Countermeasures
From problem to proof
Each case file, the problems it highlights and the countermeasures needed to combat them.
27 Things a Mobile Number Reveals: Signals for Fraud, KYC and Leads
Problems identified
- A phone number is stored as a contact field and checked only for format.
- Where a network lookup is run, it is used for live or dead and nothing else.
- Out of network numbers are accepted and dialled indefinitely.
- A passed check is taken as proof that the number is valid.
Countermeasures
- Treat the number as evidence
- For lead generation: live status and subscriber name
- Before KYC: SIM swap, exact age, subscriber name
- For fraud: SIM swap, out of network, subscriber name
- Run the cheapest check first
- Recheck out of network numbers, then reject
- Step up after a recent SIM or device swap
- Recheck stored numbers before using them
- Cover the gaps in network KYC
- Read the batch
When Identity Verification Becomes the Risk: Five Ways Checks Fail
Problems identified
- Checks the law requires before access are run after the point of entry.
- Old breached records pass as established ones.
- Data is collected and kept with no use for it.
- A customer is verified once and never again.
- AI agents make purchases for people nobody has verified.
Countermeasures
- Complete legally required checks before entry
- Sequence the rest on cost and friction
- Test the flow with everyone who has a stake
- Let the law set the floor and the board set the appetite
- Judge breach exposure by pattern, with provenance
- Keep data only if you use it
- Re-check on events and signals
- Make the cost case as insurance
- Treat an agent as someone acting for a person
- Set limits before giving an agent payment details
How Fake Leads Become KYC Data: From Breach to Identity Check
Problems identified
- Bots submit breached records to lead forms and are paid per lead.
- The details are real, so validation passes them.
- Collectors resell the records into KYC without the provenance behind them.
- A bot’s metadata trail is accepted as proof of a visit.
- The affiliate and KYC industries share nothing.
Countermeasures
- Look for the pattern, not the record
- Know the tells in a batch
- Run fraud detection on the form
- Reject the cost argument at volume
- Investigate any file over 70% breached
- Pass provenance on with every record
- Require the same at the KYC end
- Treat volume as the first warning
- Check new phone numbers with the mobile network
- Cross-check traffic sources against known fraud
How Affiliate Fraud Gets Paid: Ad Hijacking, Cloaking and Sub-Networks
Problems identified
- Commission is paid on sales the brand was already going to make.
- Lead payments go to bots submitting stolen data.
- Sub-networks hide who is sending the traffic.
- The network sees masked metadata, not the real submission.
- Evidence of fraud is held privately and never shared.
Countermeasures
- Ask for direct publishers only
- Have every sub-network labelled
- Turn off automatic approval
- Never send affiliate traffic to your own lead form
- Capture the real metadata
- Check timing, details and sources
- Use monitoring for ad hijacking
- Withhold payment on suspicion
- Make fraud controls a condition of buying
- Push for a central evidence register
Are Content Syndication Leads Real? The Content Swindlefication
Problems identified
- Real contacts are sold as people who downloaded your content.
- The supplier controls the landing page, so the buyer cannot see what happened on it.
- Buyers are told not to call, so nobody asks the lead.
- QA, certification and a perfect profile match stand in for proof.
- The lead is used as a consent record it cannot support.
Countermeasures
- Ask three questions before you buy
- Control the page or track it
- Check call recordings for splicing
- Ask for delivery in real time
- Without proof, call every lead at once
- Reject any lead that denies the download
- Ignore “do not call, nurture”
- Do not rely on the lead as consent
- Put the budget elsewhere
Where Commercial KYC Data Comes From, and Why a Match Is Not Proof
Problems identified
- A match is treated as proof, whatever the reference data is.
- Reference files hold old and breached records with no traceable origin.
- The same record sits in several databases and is counted as independent sources.
- Every party is paid on a match, so nobody is paid to question one.
Countermeasures
- Ask how the data was collected
- Get the original sign-up date
- Treat very large volumes as a warning
- Weight the match by its provenance
- Test whether sources are independent
- Test the data by originating site
- Require provenance with every record
- Audit the file you already hold
- Pay on provenance, not on a match
- Check what any persistent key is built on
Valid Data Is Not a Real Person: The Human Data Illusion
Problems identified
- Records are counted as people before anything has been proved.
- Valid details are submitted by someone other than their owner.
- Checks test the fields in a record, not who sent it or how.
- Bad data is defended by the people who paid for it.
Countermeasures
- Start from unproven
- Validate every field
- Read the submission metadata
- Test the batch as a whole
- Cross-check against an independent source
- Capture the submission yourself
- Pay only for records that pass
- Decide from the top