Your Top Affiliate Is Bidding on Your Own Brand Name

There is a specific kind of affiliate partner that looks like your best one and is actually your most expensive. Their conversion rate is superb. Their volume is steady. They almost never generate a refund. And every month you pay them a commission on customers who had already decided to buy, had already typed your brand name into Google, and were on their way to you anyway.

They are bidding on your brand name in paid search. They are buying an ad above your own listing, catching the click, redirecting the person to your site through a tracked link, and invoicing you for the introduction.

The reason this goes unnoticed for so long is that it is invisible in every report you look at. Affiliate revenue is up. Affiliate conversion rate is excellent. Cost per acquisition looks superb. Nothing in the affiliate dashboard says “this customer was already yours.” Meanwhile the damage shows up in a completely different system — your own search account, where brand keywords have quietly become more expensive — and nobody connects the two numbers because different people own them.

This is how to find out whether it is happening to you, using checks that cost nothing and take about twenty minutes, and then what to do about it — including the part nobody writes about, which is what you do when the partner doing it is genuinely one of your largest revenue sources.

What brand bidding actually is, and why it is not automatically theft

Brand bidding is an affiliate buying paid search ads against your own brand terms — your name, your misspellings, your product names, and the modifier phrases like “brand coupon” or “brand discount” — so that the click on the way to you gets attributed to them.

It matters to state clearly that this is not inherently fraud, because treating it as fraud is how these conversations go wrong. Some programs allow it deliberately. If your brand terms are being bid on by genuine competitors and you have decided not to defend them yourself, a partner holding that ad slot is doing you a service. If you are entering a market where you have no search presence at all, a partner with an established account can hold ground faster than you can.

The problem is not the activity. The problem is the activity happening without your knowledge, on terms you did not set, paid for by you twice. Twice, because two separate costs land:

• The commission. You pay a percentage on a sale you were going to get.
• The auction. The partner is now a competing bidder on your own brand keyword. Brand terms are usually the cheapest inventory you own, because you are the most relevant advertiser for your own name and nobody else is bidding. Add one determined bidder and that stops being true. Your cost per click on your own name rises, and it keeps rising as long as they stay.

The second cost is the one people miss, and it is often the larger one. It is also the one that persists after you have terminated the partner, because auction history does not reset the day you send the email.

The first check: your own brand cost per click over time

Before you look at a single affiliate report, open your search account and chart cost per click on your brand campaign for the last twelve months.

You are looking for a step, not a slope. Brand CPC drifting up across a year is normal — more competitors, more seasonality, a changing mix of query types. What indicates a new bidder is a step change that starts on a particular week and does not come back down. Someone entered the auction and stayed.

Then overlay impression share on the same campaign. The signature of a competing bidder is unmistakable once you know it: absolute top impression share falls, lost impression share to rank rises, and cost per click rises — all in the same week. If your budget did not change and your bids did not change, something else entered the auction.

This check is worth doing first because it is the only one that tells you when it started, and the start date is what you will need later when you are arguing about how much commission to reverse. If you skip straight to detection and find the ad today, you have a violation. If you find the week the CPC stepped, you have a violation with a date attached, and those are very different negotiating positions.

The mechanics of reading CPC, impression share and the rest of the paid-search vocabulary are in marketing metrics explained: CPC, CPM, CTR, CPA and ROAS in plain English if any of those terms are doing more work than they should be.

The second check: auction insights, which names them for you

Run the auction insights report on your brand campaign, because it lists the domains competing with you in the same auctions.

This is the fastest route to a name. Google’s auction insights report shows you which other advertisers appeared in the same auctions as your campaign, along with how often they were there and how often they outranked you. Segment it by month and you can usually see the exact month a new domain appears.

Two cautions on reading it:

The domain shown may not be the partner’s domain. Sophisticated brand bidders run from a domain that has nothing obvious to do with their affiliate identity, and some run through an agency account. A name you do not recognise is a lead, not a conclusion.

Absence is not proof. Auction insights has a minimum activity threshold, so a partner running a small, tightly-scheduled campaign — a few hours a day, one city — may never surface in it at all. This is not a hypothetical evasion; scheduling and geo-narrowing is precisely how an experienced brand bidder stays under a brand manager’s radar while still capturing meaningful volume.

Which is why the next check exists.

The third check: look at the actual results, properly

Search your own brand name the way a customer would, from the places your customers actually are — and understand why your first attempt will show you nothing.

If you search your own brand from your office, on your work machine, signed into an account that has clicked your own ads a hundred times, you are the worst possible observer. Personalisation, location and prior behaviour all shape what you get. More importantly, a competent brand bidder has already excluded your office IP range and your city from their targeting, because the first thing anyone doing this at scale learns is not to serve the ad to the brand’s own staff.

So run it deliberately:

That last row is the one that finds most of them. Bidding on the bare brand term is conspicuous and expensive. Bidding on “brand” plus a modifier is where the money actually is, because a person searching “brand discount code” has the highest purchase intent of anyone in your funnel and the lowest chance of being noticed by you. Check the modifier set explicitly. If you check only your bare brand name, you will find nothing and conclude you are clean.

The fourth check: read the ad, then read the redirect

When you do find an ad, the ad copy and the link chain will usually tell you who it is before you have to ask anyone.

Work down this list:

1. The display URL. Affiliates frequently display your domain — that is the whole point, it is what makes the ad convert. But the display path underneath it often carries something that is not yours.
2. The bolded text. Google bolds terms matching the query. An ad written by your own team and an ad written by a coupon site read differently once you are looking; the affiliate version is almost always more aggressive about discount language than your brand guidelines allow.
3. The final URL, after the redirect completes. Click it, then read the address bar on the page you land on. If there is a tracking parameter that belongs to your affiliate program — the network’s click ID, a partner ID, an aff_id-style value — you have your answer directly. Compare that value against your partner list.
4. The chain in between. Right-click, copy the ad’s link, and inspect where it goes before it reaches you. A direct-linking affiliate sends the click through the network and into your site in one visible hop. A cloaking affiliate bounces it through a landing page designed to be shown to Google’s reviewers but not to real traffic.

If the parameter identifies the partner, stop investigating and start documenting. Take screenshots with the query, the location and the timestamp visible, and save the full redirect chain as text. You will need it, because the partner’s first response is very often that you must have seen a competitor’s ad, or that it was a rogue sub-affiliate, or that it was one test that has since been turned off.

Why the network will not do this for you

The network is not a neutral referee. It is paid a percentage of the volume the partner generates.

This is the structural point that explains almost every frustrating conversation you will have on this topic. Networks generally have a policy on brand bidding, will enforce it when you present evidence, and are perfectly professional about it. What they will not do is proactively hunt for it on your behalf, because the traffic in question is revenue for them too. Your account manager’s incentive is aligned with volume, and this violation increases volume.

The consequence is practical, not moral: detection is your job. Assume nobody upstream is going to raise their hand. This is the same category of problem as the payout discrepancies in how affiliate tracking breaks quietly — the counterparty is not lying to you, but they are also not going to discover the problem for you, and every day you do not check is a day the number compounds.

The contract clause, and why yours probably does not work

The single most common reason a brand cannot recover anything is that its affiliate terms prohibit brand bidding in a sentence too vague to enforce.

“Affiliates may not bid on our trademarks” is the usual wording and it fails in four separate ways. A usable clause has to do all of the following:

Name the protected terms explicitly. Not “our trademarks” — an actual list: the brand name, its common misspellings, the domain with and without the TLD, product names, and the brand-plus-modifier set (coupon, discount, promo code, review, login, sign in, official, customer care). If it is not named, expect to argue about whether it was covered.

Say what is prohibited, precisely. Bidding on the term as a keyword is one thing. Using it in ad copy is another. Using it in the display URL is a third. Bidding on it in any paid platform — search, shopping, marketplace ads, app store ads — rather than only Google. And critically: prohibiting the keyword but permitting a broad-match campaign that catches it anyway is a loophole you will meet.

Cover negative-keyword obligations. The enforceable version of this clause does not just say “do not bid.” It says the partner must maintain your protected terms as negative keywords in every campaign. That converts a difficult argument about intent into a simple, checkable, binary state — either the negatives are in the account or they are not. Ask for a screenshot of the negative list. A partner acting in good faith sends it in an hour.

State the consequence, with a number. Warning, then reversal of commissions attributable to the traffic, then suspension. Without a stated remedy, the reversal is a negotiation. With one, it is a term the partner agreed to.

If your current terms fail these tests, fix the terms before you open the conversation with the partner. A clause you strengthen after the fact does not apply retroactively, and going in with a weak clause and a strong accusation is how you end up with no recovery and no partner.

What to actually do when the partner is your biggest one

Decide what outcome you want before you send anything, because the instinct — terminate immediately — is usually the wrong commercial call.

Here is the uncomfortable arithmetic. If this partner is a meaningful share of your affiliate revenue, some of that revenue is real. They almost certainly do genuine work as well: content, an audience, email, placements that produce customers who had never heard of you. The brand-bidding traffic is a layer on top of that, not the whole relationship. Terminating removes both, and it removes the incremental part along with the parasitic part.

So work it in this order:

1. Quantify before you accuse. Split their volume into brand-driven and everything else. The redirect parameters and landing pages usually make this separable — brand-bid clicks arrive from paid search on a brand query and convert almost immediately; genuine content traffic has a different referrer, a different time-to-conversion and a different device mix. You need this split anyway, because it is the only way to know whether you are having a small conversation or a large one.

2. Open with the negative-keyword request, not the accusation. “We are tightening brand-term policy across the program. Please confirm our protected terms are set as negatives in your accounts and send confirmation by Friday.” A partner who is not doing it complies in an hour. A partner who is doing it either complies — in which case the behaviour stops, which is the actual objective — or starts negotiating, which tells you everything without you having made an allegation you might have to walk back.

3. Set the go-forward rule before arguing about the past. Stopping the bleed is worth more than winning the historical argument, because the auction cost keeps accruing daily while a commission dispute is a fixed number that is not getting bigger. Get the negatives in place first. Then discuss reversals.

4. On reversals, ask for what is provable. The traffic you can tie to a specific paid brand click on a specific date, with the screenshot and the parameter. Not a round-number estimate of the whole period. Provable claims get paid; estimates get argued for three months.

5. Consider a differentiated rate instead of termination. This is the option most brands never consider and it is frequently the right one: a materially lower commission on brand-term and coupon-driven conversions, and the full rate on everything else. It aligns the partner’s incentive with the behaviour you actually want, keeps the genuine volume, and removes the reason to do it. It also, usefully, sidesteps the argument about whether the past traffic was incremental — you are changing the price of a category, not accusing a person.

That is the same reasoning as any commercial term in a partner agreement: attribution windows are a commercial term, not a technical setting. What you pay for, and what you pay less for, is a decision you make and write down — not a default you inherit from the platform.

The policy positions, and which one to pick

There are four workable positions on brand bidding, and the mistake is not picking the wrong one — it is not picking one at all.

The default recommendation for most businesses is prohibited, with a written whitelist exception process — because it is the only one that makes the answer to “is this partner allowed to do this?” a lookup rather than an argument. But whichever you pick, publish it in the program terms, restate it in the partner onboarding, and put it in the monthly partner email. Enforcement is dramatically easier against a rule the partner has been reminded of four times than against one buried in terms they accepted two years ago.

When you should buy a monitoring tool

Buy monitoring when the arithmetic works, and not before — the checks above are free and catch the majority of cases.

The honest position is that paid monitoring services do something you genuinely cannot do manually: they run brand queries continuously, from many locations and devices, at all hours, which defeats geo-narrowing and ad scheduling. That is real capability. What they are sold on, though, is fear, and the sales conversation will not help you decide.

The decision rule is straightforward. Monitoring is worth it when the recoverable commission plus the brand CPC inflation you would prevent exceeds the subscription — and you can only estimate that after you have done the manual checks once and found out whether you have a problem at all. Run the free checks first, quarterly, on the modifier set as well as the bare brand. If they keep coming back clean and your brand CPC is stable, you are paying for insurance against a risk you do not currently carry. If they keep finding things, or if you are running a program large enough that manual checks cannot cover the partner list, the tool pays for itself.

The other thing worth knowing: Google’s trademark policy lets a trademark owner complain about use of the mark in ad text, which is a narrower remedy than most people assume. Bidding on your trademark as a keyword is generally permitted by the platform in most regions; using it in the ad copy is what the complaint process addresses. So the platform complaint is a supplement to your contract, not a replacement for it. Your leverage is commercial, not platform-level. The contract is the instrument that actually works.

The check to run every quarter

Put four things in a recurring calendar entry and the whole problem stays small:

1. Brand campaign CPC and impression share, charted over twelve months. Looking for a step, not a slope.
2. Auction insights on the brand campaign, segmented by month. Looking for a new domain.
3. Manual searches on the modifier set — coupon, discount, promo code, review, login — from at least two cities and both devices, outside working hours.
4. Negative-keyword confirmations from your top partners by revenue. One email, screenshot in reply.

That takes under an hour a quarter and it is the entire defence. The reason it works is that brand bidding is not a permanent state — partners test it, and if nothing happens, they scale it. The quarterly check catches it in the testing phase, when the commission at stake is small and the conversation is easy.

Two related things are worth auditing on the same schedule, because they fail the same way — quietly, in a system somebody else owns. One is whether you actually control the accounts this evidence lives in, covered in who owns the ad account. The other is the reconciliation habit itself: the discipline of comparing what a partner reports against what your own systems observed, rather than accepting one number because it arrived in a nicely formatted email. Partners are not usually dishonest. But nobody else in the chain is paid to find your problems, and a metric that only ever gets read in the counterparty’s dashboard is a metric that will eventually surprise you.