The networks settle on the last click. Google's own analytics stopped doing that.
Across the Awin, Rakuten Advertising and CJ Affiliate programmes we run, commission still settles on the last click before a sale. Google moved the other way in 2023 to 2024, making data-driven attribution the default in Google Analytics 4 and retiring last-click as its standard model. That divergence is the whole game. It lets an affiliate run a paid ad against your own company name, take the closing click from a customer who already typed your brand, and bill you a commission on a sale that was coming anyway.
The move that follows is the one most agencies will not recommend: reprice those bare-brand clicks to zero commission without expelling the partner, so the incentive to bid on your name disappears while everything genuinely incremental they do still pays. Treat any affiliate bidding on your bare brand name as non-incremental until a pause test proves otherwise, because the last-click model rewards standing where your demand already flows. To make the cost concrete, take an illustrative £100 order at a 10% commission: that is £10 handed over for a sale you had already won, before you count the auction cost of the click itself. That single example carries the rest of this edition.
An affiliate brand-name click bills you twice on one sale
Both costs land on a single click. A customer searches your brand, clicks the affiliate's paid ad, buys, and the network credits the sale to the affiliate. That is the commission described above, paid to a partner for a buyer who was already on their way to you.
Then add the auction effect. A brand term you can hold cheaply as the only serious bidder becomes more expensive once a partner you also pay is bidding against you in the same auction, so on that one sale you cover both a higher click cost and the commission, for demand you already had. Across the brand-name conversions a partner intercepts in a month, that commission adds up, none of it buying a customer you did not already have. Use your own rate and the shape does not change.
Last-click bidding also strips credit from the channels that created the demand
The damage is not confined to brand PPC. Last-click hands the whole sale to whoever touched the buyer last, so a partner sitting on your brand term takes 100% of the credit even when the buyer first arrived through another channel. Picture a newsletter click earlier in the week and a comparison page or two read before the buyer searched your name: in GA4 those earlier touches register as assisted conversions and data-driven attribution shares the credit across them, while under the network's last-click model they get nothing. Your email and content lines then read weaker than they are, and you risk cutting the spend that actually created the customer while paying the partner that stood at the exit. A review or comparison that reaches someone before they knew your name is incremental, and commission on that is money well spent. The test is not the partner's reputation but whether the sale would have happened without them, and on a bare brand search the honest answer is yes.
Cashback and voucher affiliates call brand bidding defensive. Make them prove it.
The pushback comes mainly from cashback, voucher and loyalty affiliates and the outsourced programme managers who represent them. Their standard line is that a presence on your brand results is defensive, capturing buyers who would otherwise drift to a rival who is bidding on your name. There are two versions worth a straight answer, and both have limits.
The first is genuine defensive occupancy: if a competitor or a rogue affiliate is squatting your brand results, some presence there is worth paying for. True, but you can hold that slot with your own ad. Defending it yourself costs you only the click on your own brand term; paying a partner to guard it costs you the commission described above on every order they close, order after order, for a door you already own. The second is long-tail trademark-plus bidding: pairing your brand with a product modifier can reach queries you do not already rank for, and that can be incremental, so permit it only where you do not already occupy the result. What neither argument defends is a partner bidding on your bare brand name where your own listing already sits. The burden of proof stays on the partner.
The pause test settles incrementality, if you control the confounds
Do not argue incrementality, measure it, but measure it in a way that survives the obvious objection. A simple two to four week pause on all affiliate brand-term bidding is confounded by seasonality, promotions and every other campaign change running at the same time, so a flat brand-revenue line proves little on its own. The fix is a geographic holdout: switch affiliate brand bidding off in one region and leave it live in a comparable one for the same weeks. Both regions share the same season, the same promotions and the same wider campaigns, so the difference between them isolates the partner's actual contribution.
Watch total brand-search revenue, not the affiliate line. Run it long enough, and across enough brand-search volume in each region, that the gap between the two clears normal week-to-week noise; a handful of conversions either way tells you nothing. If total brand revenue holds in the paused region while the affiliate's attributed sales fall, the partner was moving credit, not creating demand, and you stop paying. If total revenue drops, the partner was doing real work and you keep them. If the result is noisy and flat, read it as incrementality unproven, which leaves the burden where it started, on the partner. In the programmes we have audited, we have yet to inherit one already running this test.
Three checks show whether a partner is billing you for your own brand searches
The evidence sits in three places your team can check without new tools:
- The live results on your brand terms. Search your company name on a phone and a desktop across a normal week and note which partners appear above or beside your own ad.
- Your affiliate report. In Awin, Rakuten or CJ, filter for conversions whose last touch was a search for your brand name rather than a content or comparison page.
- Your own programme terms. Networks let you set a prohibited keyword list, ban brand-term bidding outright, or permit only trademark-plus bidding. A rule nobody monitors is not a rule.
Google will not police your affiliate deals, because its trademark rules cover ad text, not your commissions
Google will not intervene. Under its trademark policy, it allows advertisers to bid on trademarked terms as keywords, and its complaint process addresses only the use of a trademark inside ad text, with exceptions even there for resellers and informational sites. It does not arbitrate whether a partner should earn a commission on your brand traffic, monitor your affiliate agreement, or decide who bids on your name. The control sits entirely in your programme terms and your monitoring. Waiting for Google to step in is waiting for a referee who has already left the pitch.
Reprice brand-triggered sales rather than just policing them
The one lever most programmes never pull is pricing the behaviour out instead of chasing it. In our own hands-on configuration of Awin, Rakuten and CJ programmes, each network lets you set commission by traffic type or rule group, so you can drop the rate on brand-term-triggered conversions to zero or near it while the partner keeps full commission on the incremental traffic you actually want. Prohibition relies on someone catching a rule being broken; repricing removes the reward whether you are watching or not.
Two objections come straight back. First, the partner reclassifies the traffic, tagging a brand-triggered click as content to reclaim the commission. That is exactly what checks one and two catch: the live search results show who is bidding on your name, and the report shows the last touch, so a rule set by observed traffic type is enforced continuously rather than on trust. Second, the partner churns off the programme entirely. If their contribution was only intercepting demand you already had, losing them costs you nothing, because that traffic was non-incremental by definition: you keep the customer, you keep the sale, you simply stop paying the toll. The partners who create genuine demand keep full commission, so they have no reason to leave.
