ServiceMethodTeamFAQThe BlogLet’s talk
← All articles

Paid search4 November 2026

Should you bid on your own brand name? Three experiments, three answers

Paying for a click you would have had for free: the question irritates every advertiser. Controlled experiments have settled it, and the answer is not the one usually repeated.

By Léo Baumgarten
EU Market Lead, Affiliate

Someone types your brand name. They already know who you are; they are looking for you. At the top of the page sits your paid ad. Just below it, your free organic result. If they click the ad, you pay for a visitor who was coming anyway. The question is as old as search, and it is settled by experiment rather than intuition.

The eBay experiment, 2015

The first answer comes from eBay. In 2015, three researchers (Thomas Blake, Chris Nosko and Steven Tadelis) published in Econometrica the results of large-scale controlled experiments run on eBay’s own account: brand-keyword advertising was switched off across part of the United States and traffic compared with control areas. The result is blunt. Traffic stayed virtually unchanged. Visitors deprived of the ad simply clicked the organic link just below. Substitution approached 100%.

That study has been read everywhere as a proof: bidding on your own brand is pointless. That goes too fast, and a second experiment shows why.

The Edmunds experiment, 2018

In 2018, Lorenzo Coviello, Uri Gneezy and Lorenz Goette published a comparable protocol, this time on Edmunds.com, a US automotive information site. Same method: markets randomly assigned, brand advertising switched off in the treatment group. Same question, opposite result. Only about half the traffic that used to come through brand ads still arrived via the organic link. The other half vanished.

Why two rigorous experiments diverge

Why do two rigorous experiments reach opposite conclusions? Because they do not measure the same brand. eBay is a destination: when someone types "eBay", the organic result owns the page, it is impossible to miss, and the ad merely duplicates an already dominant link. A smaller brand has no such pull. Its organic link exists, but it does not hold the page the same way.

The decisive variable, your competitors

That leaves the variable neither study manipulated, and which turns out to be decisive: your competitors. Also in 2018, Andrey Simonov, Chris Nosko and Justin Rao published in Marketing Science a series of experiments run on Bing across thousands of brands. Their figures are precise and deserve to be read in the right order.

When no competitor bids on your name, your own brand ad adds only 1% to 4% more clicks: and the effect is smaller the larger the brand. In that case the cost of a genuinely incremental click becomes very high. When competitors hold paid positions 2 to 4 and your ad is present, they still take 1% to 5% of your clicks. And when competitors bid on your name while you have chosen not to defend it, they capture 18% to 42% of the clicks.

It is that last figure which changes the decision. The authors conclude that the return on investment of defensive advertising is strongly positive. In other words: the useful question is not "does brand search work", but "is anyone bidding on my name". In the first case you pay dearly for a few percent; in the second you are protecting between a fifth and two fifths of your own clicks.

Scroll the table

ExperimentFieldPublicationTraffic retained without brand ads
Blake, Nosko, TadeliseBay, United StatesEconometrica, 2015Close to 100%
Coviello, Gneezy, GoetteEdmunds.com, United States2018About 50%
Simonov, Nosko, RaoBing, thousands of brandsMarketing Science, 2018Depends on which competitors are present

The test that takes thirty seconds

That question takes thirty seconds to answer: type your brand name in a private window and look at who appears above you. If nobody is contesting the spot, the experiments suggest your budget is better spent elsewhere. If a competitor is there, the arithmetic reverses.

One caution, because it matters: working versions of that last study, circulated before publication, reported different orders of magnitude, and several press write-ups still quote those. The figures given here are those of the article published in Marketing Science in 2018. It is the only peer-reviewed version.

The same question, in affiliate

Finally, this reasoning is not confined to search. In affiliate, a cashback site or a coupon that slots in at the payment step occupies exactly the position of a competitor bidding on your brand name: it places itself between an already-decided buyer and your site. The question is identical, and so is the method for answering it: a controlled switch-off, watching total revenue rather than attributed sales.

Sources

  1. Blake, Nosko & Tadelis, Consumer Heterogeneity and Paid Search EffectivenessExperiments run on eBay, published in Econometrica vol. 83 (2015), pp. 155-174. NBER working paper no. 20171, May 2014.
  2. Coviello, Gneezy & Goette, An experiment to estimate the effectiveness of branded search adsThe authors’ own write-up of their Edmunds.com experiment, CEPR/VoxEU, 18 February 2018. Full paper: CEPR Discussion Paper 12333.
  3. Simonov, Nosko & Rao, Competition and Crowd-Out for Brand Keywords in Sponsored SearchMarketing Science, vol. 37 no. 2 (2018), pp. 200-215. Experiments on Bing. This is the published, peer-reviewed version: working-paper figures differ.

Share this article