How Fake Leads Become KYC Data: From Breach to Identity Check
Breached records are replayed through lead forms by bots, paid for as leads, resold as co-registration data and end up confirming identities in KYC checks.
Key findings
- A fake lead is usually a real person’s details submitted by a bot. The data comes from an old breach. The submission is the only invented part.
- The chain has six stages: breach, resale, replay through a lead form, payment, resale as co-registration data, and use as reference data in KYC checks.
- It passes because the data is real. At about £1 a lead, there is no budget for checks beyond whether the details look valid.
- The bot leaves a metadata trail: time, device, browser, IP address. If the lead is ever challenged, that trail is produced as proof that the person visited.
- Lead fraud and KYC fraud appear to be run by different people. If they converge, someone trying to defeat an identity check could write the reference data it is checked against.
Summary assessment
A fake lead is rarely made-up data. It is usually a real person’s details, taken from an old breach and submitted to a lead form by a bot. The lead is paid for, resold as co-registration data and passed into the commercial databases used for KYC checks. A record that began as stolen property ends as evidence that an identity is genuine.
The assessment of this file is that affiliate lead fraud is a supply line for commercial KYC data. The two industries are treated as separate. One feeds the other.
This file follows where commercial KYC data comes from, which found breached records in KYC inflow. This file sets out the route by which they arrive.
Source: high on stages 1 to 4 of the chain, which Simon Delaney detects first-hand in lead generation traffic. Medium on stages 5 and 6, which rest on his working knowledge of how co-registration data is resold. Background in lead generation, identity, verification and software for data processing.
Evidence: stated for stages 1 to 4. Contact source for evidence. None on file for stages 5 and 6. Hypothetical: the convergence threat in the final section. It is reasoning about what comes next, not a record of what has happened. What share of commercial KYC data arrives this way has not been measured.
The chain: six stages
Stage 1: breach
Data is stolen from a company’s systems, for example a retailer’s customer database of several million records. The thief demands a ransom. Companies are advised not to pay. When they do not, the breach becomes known.
Stage 2: ageing and resale
Fresh stolen data may be used first for high-value fraud: applications for loans, credit cards and bank accounts. Or it is held. Over time it is sold and resold until it trades like any other list. By the time it reaches a lead fraudster it is typically five to seven years old. It holds names, email addresses, phone numbers, full addresses and dates of birth.
Stage 3: replay through a lead form
The fraudster signs up as a traffic source with an affiliate network, a brand or a co-registration site, and agrees a price for each valid lead. A typical figure is £1.
No people are sent to the form. A script submits the breached records instead.
- The real device is masked. The network sees a different phone or browser for each visit.
- Clicks are faked to make the traffic look like a campaign.
- Residential proxies supply the IP addresses. Each submission appears to come from a different household.
- The order is irregular, to avoid an obvious pattern.
There is always a tell. It is found only by someone running fraud detection who knows what to look for.
Stage 4: payment
The buyer may run phone and email verification. The records pass, because the details are real. At the end of the month the buyer states how many leads it accepts. In the source’s example, 40,000 are submitted, 30,000 are accepted and £30,000 is paid. The buyer asks for more.
Stage 5: resale
The collector has paid for the data and needs to recover the cost. It sells the records as co-sponsor data to as many buyers as it can:
- call centres
- email marketers
- commercial KYC data companies
The source reports that most of the money is made here, not from the co-registration offers shown to visitors, and that some collectors now earn almost entirely from KYC.
Stage 6: corroboration
KYC data companies aggregate the records. They match them against what they already hold, append new fields, note changes of address and overwrite older entries. The record becomes reference data. When a firm later checks an applicant called John Smith at a given address, this is the record that answers yes.
Everyone in the chain has been paid. The person in the record does not know any of it happened.
Why it is not caught
- The data is real. One industry contact, who knew that a database feeding KYC was built from stolen data, put it to the source plainly: stolen data is good data. The emails work, the numbers connect and the addresses exist.
- The price leaves no room for checks. At £1 a lead, anything that looks real is accepted.
- The collector has already paid. Its interest is in reselling, not rejecting.
- Fraud detection is absent. Most collectors do not run it because of the cost.
- KYC data is never used to contact anyone. It sits in a database and is matched against. Nobody receives a call or an email, so nobody complains.
The metadata trail and plausible deniability
Suppose the person finds out. They receive a call from a call centre, or make a subject access request to a KYC company, and ask how their data was obtained.
The enquiry passes back through each reseller, perhaps three steps, to the site that collected the lead. That site holds a full record:
- the date and time of the visit
- the device, operating system and browser
- the IP address
- the answers given
- how long the form took to complete
The site can say in good faith that, as far as it can see, the person visited and filled in the form. The burden now falls on the individual to prove they did not. It could have been a neighbour. It could have been a friend. Nothing further happens.
The trail was written by the bot. See why valid data is not a real person.
Reading: the metadata of one record proves nothing, because a bot produces it as easily as a person does. The fabrication shows only across many records from one source, where the same patterns repeat.
The same structure protects the supply side. Most of this traffic arrives through sub-networks. Each can say the traffic came from the publisher below it. See how affiliate fraud gets paid.
Two frauds that have not met
Lead fraud. The aim is the payment per lead. The identity is raw material. Once paid, the fraudster has no further interest in the record.
KYC fraud. The aim is to pass an identity check as someone else. Methods include synthetic identities, the use of real people’s details and SIM swapping to intercept security codes.
Status of the scale figures: corroborated by published industry data. Cifas recorded close to 130,000 identity fraud cases in the UK in the first half of 2026, 59% of all cases filed, with unauthorised SIM swap cases up 402%. UK Finance reports £1.28 billion stolen through payment fraud in 2025.
The source’s assessment is that these are two different sets of people in two unrelated industries. One is paid at the start of the chain. The other attacks the end of it.
Threat assessment: if the two converge
Status: hypothetical. There is no evidence that this is happening. The source suspects it is not.
A person trying to defeat KYC checks who also ran an affiliate sub-network would control what enters the reference data. That person could:
- submit chosen identities into the co-registration market in volume
- place the same records with many collectors, so that several KYC databases corroborate each other
- pair a real person’s details with a different phone number. Most KYC databases hold only the data spine. Where numbers are held, a new one is stored beside the old and either will match
The check would then confirm an identity against data the fraudster wrote.
At present the weakness is that stolen data circulates as reference data. The larger risk is that somebody uses the route on purpose.
Unverified
The source holds a suspicion that some collectors do not buy fraudulent traffic at all and generate the submissions on their own pages. There is no evidence for this. It is recorded here as a suspicion only.
Assessment
Affiliate lead fraud and commercial KYC data are connected by the resale of co-registration data. A weakness at the cheap end of the chain, a £1 lead that nobody can afford to check, becomes a weakness in identity verification for financial services.
The assessment is open to challenge on evidence.
The countermeasures follow below.
Countermeasures: Breached records entering KYC as leads
Old breach data is replayed through lead forms by bots, passes because the details are real, and is resold into the databases used for identity checks.
- Look for the pattern, not the record
These are files of tens or hundreds of millions of records. Catching one bad record makes no difference. Test in aggregate. The aim is to find the pattern that infects many records and remove its source.
- Know the tells in a batch
Mechanical patterns in the metadata. Mechanical timing in when the data arrives. Many suspicious IP addresses. Breach results that repeat. Every record formatted in exactly the same way. Very large volume with a doubtful account of how it was generated. Ask for the data as the user entered it, so that formatting can be judged.
- Run fraud detection on the form
Residential proxy traffic is caught by fraud detection and IP fraud checks at the point of capture. Without them it looks like ordinary households.
- Reject the cost argument at volume
A collector taking 3,000 leads a month at £1 may have little to spend. A collector claiming 250,000 a month is spending a quarter of a million pounds. Checks cost a fraction of a penny a record. At that volume, “we cannot afford to check” does not make sense.
- Investigate any file over 70% breached
Do not block an individual record for breach exposure. Look at the file. Above 70% it needs investigating. Then look at which breaches the records appear in. Row after row from the same breaches points to one stolen source.
- Pass provenance on with every record
Before reselling, a collector should run fraud detection and IP fraud checks and remove the bots. Each record passed on should carry the URL it was collected on, the date and time of the original sign-up, how the traffic was generated and the metadata of the submission.
- Require the same at the KYC end
A KYC data company should demand that provenance from every supplier: the URL, the original sign-up date and time, the submission metadata, the IP address and how the traffic was generated.
- Treat volume as the first warning
KYC wholesale companies are best placed to see this route being misused. The first tell is volume. Ask how the supplier is producing that much. Then ask for all the corroborating data that should come with it.
- Check new phone numbers with the mobile network
Most KYC databases hold the data spine and not phone numbers. Where a new number does arrive for a person already held, it is stored beside the old one and either will match. Run a mobile network operator check on the new number against the details held. It confirms the person against that number, or it does not.
- Cross-check traffic sources against known fraud
The affiliate and KYC industries are unlikely ever to talk. If the affiliate industry kept a record of fraudulent sources, a KYC buyer could check it against the traffic sources named in a provenance trail.
What does not work
- Phone and email verificationStolen data passes. The details are real.
- One record’s metadataA bot writes metadata as easily as a person does.
- Catching single recordsPointless at this scale. Find the pattern.
- “Checks cost too much”Not from anyone buying leads by the hundred thousand.
- A subject access responseAn honest collector can only repeat what its metadata says. A dishonest one has deniability.
- Looking validTrust nothing until your own checks show the data is genuine.
Limits
Many companies in this chain earn more when match rates are high. That interest is part of the problem, and no check removes it. A collector answering a subject access request also has little it can verify. It can only report the trail it holds.
Example checks
- valid
- phone
- connects
- address
- real
- metadata
- device, IP, time present
- formatting
- identical on every row
- timing
- mechanical
- breach
- 90%, same breaches
- ip
- residential proxies
Example checks are illustrative. They are not real records.
Operational resources
Relevant services and organisations that can help implement these countermeasures.
Provero was founded by Simon Delaney, who produces The Verification Files.
- Provero: Form verification
Runs checks behind a form as it is submitted: contact details, connection signals and submission metadata. It does not prove on its own that a real person submitted the details.
Supports: 03 Run fraud detection on the form
- Provero: Source audit
Assesses a supplier’s file or a sample for recycled or fabricated data, and reads breach history across the file. Breach exposure is not treated as proof on its own, and it cannot establish the original provenance of every record.
Supports: 01 Look for the pattern, not the record · 05 Investigate any file over 70% breached
- Provero: Mobile network KYC
Compares the details held for a person with the subscriber record for a mobile number. It covers the UK and Spain, and mainly contract numbers.
Supports: 09 Check new phone numbers with the mobile network
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
How do fake leads end up in KYC databases?
Breached records are submitted to lead generation forms by bots. The site that collects them, often a co-registration site, pays for them as leads and then sells the data on. One of the buyers is the commercial KYC data market. The record is aggregated into reference files and used to confirm identities.
How is breached data used to create fake leads?
An affiliate agrees a price per lead with a network, brand or co-registration site. Instead of sending people to the form, the affiliate uses a script to submit records taken from an old data breach. Each submission is dressed to look like a different visitor on a different device. The details belong to real people who took no part.
Why do fake leads built from breached data pass validation?
Because the data is real. The email addresses deliver, the phone numbers connect, and the names and addresses belong together. Validation checks whether details are genuine. It does not check who submitted them.
What is a residential proxy and why does it matter in lead fraud?
A residential proxy routes traffic through household internet connections. Services sell access to very large pools of residential IP addresses in any chosen country. Used with a bot, it makes each submission appear to come from a different home. Without fraud detection on the form, the traffic looks like ordinary consumers.
How do you detect breached data being replayed as leads?
By testing the batch, not the record. The signs are mechanical patterns in the metadata and in the timing of arrival, many suspicious IP addresses, identical formatting on every record, and very large volume with a weak explanation. A file that is more than 70% breached needs investigating, more so when the same breaches appear in row after row.
Can submission metadata prove that a lead is genuine?
Not for a single record. A bot produces a complete trail: date and time, device, operating system, browser, IP address, answers given and time taken. That trail can be shown to a complainant as evidence of a visit. The fabrication shows only when many records from the same source are compared and the same patterns repeat.
What is co-registration data?
Co-registration, or co-reg, is the collection of consumer details on a site together with consent to pass them to third parties, known as co-sponsors. The data is sold many times: to call centres, to email marketers and into KYC reference files. For some collectors, resale into KYC is now the main source of income.
Could lead fraud be used to defeat KYC checks?
In principle. Someone who controlled a traffic source could choose what records enter the co-registration market and, through it, KYC reference data. That could include pairing a real person’s details with a phone number the fraudster controls. The source’s assessment is that this is not yet happening, because lead fraud and KYC fraud are run by separate groups. There is no evidence either way.