Card Testing Fraud: Protect African Payment Platforms

Content authorBy EGSPublished onReading time11 min read
A woman interacts with her smartphone at a casual desk, navigating a banking app, with a blurred laptop and plant in the background.

This article explains how to investigate and contain automated authorization abuse across a merchant portfolio without shutting out paying customers. It covers the signals worth correlating and a control matrix you can hand to the team on shift.

Card testing fraud costs

Card testing fraud is what happens when someone points a script at your checkout and runs stolen or guessed card numbers through it in small amounts to find out which ones still work. Visa puts the annual loss from enumeration at $1.1 billion, and almost none of that lands on the merchant being used as the test bench. The damage sits somewhere else.

For an acquirer or payment service provider (PSP), the cost shows up in four places. Authorization traffic climbs while settled volume stays flat, which drags your scheme ratios. Visa flags acquirers at an enumeration ratio of 20% with 300,000 or more enumerated attempts in a month, measured on authorization attempts rather than disputes.

Then there is the per-attempt processing cost. Fraud screening is metered, and a ten-thousand-attempt overnight run against a small merchant can cost roughly $0.02 per blocked review on some platforms. A single carding attack also burns analyst hours and goodwill with the merchant whose store was chosen.

The fourth cost is the one teams underestimate. Clamp down too hard and you start declining real buyers, and false declines already cost more than the fraud they prevent. Datos Insights puts the industry false decline rate at 1.51% of annual sales across e-commerce merchants. That's why a card testing attack is a decisioning problem before it's a blocking problem.

South African portfolio example

Picture the data. A portfolio alert fires because one merchant identifier in your book shows authorization attempts running several multiples above its own weekday baseline, while settled value barely moves. Average ticket drops toward the floor of what the merchant normally sells.

The card estate behind those attempts is wide and shallow. Hundreds of distinct primary account numbers, most appearing once or twice, many clustered inside a handful of bank identification number (BIN) ranges, submitted from a small number of newly created accounts. Visa's own guidance describes account testing as one or two low-dollar transactions aimed at a single BIN range to confirm a card is live.

The decline mix is the part that moves. Early in the run you see invalid account number declines dominate, then the mix shifts toward invalid credentials as the script finds live numbers but can't produce the matching expiry or card verification value. Stripe exposes exactly those two rates, invalid account number and invalid credentials decline rates over the past hour, as card testing fraud risk indicators.

Context matters in a South African book. Domestic card fraud losses rose 43.3% to R857.5 million in 2025 according to the South African Banking Risk Information Centre, with the increase tied to the shift of card use into online payments. A card testing attack against one small merchant sits inside that broader move, which is why portfolio-level correlation beats merchant-by-merchant firefighting.

Separate attacks from noise

Flowchart illustrating financial transaction data points with modular blocks, curved arrows, and a pale-blue gradient background.

One signal on its own proves nothing. Shared mobile networks and office IP ranges across Lagos or Nairobi produce IP concentration that looks identical to a botnet if you read it alone. Failed payments cluster for boring reasons too, because a customer whose card was soft-declined will retry.

Correlation is the test. These are the dimensions worth joining before anyone touches a control:

  • Velocity and value together: attempt counts per card, per account, and per device against the merchant's own baseline, paired with ticket size at or near the minimum the merchant accepts.

  • Identity and device: account age at first payment attempt, email reputation, and whether one device fingerprint is carrying many unrelated cards.

  • Network and merchant: autonomous system and proxy or data-center origin, read against which merchant identifiers in the portfolio are absorbing the traffic.

  • BIN and decline codes: how tightly attempts cluster in issuer ranges, and whether the decline mix is drifting rather than holding steady.

Legitimate retries look different in a way that's easy to confirm. They repeat the same card on the same account two or three times within minutes, and the decline reason stays constant. A card testing attack repeats the merchant endpoint and cycles everything else.

Card testing fraud also leaves an approval footprint. A small share of attempts will succeed, and those approvals are the attacker's product, so approved low-value transactions inside a declining burst deserve more scrutiny than the declines around them.

Start building your financial platform?

Speak with EGS engineers about open banking, payment infrastructure, cloud systems, and enterprise software.

Get in Touch →

Detection to response

The workflow matters more than any single rule, because the sequence determines whether you contain the attack or create a second incident. Each stage needs a named owner and a decision that can be written down.

  1. Alert and triage. Monitoring surfaces an authorization anomaly at merchant level. The fraud duty analyst owns it and has one job, which is confirming whether settled value moved with attempt volume.

  2. Signal validation. Risk analytics pulls card, device, BIN, and decline-code distributions over a comparable window. The decision point is whether two independent dimensions agree.

  3. Merchant investigation. Payment operations checks the integration itself for exposed publishable keys and unauthenticated payment endpoints. It also checks whether a cart flow allows unlimited payment-method attachment.

  4. Containment. Risk leadership approves a reversible control at a defined processing point, with a stated monitoring window and a rollback trigger agreed before deployment.

  5. Escalation and monitoring. Scheme ratio exposure goes to the compliance owner, since Mastercard expects acquirers to produce an internal investigation report with corrective actions for threshold events.

  6. Controlled recovery. Controls come off in stages against measured traffic rather than in one release.

Card testing fraud response fails most often between steps two and four, when a team skips validation because the volume looks obvious. Enumeration shows up in authorization traffic, which means the evidence is available in real time and there is no good reason to decide without it.

Card testing control matrix

Every control below is a test with an expected result. Write the rollback condition before deployment, because the pressure to leave a blunt control in place is strongest right after it works.Each control should be evaluated against its required evidence, accountable owner, success measure, processing impact, false-positive risk, and rollback condition.

A velocity cap per card and account should be supported by evidence of attempt distribution compared with the 30-day merchant baseline. Risk analytics owns this control. Success is achieved when attempt volume returns to the baseline band. The processing impact is negligible, and the false-positive risk is low. Rollback should occur if legitimate retry declines rise.

An IP and device rate limit at checkout should be supported by evidence of ASN and device fingerprint concentration. Payment operations owns this control. Success is measured by a drop in bot attempts while approval rates hold. The processing impact is added request latency, with a medium false-positive risk on shared networks. Rollback should occur if approval rates fall below pre-attack levels.

A bot challenge on the payment endpoint should be supported by automation evidence in session data. Merchant engineering owns this control. Success is measured by fewer scripted submissions. The processing impact is an additional checkout step, with a medium false-positive risk. Rollback should occur if cart abandonment climbs.

Step-up authentication on risky attempts should be supported by risk scores and BIN clustering. Risk leadership owns this control. Success is measured by fewer approved test transactions. The processing impact is issuer challenge time, with a medium false-positive risk. Rollback should occur if the challenge rate exceeds the agreed ceiling.

Minimum-value and zero-auth restrictions should be supported by evidence comparing ticket sizes with the merchant catalogue. Payment operations owns this control. Success is measured by a decline in low-value probes. The processing impact is none, with a low false-positive risk. Rollback should occur if genuine small-ticket sales are blocked.

A temporary merchant identifier block should be supported by evidence of a confirmed coordinated pattern. Risk leadership owns this control. Success is measured by attack traffic stopping. The processing impact is halted merchant revenue, with a high false-positive risk.

Start building your financial platform?

Speak with EGS engineers about open banking, payment infrastructure, cloud systems, and enterprise software.

Get in Touch →

Investigate the carding attack

The investigative question is whether the data is coordinated or coincidental. Join cards to accounts and all of it to timestamps, then ask whether a single actor could plausibly produce the result. Scripted traffic arrives on an inhuman cadence, with inter-attempt gaps that are too regular to be people.

Decline codes are your most honest witness. A drifting mix, from invalid account number toward invalid credentials, tells you the script is making progress. A static mix across thousands of attempts means an integration problem or a misconfigured recurring billing run, and treating that as a carding attack will cost you a merchant relationship.

Card testing fraud investigations also check what the attacker reached. If only a publishable key was abused, the account itself is intact and the fix belongs at the checkout layer. API calls using a secret key from unrecognized addresses are a different incident entirely, and keys get rolled immediately.

Contain suspicious activity

Reversible first. Velocity limits scoped to the affected merchant or the affected BIN ranges will stop most activity without touching anyone else's traffic. Stripe reports that adding managed challenges across its Checkout users cut card testing fraud by 80% with under 2bps impact on authorization rates, which is the shape of outcome worth aiming for.

Placement decides the cost. Rate limiting at the web front end stops traffic before it becomes a billable authorization, while scoring at the payment layer gives you better signal but after you've paid for it. Stripe's documentation recommends limiting new customers created per IP address in a day as a targeted deterrent, tailored to the specific pattern you're seeing.

Step-up authentication is the middle setting, and risk-based challenge selection keeps it from becoming a tax on everyone. One Peruvian issuer using Visa Consumer Authentication Service recorded a 30-percentage-point reduction in challenge rates alongside better authentication success. While any containment runs, watch authorization rate and merchant complaint volume on the same dashboard.

Communicate with merchants

Tell the merchant what you observed and what you did, in that order, without claiming more certainty than the data supports. They need the affected window and the control you applied. Vague reassurance is worse than silence because it stops them acting.

Ask for specifics in return. Server logs covering the attack window and a list of every live API key and where it sits. Checkout changes that help include capping payment attempts per session and removing zero-value or minimum-value flows that make validation cheap.

Card testing fraud updates work best on a fixed cadence agreed at the start, because merchants left guessing will either disable your controls or escalate above you. Say what you know and give the next update time.

Recover and validate

Remove controls in stages, loosest first, with a monitoring window between each step. If attempt volume stays flat when the device rate limit comes off, move to the velocity cap next. A single bulk rollback teaches you nothing about which control was doing the work.

Hold heightened monitoring on the merchant for longer than feels necessary. Visa found that 33% of enumerated accounts experienced fraud within five days of the attacker obtaining the payment data, so the second wave arrives while the first incident is still open in your tracker. Recurrence on the same merchant is a finding.

The incident review compares attack attempt volume and approval rate against the pre-attack baseline. If a control reduced the card testing attack but moved approval rate more than the attack itself did, retire it. Write the comparison down, because the next alert will arrive on someone else's shift.

How EGS could assess

EGS anti-fraud and payment-processing engineers work on the layer where this problem actually lives, which is signal quality and decision placement inside your own stack. An assessment scope covers what your authorization data captures and where in the request path decisions are made.

Deliverables are a signal inventory against the correlation dimensions earlier in this article and a tested control matrix with rollback conditions. Dependencies are read access to authorization and decline data and time with the people who run the controls.

Objectives are measurable and bounded. Reduce card testing fraud attempts reaching paid authorization and hold or improve approval rate on legitimate traffic. No assessment eliminates a card testing attack, and anyone promising that is selling something else.

Review platform exposure

Layered controls that have been tested and can be rolled back beat a single aggressive rule every time, because the second one costs you customers the moment the pattern shifts. The teams that come out of these incidents well are the ones who validated signals before acting and measured what each control did to real traffic.

Talk to EGS about card testing fraud exposure across your payment platform and the controls your team needs to evaluate. Book a call with our payment and anti-fraud engineers to review where your decisions sit and what your data can actually prove about card testing fraud.

Start building your financial platform?

Speak with EGS engineers about open banking, payment infrastructure, cloud systems, and enterprise software.

Get in Touch →

Preserve the alert window, raw authorization records, control settings, and baseline comparisons before changing anything. Export timestamps, merchant identifiers, response codes, BIN groupings, account and device IDs, and approval outcomes. Record who approved each action. This evidence lets the team compare attack volume with approval impact and supports later scheme or merchant reviews.

Compare requests made with the publishable key against normal checkout traffic, then inspect server logs for unusual origins, timing, and endpoint use. A publishable-key abuse pattern points to checkout controls, while secret-key calls from unknown addresses require immediate key rotation and a separate investigation. Don’t treat the two incidents as equivalent.

Notify the acquiring or compliance contact when the event creates scheme-ratio exposure, reaches a reporting threshold, or requires action outside the PSP’s authority. Send the affected period, merchant identifier, evidence used, controls applied, and rollback plan. Keep the report factual, because unsupported claims can complicate the merchant and scheme review.

Yes. A drop in authorization volume doesn’t prove the actor stopped. Attackers can change merchants, devices, BIN ranges, or request timing, and a second wave can follow after the first control. Continue watching approval rates, low-value approvals, decline-code shifts, and related merchants during the agreed monitoring period. This is card testing fraud.

Use a challenge only where correlated signals justify it, rather than placing one on every checkout. Apply risk-based challenges to suspicious combinations of velocity, device, BIN, and decline behavior, then track approval rate and abandonment. EGS can help a client test placement and set rollback thresholds before deployment.

Schedule a Meeting

Book a time that works best for you

You Might Also Like

Discover more insights and articles

A project manager in business casual attire discusses a document intake process with a team in a modern office, focusing on a large monitor.

Bank document intake workflow: A readiness checklist

This article gives you a checkpoint-by-checkpoint method for testing a document intake process before you turn it on. It includes a worksheet structure you can copy into a spreadsheet and the evidence each control needs to produce.

Two business partners collaborate on configuring a POS terminal in a modern office, surrounded by payment devices and colleagues.

EthioPay for Ethiopian Banks: Turn Readiness into a Delivery Plan

This article explains how an Ethiopian bank moves from "we want to participate" to a dated, costed integration plan for the country's domestic card scheme and instant payment rails. It covers service selection and a phased release sequence.

A modern smartphone displaying a mobile banking app lies on a wooden desk, surrounded by abstract payment cards and a notebook.

AfriGO Card Readiness: Plan Your Nigerian Bank’s Rollout

This article is a readiness guide for moving from "we're considering it" to a controlled pilot on Nigeria's domestic card rails. It covers role definition and launch scope, then maps the transaction flow so you can assess systems and assign requirement ownership through the workstream matrix that holds the whole program together.

A professional seated at a desk uses a smartphone and payment card, engaged in online banking in a bright, modern office.

G2P Payments in Africa: Build Reliable Social Payment Infrastructure

This article traces the full lifecycle of a government social payment in Africa, from eligibility approval through to reconciliation and investigation evidence. It assigns operational ownership at each system boundary and sets out how to control exceptions and assisted access.