Bidding on your own brand: a calculation, not a matter of faith

The question has a threshold, and you can calculate it

In my view, whether you bid on your own brand is a calculation. A brand campaign costs money and reports revenue, part of which would have arrived without the ad. It covers its costs once the genuinely incremental part is large enough:

Required incremental share = 1 / (brand ROAS × contribution margin rate)

Suppose a shop spends €2,000 on brand, the account reports €40,000 in revenue (ROAS 20) and the margin is 30%. Then 1 / (20 × 0.3), so around 17%, is needed. Around €6,700 of the reported €40,000 has to be genuinely incremental. The margin here is the same as for the break-even ROAS, so it includes the costs that are often left out.

Turned around, the formula shows which reported ROAS is just enough at a given incremental share:

A brand ROAS of 20 looks brilliant in the report. At a 20% margin it means a quarter of the revenue has to be incremental just for the campaign to break even.

Doesn't the shop get the click organically anyway?

Often it does, at least for strong brands. When eBay stopped brand ads on Yahoo and MSN, organic clicks made up for 99.5% of the lost clicks. The Bing experiments by Simonov, Nosko and Rao cover more ground, with 824 companies: without competitors, a brand's own ad brings 2.27 additional clicks per 100 searches, out of 36.4 paid clicks. So you pay for around 16 clicks to get one additional click, which is around 6% of paid clicks.

With that 6%, at a 30% margin you would need a brand ROAS of just over 50. That is only a thought experiment, because the study measures clicks, not revenue. But it shows why a ROAS of 20 or 30 proves nothing by itself.

For your shop these figures only provide a hypothesis. The answer comes from your own test. For less well-known brands the effect is considerably larger, though in this study still fairly moderate. According to Google, on average 50% of ad clicks are incremental when you rank first organically, measured across all search ads rather than brand only. I know of no robust public study for small online shops in the DACH region (Germany, Austria and Switzerland).

What changes when competitors bid on your brand

With a competitor in first position, a pause costs you real clicks. In the Bing data, a high-traffic brand without its own ad loses on average 18% of its clicks to a single competitor at the top, and 12% across all companies. If the brand's own ad sits above, the loss is only 1 to 5%. The comparison is not fully randomised, but I still consider the direction reliable.

It is not a given that competitors move in after a pause: at eBay the feared bidding did not happen, although eBay is an edge case.

Legal or policy action rarely replaces the calculation. Google does not restrict trademarks as keywords; following a complaint, their use in ad text can be limited. Legally, too, booking someone else's trademark is permissible in principle according to a law firm's overview (CJEU, Interflora), for example if the ad does not name the trademark, does not suggest a link to its owner and the domain points to a different origin (German Federal Court of Justice, Bananabay II). This is not legal advice; in a dispute the individual case decides. The other side of the question is covered under bidding on competitor brands.

My rule from this: if nobody regularly appears on your brand searches, brand is a candidate for a test pause. If someone regularly sits at the top, brand keeps running, with a cost cap. You can read this from the auction insights of the brand campaign.

How to test this without fooling yourself

1. Set the threshold in advance. Calculate the required share and write it down before any results come in.
2. Take brand out of the other campaigns. Brand exclusions exist for Performance Max and Search; in Search, depending on the account, they require AI Max, which you should not switch on in passing for the test. The option to allow Shopping ads on excluded brands has to be off in PMax, otherwise Shopping catches the searches and you measure a shift. Details are under brand in Performance Max. Keep Standard Shopping away from brand searches with a negative keyword list and check this in the search terms report. All of this happens before the comparison period, because the separation itself changes the numbers.
3. Split regions at random. In the test regions brand is paused, in the control regions everything keeps running. The study authors recommend exactly this kind of pause or geo test. Conversion Lift based on geography supports Search, Shopping and PMax, but is not open to every account, runs through your Google account representative and requires campaigns that target only one country. Without access, your own geo test does the job.
4. Measure the shop's total revenue. Take revenue per region from the shop system. The platform figure is useless, because in the test area it inevitably falls to zero.
5. Long enough and without promotions. I would run at least two full purchase cycles and avoid any discount period.

The evaluation stays simple. Suppose the test regions go from €100,000 to €99,000 and the control regions from €200,000 to €204,000, so up 2%. You would have expected €102,000 in the test, and €3,000 is missing. Reported brand revenue there would have been €20,000, at a cost of €1,000. So 15% was incremental, below the 17% threshold. €900 of contribution margin is lost and €1,000 of ad spend is saved, so the pause only just pays off.

Before you start, check whether the test can answer the question at all. If brand accounts for 20% of shop revenue and the threshold is 17%, everything hinges on a difference of 3.4% of shop revenue. If the ratio of test to control regions fluctuates more than that in your historical data, you are measuring noise. A before and after comparison without a control group is the weaker option: at eBay, Google traffic fell by three per cent on that basis, and according to the authors it would probably have been less with a clean control group. If the measurement stays inconclusive, decide based on the competitive situation. What is at stake then is the brand spend, and with a high ROAS that is small relative to shop revenue.

Which bid strategy suits your brand

I consider target ROAS the wrong lever for brand, because it optimises for reported revenue, and the incrementality of exactly that revenue is what is in dispute here. According to Google, Target impression share can make sense for brand terms, as long as the maximum CPC bid limit is not set too low. That fits the job, provided you derive the limit from the economics:

Max. CPC = revenue per click × incremental share × margin

Suppose a click brings €5 in revenue, the test shows 15% and the margin is 30%. That gives around €0.23. If instead the test shows 50% in a situation with a competitor in first position, it is €0.75. If the strategy then misses its target because the cap kicks in, you know that first position does not pay for you at that price.

Taking brand out of your ROAS reporting

Whatever the test shows, brand belongs outside the figure you steer by:

Non-brand ROAS = (total revenue minus brand revenue) / (total cost minus brand cost)

A worked example at a 30% margin, so a break-even ROAS of 3.3:

At ROAS 5 the account is well above break-even, while the part you can actually steer is well below it. If only 10% of brand is incremental, the €2,000 really brings €1,800 of contribution margin. The reported €10,000 profit turns into a €6,200 loss. That is why brand revenue is not automatically Google Ads performance.

Even the adjusted figure is not exact. Experimentally measured returns from paid search come to only a fraction of the usual estimates, and that applies to non-brand too. Conversely, some brand searches only happen because someone found you earlier through a generic ad, so non-brand can also look too bad. I would therefore use the figure as a comparison with the break-even ROAS, not as a measurement. And it is only correct if PMax and Shopping no longer contain any brand traffic.

Without a safety net, brand keeps running

Low incremental values are only plausible if an organic result catches the click. If marketplaces, other retailers or comparison sites rank above you for your own name, that safety net is missing, and I would keep brand running without a test. With a young or little-known brand I would not start with a pause either, because the gain in clicks is larger there.

A test over a few purchase cycles also only measures the short term. It does not show whether a pause changes brand perception over the years, and its result only applies to the auction in which it ran. My calculation holds for shops with their own brand and meaningful brand search volume. If hardly anyone searches for your name, the amount is usually too small to justify a test.