Branded key words, the split that changes every report you read, and whether to bid on your own name

A branded keyword is any search that contains your company name, a product name you own, a misspelling of either, or a phrase only a person who already knows you would type. Everything else is non brand. That one division changes the meaning of almost every number in a marketing report, because branded searches come from demand you already created and non brand searches come from demand you are capturing for the first time. A report that mixes the two can show traffic rising while new customer discovery falls, and most reports mix the two by default.

So this page does three things: defines the boundary precisely enough to apply, shows how to split the reporting in two tools, and answers the bidding question, which is the one that costs real money.

What counts as branded, and the three edge cases that cause arguments

• Clearly branded. Your company name alone, your name plus a service, your name plus a location, your name plus "reviews" or "pricing" or "login", your product names, and every misspelling and spacing variant of all of those.
• Clearly non brand. A category or problem query with no name in it at all.
• Edge case one, a competitor's name. Not your branded traffic, and not generic either. It belongs in a third bucket called competitor, because its conversion behaviour sits between the two and averaging it into either one distorts both.
• Edge case two, a generic company name. If your company is called something like a common noun, a plain name match will pull in searches that have nothing to do with you. Build the brand filter from your name plus a qualifier and accept a small undercount rather than a large overcount.
• Edge case three, a product name that became the category. If your product name is used generically by people who do not know it is yours, those searches are branded in form and non brand in intent. Track them separately and do not let them inflate your brand demand figure.

Splitting the report in Search Console

This takes five minutes and it is the highest value five minutes available in most reporting setups.

1. Open the performance report and set a period of at least three months, so seasonality does not swamp the comparison.
2. Add a query filter containing your brand name. Record clicks, impressions and average position. That is your branded line.
3. Change the filter to exclude the same string. Record the same three figures. That is your non brand line.
4. Repeat step two for every misspelling and spacing variant you can think of, and add them to the branded total. The variants are usually five to ten per cent of brand volume and they are always missing from the first attempt.
5. Write the two lines side by side every month, as two rows rather than one total. From then on the report answers a question the combined figure never could.

What the split tells you, in practice. If brand clicks are rising and non brand clicks are flat, your other marketing is working and your search presence is not growing. If non brand is rising and brand is flat, you are being discovered but the discovery is not converting into people who come back by name. If both are flat and total traffic is up, something else is happening and you should find out what before you attribute it to anything.

Branded queries behave differently from non brand ones in one more way that is easy to misread: your average position on your own name will be close to one, which pulls the site average down towards it and makes the overall figure meaningless. Read position by bucket or do not read it at all, which is the same discipline as reading click through rate by source rather than as a single average.

Splitting the report in a paid search account

In an advertising account the split has to be structural rather than a filter, because budgets and bids are set at campaign level.

• One campaign for brand, separate from everything else, with its own budget. Mixed campaigns let cheap brand clicks subsidise the reported cost per acquisition of the whole account, which is the most common way a search account looks better than it is.
• Your brand terms as negatives on every non brand campaign. Without this, broad and phrase matching will serve brand queries from generic campaigns and the separation collapses.
• Report the two cost per acquisition figures separately, always. A blended figure that includes brand is not a measure of acquisition, it is a measure of how much existing demand you have.
• Watch the brand impression share for competitor pressure. A sudden fall usually means somebody has started bidding on your name, which is the trigger for the decision below.

Whether to bid on your own brand

The argument against is simple and often right: you already rank first for your own name, so you are paying for a click you would have received free. The argument for depends on four specific situations, and outside them the sceptics are correct.

• A competitor is bidding on your name. Then the top of the page is theirs and the first thing your prospect reads is their pitch. Bidding restores the position, and brand clicks are usually cheap because your own relevance is high.
• Resellers, affiliates or marketplaces outrank you for your own name. Common for consumer brands sold through third parties. An ad is the only way to be the first result that sells at your own margin.
• You need to control the message for a specific moment. A price change, a recall, a rebrand, a service incident. An ad lets you put the right sentence at the top of the page today rather than waiting for the organic result to catch up.
• Your organic result does not serve the query well. If somebody searching your name plus "cancel" or "pricing" lands on a homepage, an ad pointing at the right page is a better experience and a cheaper support ticket.

The test to run before you commit a budget to it is an incrementality test, not a performance report. Pause brand bidding in half your geographies for two to four weeks and compare total conversions, brand plus organic, between the two halves. If total conversions hold where the ads are off, the ads were buying clicks you already had. That experiment is cheap and almost nobody runs it, which is why this question stays unsettled in so many accounts.

Growing branded demand on purpose

Brand searches are an output of everything else you do, which makes them the most honest lagging indicator in marketing. Three things move them reliably. Consistent naming everywhere your company appears, so the searches people type are the words you chose rather than four variants. Being present in the places where people first hear the name, which is usually not search. And making the branded result itself complete, so your own name query returns the pages a buyer needs rather than one homepage, which is the subject of how to control what Google shows when somebody searches your name.

On the non brand side, the decision about which generic queries are worth chasing in what order is a separate discipline, and the sequencing method is in how to decide what to chase in what order and when to stop adding. The two lists should be built and reported as two lists, forever.

Questions about branded keywords

What is a branded keyword?

A search containing your company name, a product name you own, or a misspelling of either, plus any phrase only somebody who already knows you would type. Everything without a name in it is non brand.

Why separate brand and non brand in reporting?

Because they measure different things. Brand searches count demand you already created; non brand searches count new discovery. A combined total can rise while discovery falls, and nothing in a single figure will show you that.

Should I bid on my own brand name?

Yes in four situations: a competitor is bidding on it, resellers outrank you, you need message control for a specific event, or your organic result answers the query badly. Otherwise run the incrementality test described above and let your own data decide.

Is it legal for a competitor to bid on my brand name?

Bidding on a competitor's name as a keyword is generally permitted by the major ad platforms, while using a trademark in the ad text is treated differently and platforms have complaint processes for it. The practical first step is the platform's own trademark complaint route, and the legal question belongs with a lawyer rather than a marketing page.

How do I handle misspellings of my brand?

Add every variant to the brand filter in reporting, and in paid search let broad matching catch them rather than listing each one. Misspellings are usually a meaningful share of brand volume and they are almost always missing from the first version of a brand report.

What if my brand name is a common word?

Build the filter from your name plus a qualifier rather than from the name alone, and accept that you will undercount slightly. An undercount you understand is better than an overcount that makes your brand demand look like generic traffic.

The branded search that never gets typed

A brand report counts people who typed your name. There is a growing step before that one where nobody types anything: a buyer asks an assistant who to use for a thing, and the answer names two or three companies. If you are in that answer, the buyer may then search your name, and the branded search you record is the last step of a process you had no visibility into. If you are not in it, the branded search never happens at all and your report shows nothing, because an absence leaves no row. That is the practical limit of the brand and non brand split: it measures demand that reached a search box, and it cannot see the shortlist being formed one step earlier. AI Knows Us asks those buying questions across the main assistants and records which companies get named and which sources each answer was built from, which is the reading that sits upstream of the brand line in your report.