The Verification Files
TVF-005 · Affiliate fraud · 35 min listen

How Affiliate Fraud Gets Paid: Ad Hijacking, Cloaking and Sub-Networks

Affiliate fraud survives in plain sight. How ad hijackers copy brand ads, hide behind cloaking and collect commission through sub-networks nobody vets.

Key findings

  • Ad hijacking copies a brand’s own search ad and adds an affiliate ID to the click. The brand pays commission on a sale it was already going to make.
  • The operator is usually unknown. The payment route is not. Commission has to pass through an affiliate programme, and sub-networks are the most common way on to one.
  • Cloaking hides the affiliate ID from anyone who looks like an investigator. Tools for it are sold openly.
  • In lead generation, every campaign observed by Simon Delaney has contained fraud. The lowest rate seen was 15% to 20%. Some campaigns were entirely fraudulent.
  • Affiliate marketing has no central vetting and no shared blacklist. Stands, awards and familiar faces at events stand in for scrutiny.
  • The advertiser holds the money. Change in programmatic advertising came when advertisers made fraud detection a condition of buying.

Contributing intelligence

  • Simon Delaney

    Founder, Provero · LinkedIn profile

    Background in lead generation, identity, verification and software for data processing. Ten years on the lead generation side of affiliate marketing, building landing pages and detecting fraud in the traffic sent to them.

  • Andy Cooney

    Co-founder, Marcode · LinkedIn profile

    Marcode detects ad hijacking, brand bidding and affiliate rule-breaking in paid search, and helps advertisers stop commission going to the affiliates behind it. Contributed expertise on ad hijacking, cloaking, sub-networks and how affiliate programmes are policed.

Contributors are sources, not authors. The Verification Files is responsible for the analysis and the assessments in this file.

Summary assessment

Affiliate fraud is the earning of commission that was not earned. It takes two main forms. In sales-based programmes, an affiliate claims credit for a sale the brand would have made anyway. In lead generation, an affiliate is paid for leads that no person submitted.

The assessment of this file is that affiliate fraud persists because of how the channel is built. The operators are hard to identify. The route by which they are paid is visible and is rarely closed. No central body vets participants. Being seen at events passes for being legitimate.

Source: high. Delaney and Cooney each report from their own work, on problems they deal with directly.

Evidence: stated for the methods and the figures. Contact sources for evidence. The link between ad hijacking and lead fraud is inferred. Every lead generation campaign Delaney has worked on contained fraud. The lowest rate he has seen is 15% to 20%. That is the minimum, and some campaigns were entirely fraudulent. Each source sees part of the market. The rate across affiliate marketing as a whole has not been measured.

Method: ad hijacking

Ad hijacking is the best known form. It is also called brand poaching, direct linking and URL jacking. Cooney describes the steps:

  1. Open a Google Ads account in another name. A company name or an individual’s. Verification needs only a few details about a company, all publicly available.
  2. Run ads that look identical to the brand’s own. Same name, same display URL.
  3. Add an affiliate ID to the click. The customer lands on the brand’s site as normal.
  4. Collect commission through the affiliate programme.

The customer was searching for the brand. The click would have reached it anyway. The brand now pays commission on that sale and pays more for its own clicks, because it is bidding against a copy of itself.

Cooney’s firm found the practice by accident. It saw ad accounts registered around the world running ads in the names of brands, and traced them to affiliates. He had spent fifteen years in agencies running paid search for some of the largest advertisers and had never seen it raised.

Related methods named in the file:

  • Cookie stuffing through browser extensions. An extension with many users is a very cheap source of traffic to claim.
  • Brand bidding through a third-party site. The click passes through an intermediate page before reaching the brand.
  • AI-generated publishers. Websites and influencer profiles produced quickly with AI, then used to apply to programmes. Social accounts can be bought and filled with fake posts.

Who is exposed

Status: Cooney, from his firm’s own studies.

The brands most affected have a large affiliate programme and one of two features:

  • A low cost per click on search. Retail, travel, telecoms and parts of finance.
  • A high average order value. Travel again, where one booking can be a package holiday.

Within that profile, more than half of the brands checked had at least one issue. Many had several. One client stopped commission payments in the high six figures within three months, all of it on sales that were not incremental.

Who is behind it

The operator is usually unknown. Ad accounts are registered in borrowed names. AI-generated profiles have nobody visible behind them. Cooney cannot say which group of people is responsible.

The payment route is known. Running the ads earns nothing until the operator is on the brand’s affiliate programme. That is the point at which the fraud can be seen and stopped: the affiliate ID, and the network it sits on.

The payment route: sub-networks

The most common route on to a programme is a sub-network. A sub-network holds an account with a main network and is approved by the advertiser through it. The operator joins the sub-network as a sub-publisher and is never seen by the advertiser.

Both sources are clear that sub-networks are not all alike. Some vet actively. Some control their own audience, such as an email list. A smaller number account for a large share of the problems. Awareness has grown and Cooney reports the route is used less than it was.

Delaney, from lead generation: fraud appears where traffic leaves the main network and passes from one sub-network to the next in a chain.

  • Every lead generation campaign observed has contained fraud.
  • The lowest rate seen is 15% to 20%.
  • Some campaigns have been 100% fraudulent.
  • The usual form is a bot submitting stolen data to a form.

Concealment: cloaking

Fraudulent affiliates cloak their links. If the visitor looks like a bot or an investigator, the affiliate ID is never shown. The click appears to be an ordinary ad.

  • Third-party cloaking tools are sold online.
  • Some name the monitoring firms they are built to evade.
  • Cooney describes a supplier industry serving people who commit fraud.

How each source detects it:

  • Cooney hunts. His firm runs millions of searches a day to find suspect ads, then defeats the cloaking to recover the affiliate ID. The advertiser can then tie that ID to the sales it was paid for. The target keeps moving.
  • Delaney catches. Hosting the landing page records two things: the browser, device and operating system the visitor presents to the network, and the bot actually executing the submission behind that mask.

Two kinds of fraud, one set of actors

The frauds differ in what is fake.

  • In sales programmes, the sale is real. Almost every hijacked sale is a real purchase by a real customer. The commission is what is false. The sale was not incremental.
  • In lead generation, the lead is false. No person made the submission.

Delaney’s assessment is that the same people are likely to be behind both. Somebody prepared to do one has no reason to stop at the other. This has not been confirmed. The two sources have not compared names.

Why it survives in plain sight

Delaney identifies four conditions.

  1. The barrier to entry is low. A tracking platform can be bought off the shelf. A network will supply offers to anyone who says they have traffic. Buying media on the large ad platforms took skill and capital, which kept fraudsters out. AI has lowered that barrier too.
  2. The skill required is sales. No technical ability is needed.
  3. Buyers want to believe. High volume, low cost and a plausible story get a yes.
  4. Familiarity replaces scrutiny. The same people attend the same events, take stands, win awards and produce polished marketing. Being present comes to pass for being legitimate.

Cooney confirms the pattern. A firm selling fraud detection takes a stand at a trade show and gives away T-shirts. A firm selling fraud would do exactly the same.

The names are known and not shared. Put experienced people in a room and ask who to avoid, and they produce the same list. That list exists only in people’s heads. Cooney notes he sees only part of any publisher’s activity and cannot put a percentage on it. Some of those named may not be wholly fraudulent. They have loose controls and do not vet.

Honest operators are damaged twice. A well-meaning person who is good at sales can start a sub-network with no fraud controls. Fraudsters seek out exactly that sub-network. Then advertisers who have been burned stop working with sub-networks altogether, including the careful ones.

Why nobody polices it

  • The channel is decentralised. No platform owns it. That is an advantage, because no single company can change the rules overnight. It also means nobody can ban anyone.
  • Responsibility is passed around. Advertiser, network, agency, event organiser. An organiser selling stands has no reason to refuse a buyer with a client list and a brochure.
  • The trade body has limited reach. The Affiliate and Partner Marketing Association publishes guidance and opened a Compliance Reporting Portal on 6 July 2026. Cooney reports that its earlier whistleblowing route received very few submissions. Member firms that monitor fraud are to vet submissions against evidence.
  • Evidence is not requested. Delaney holds ten years of it: who, how, and on which campaigns. Nobody has asked for it.
  • Nothing happens until someone looks. In Delaney’s experience the activity continues unabated until an advertiser or a technology provider notices it.
  • New programme managers are not warned. Cooney can tell within days of auditing a programme whether its manager restricts who joins. Where there are many problems, the manager did not know to look and approved everyone. Automatic approval of applicants is still in use.

Cost to the channel

  • A chief marketing officer finds the problem and shuts the affiliate programme. Both sources have seen it.
  • A chief digital officer at a company with revenue above a billion told Cooney that he does not run affiliate at all, because he does not trust it.
  • One of the largest networks has stopped running lead generation, according to Delaney, because of the volume of fraud it saw.

Money that could circulate in the channel leaves it. Cooney calls the lack of trust an existential threat to its growth.

Precedent: programmatic advertising

Programmatic is also decentralised and has had the same problems. Cooney, who bought it in a previous role, reports that change came from advertisers. Fraud detection and brand safety became a required part of every tender. The problem is not solved. It is taken seriously.

Assessment

The advertiser is the party spending the money and is the party with the strongest reason to act. Networks, agencies and trade bodies have a part. None of them will move until advertisers require it.

Decisions about who to exclude need evidence that others can inspect. A list held in people’s heads protects nobody who was not in the room.

The countermeasures follow below.

Listen to episode 005.

Countermeasures · TVF-005

Countermeasures: Affiliate fraud that still gets paid

The operator is hidden. The payment is not. Commission and lead payments pass through programmes that few advertisers can see into, and it carries on until somebody looks.

  1. Ask for direct publishers only

    Question whether to work with sub-networks at all. If you run through a network, ask whether the programme can be limited to direct publishers. If sub-networks are included, require the technology to police them as a condition.

  2. Have every sub-network labelled

    Networks and agencies can tell a direct publisher from a sub-network, and a sub-network that names its publishers from one that will not. Require that label on every partner in your programme, with a plain account of what it means. They could do this easily. They will not unless advertisers force it.

  3. Turn off automatic approval

    Nobody should join a programme unreviewed. Automatic approval lets a fraudulent affiliate run until somebody notices.

  4. Never send affiliate traffic to your own lead form

    Most advertisers have no fraud technology on their own forms. Affiliate traffic sent there cannot be checked, and it ends badly. Send it to a landing page that carries its own tracking, separate from the network’s link.

  5. Capture the real metadata

    When traffic is redirected from the network’s tracking link to a page you control, your tracking records the real metadata, which is masked to the network and the agency. This is where the fraud shows up.

  6. Check timing, details and sources

    On a page you control you can read the timing of submissions, verify the personal details and first-party data, and compare each source against the others and against itself. Add breach detection, IP geolocation, IP fraud checks and name-to-phone matching. Mechanical patterns, fraud flags and high breach rates are the signs.

  7. Use monitoring for ad hijacking

    There is not much an advertiser can do about ad hijacking by hand. The ads are cloaked from anyone who looks like an investigator. Recovering the affiliate ID behind them takes monitoring at scale.

  8. Withhold payment on suspicion

    Put it in the terms. Where fraud is suspected, commission and lead payments are withheld until it is shown that fraud did not take place. Set a time limit, such as one month, so it does not drag on.

  9. Make fraud controls a condition of buying

    Advertisers hold the money. Write fraud detection, sub-network transparency and withholding terms into every tender and contract, as advertisers did in programmatic. Networks and agencies act when it is required of them.

  10. Push for a central evidence register

    The industry needs one body that receives evidence of fraud from every network and technology provider: the metadata, the supplier names, the supporting evidence and the remedy applied. Every case that has not been disproven goes in. It does not exist because nobody wants to admit that fraud happened on their network or in their agency.

What does not work

  • AwardsProblem publishers win them too.
  • A stand at an eventAnyone can buy one.
  • A familiar faceFamiliarity is not scrutiny.
  • A network recommendationNetworks do not reliably separate direct publishers from sub-networks.
  • An informal blacklistA list held in people’s heads protects nobody outside the room.
  • Waiting for it to stopIt carries on unabated until an advertiser or a technology provider notices.

Limits

Nothing fixes an advertiser who does not know to look. Whether that naivety is the advertiser’s fault, or the industry’s for not telling them, is an open question. The people who know which operators to avoid keep it to themselves. The advertiser is still the one who pays.

Example checks

Example checkIllustrative
Lead as the network sees it
device
iPhone, Safari
location
UK
source
sub-publisher
status
payable
PayableLooks like a customer. Payment due.
Example checkIllustrative
Same lead on a page you control
real device
automated browser
timing
mechanical
breach
details in known breaches
sources
same pattern across three
WithheldBot with stolen data. Payment withheld.

Example checks are illustrative. They are not real records.

Operational resources

Relevant services and organisations that can help implement these countermeasures.

Marcode

Andy Cooney, a contributor to this file, is its co-founder.

  • Marcode: Paid search and affiliate monitoring

    Tracks who is bidding on a brand’s terms and identifies the affiliates behind hijacked ads. It gives the advertiser the grounds to block them and stop their commission.

    Supports: 07 Use monitoring for ad hijacking · 08 Withhold payment on suspicion

Provero

Provero was founded by Simon Delaney, who produces The Verification Files.

  • Provero: Affiliate fraud detection

    Assesses each affiliate lead before it is accepted or paid for: how and where it was submitted, and whether the details support that picture. It does not cover ad hijacking or brand bidding in paid search.

    Supports: 05 Capture the real metadata · 06 Check timing, details and sources

A resource is one way to carry out a countermeasure. Listing is not an endorsement. Some countermeasures may only be able to be applied via a resource. How resources are chosen

Frequently asked questions

What is ad hijacking in affiliate marketing?

Ad hijacking is a form of affiliate fraud in which an affiliate runs paid search ads that look identical to a brand’s own ads and adds an affiliate ID to each click. The customer was already searching for the brand. The affiliate collects commission on a sale the brand would have made without it. It breaches programme terms and pushes up the brand’s own cost per click.

How do fraudulent affiliates get paid?

Through an affiliate network or platform. The fraud only earns money once the operator is approved on the advertiser’s programme. The most common route is a sub-network: the sub-network is approved by the main network, and the operator joins the sub-network as a sub-publisher.

What is an affiliate sub-network and why is it a risk?

A sub-network is an affiliate that holds one account with a network and passes offers on to publishers of its own. The advertiser sees one partner and cannot see who is behind it. Not all sub-networks are a problem, and some vet actively. The risk rises where traffic passes from one sub-network to another in a chain.

What is cloaking in affiliate fraud?

Cloaking shows different content to different visitors. A fraudulent affiliate uses it to hide the affiliate ID from bots, auditors and anyone who looks like an investigator, so the click appears clean when checked. Third-party cloaking tools are sold online, and some advertise the monitoring firms they evade.

Which brands are most exposed to ad hijacking?

Brands with a large affiliate programme and either a low cost per click or a high average order value. Retail, travel, telecoms and parts of finance fit the profile. Travel is hit hard because a single booking can be large.

Are sales from ad hijacking real sales?

Almost always. The customer exists and the purchase happened. The fraud is in the commission, which is paid on a sale that was not caused by the affiliate. These are known as non-incremental sales.

Should advertisers send affiliate traffic to their own lead forms?

No. Most advertisers have no fraud detection on their own forms, so affiliate traffic sent there cannot be checked. The traffic should go to a landing page with its own tracking, separate from the network’s tracking link. That page records the real metadata of each submission, which is masked to the network, along with timing and source patterns.

Why does affiliate fraud persist?

Four conditions combine. The barrier to entry is low. The skill needed is sales, not technology. Buyers want to believe offers of high volume at low cost. Familiarity at events passes for legitimacy. There is no central body that vets participants or shares a blacklist.

Who is responsible for stopping affiliate fraud?

Responsibility is disputed between advertisers, networks, agencies and event organisers. The assessment in this file is that it falls to the advertiser, because the advertiser is spending the money and has the strongest incentive.

Exhibits: sources and further reading

Related files