full 16 months Search Console holds. You control what counts as branded, and you can see exactly why a query matched. Use Google's filter as a cross-check, and the regex as your working KPI.
The branded regex, and how to apply it
The filter is a single line in the Queries filter, set to Custom (regex). Search Console uses RE2 syntax, matches anywhere in the query, and is case-insensitive by default, so you do not need to list capitals.
For a company called Growthcart the filter would be:
growth ?cart|growthcart\.co\.uk|growthkart|grothcart
Write your own in four steps:
- Start with the brand name as one word and as two words, joined with
?so the space is optional:growth ?cart. - Add the domain with the dot escaped, because people search URLs:
growthcart\.co\.uk. - Add the misspellings you see in the Queries table (sort by impressions, scroll, and note anything that is obviously you). Three or four is usually enough.
- Add product or service names that only you use, if you have any. A named package, a trademarked tool, a founder's name if clients search for it.
Separate each alternative with |. Apply with "Matches regex" to see branded, then flip to "Doesn't match regex" for non-branded. The two views add up to the total, with one caveat below.
Avoid patterns that catch generic words. If your company is "Northampton Roofing", a filter on northampton roofing will match a non-branded search for roofers in the town. Either accept that the split is fuzzy for a descriptive name, or restrict it to the full name plus ltd and the domain, and treat the result as a floor.
The caveat is anonymised queries. Google omits some rare queries from the table to protect privacy. They count in the chart totals, but once a query filter is applied they are excluded. So branded plus non-branded will come to slightly less than the unfiltered total. The gap is normal and stable, and it does not affect the trend.
How to read the branded trend
Read branded impressions, not clicks, at monthly granularity over the full 16 months. Impressions measure how many times someone searched your name and your site appeared, which for a brand query is nearly every time. Clicks are distorted by sitelinks, by a Google Business Profile panel that answers the question without a click, and by your own brand ads.
Set the date range to 16 months, switch the chart to Monthly, apply the regex, and read the line. Then do the same for non-branded. Non-branded tells you whether SEO is bringing in strangers. Branded tells you whether anything you did, in any channel, made people remember you.
The chart below is illustrative. The numbers are invented to show the shape to look for, not data from any client.
xychart-beta
title "Illustrative: branded impressions by month, campaign starts in month 8"
x-axis ["M1","M2","M3","M4","M5","M6","M7","M8","M9","M10","M11","M12"]
y-axis "Branded impressions" 0 --> 1400
line [520, 540, 510, 560, 580, 550, 570, 760, 910, 980, 1040, 1020]
The feature to look for is a step change that holds. Months one to seven are the baseline, drifting between 510 and 580. Month eight jumps by a third and the following months stay up. That is the signature of a campaign that reached people and stuck.
A spike that falls back the next month is PR, a viral post or a news item. A slow steady rise with no step is organic brand growth from word of mouth and repeat clients.
Two things make the line lie. Seasonality, which is why you compare each month with the same month a year earlier. And a change to the regex, which is why you should fix the pattern, write it down, and not touch it for a year.
What a healthy branded share looks like by business type
There is no published benchmark table for branded share by business type, and anyone who gives you one to a decimal place is guessing. What exists are a few published points.
Ben Garry at Impression wrote in The Drum that one UK high street retailer took 62% of organic traffic from brand queries in 2023, and one multi-country company took over 90%. In Google Shopping, a 2019 study of 750 European accounts reported by Search Engine Land put the branded share of ad cost between 16% and 56% depending on the sector.
From those points and from the Search Console accounts we work in, the table below is the working guide we use. Treat the percentages as the middle of a range, not a pass mark.
| Business type | Branded share of organic impressions | What low means | What high means |
|---|---|---|---|
| Local trade or service (plumber, roofer, cleaner) | 10% to 25% | Nobody knows the name; work comes from directories and "near me" | Strong local reputation; growth now depends on non-branded reach |
| Professional services (solicitor, accountant, consultant) | 25% to 50% | Referrals are not converting into search; check the Google Business Profile | Reputation-led practice; protect the brand page and reviews |
| Ecommerce brand selling own products | 40% to 70% | Advertising is not building memory; paid traffic will not get cheaper | Repeat buyers and word of mouth; watch for complacency on category terms |
| Retailer selling other people's brands | 15% to 40% | Customers remember the product, not the shop | The shop is a destination; defend it with brand ads |
| B2B software or SaaS | 50% to 80% | Early stage, or marketing is reaching nobody who remembers | Category leader signals; most new demand still arrives by name |
Two readings of the table matter more than the numbers. First, a branded share that is high because non-branded is tiny is not a good sign; a new site with 20 branded searches and 10 non-branded has a 67% share and no business. Always read the share next to the absolute branded count.
Second, the direction over a year matters more than the level. A professional services firm going from 30% to 42% with both lines rising has had a good year, whatever the table says.
How to attribute a branded lift to a campaign
Attribute a lift by comparing branded impressions in the campaign window with a baseline, while checking that non-branded did not move the same way. If both lines jumped, something changed in Google or on your site. If only branded jumped, and it jumped when the campaign started, you have your answer.
The method in five steps:
- Fix the baseline before the campaign starts: the average monthly branded impressions over the previous three months, and the same months a year earlier.
- Note the campaign dates and the channel. Radio, print, sponsorship, a LinkedIn push, a PR placement, an AI visibility programme, whatever it is.
- During and after the campaign, read branded impressions monthly. Lift is the amount above the baseline, adjusted for whatever the year-on-year trend was already doing.
- Use the Countries and Devices tabs to narrow it. A Northampton radio campaign should show up in UK impressions and on mobile, and should not show up in Ireland.
- If the campaign ran for a fixed period, watch whether the new level holds for three months after it ends. If it holds, the campaign built memory; if it decays, it bought attention.
Two channels need an extra step. Social campaigns often show up first as searches for the founder's name or the post topic, so add those terms to the regex for the campaign window and remove them after.
AI mentions show up as branded searches with no referral, which is why Visibility Labs recommended a post-purchase survey question asking where the buyer first heard of you. One question on the enquiry form does the same job for a service business.
What you cannot do is split a single month's lift between two campaigns that ran at the same time. Search Console does not know, and neither does anyone else. Stagger campaigns by a few weeks where you can, and accept a shared credit where you cannot.
Putting it in the monthly report
Add one line to whatever report you already run: branded impressions this month, against the same month last year, with a one-sentence note on anything that could explain a change. That is the whole job. It takes ten minutes once the regex is saved, and it is the only number in the report that every channel feeds into.
If the line is flat after a year of marketing spend, the spend reached nobody who remembered it, and that is worth knowing before the next invoice. If it is rising and enquiries are flat, the problem has moved to the website or the sales process, which is a cheaper problem to fix.
A rising branded line is the point of marketing. Rankings, impressions and followers are the means. If the branded number is not in your report, ask whoever writes it why not. Branded search is one line every agency report should carry; our guide to what a UK SEO agency should cost and deliver covers the rest of the report. If you would like us to set up the split and read it with you as part of an SEO programme or an AI visibility programme, get in touch.