Building a Fraud-Resistant Publisher Network: Strategy + Process

Every pay-per-call broker eventually deals with publisher fraud. The question is not whether it happens, but how much damage it does before you catch it and how your network is structured to minimize that damage.

Most operations focus on detection: listen to calls, find the bad ones, take action. Detection matters. But if detection is your only line of defense, you are always playing catch-up. By the time you flag a coached call, the payout is already committed. By the time you identify a fraud pattern, dozens or hundreds of bad calls have already flowed through.

The better approach is structural — building a publisher network where fraud is difficult to execute, expensive to attempt, and caught quickly when it happens. Here is how.

Why Detection Alone Falls Short

Consider the economics from a bad actor's perspective. They sign up as a publisher, send 200 coached calls over two weeks, collect payouts, and disappear before your monthly QC review catches the pattern. Even if you eventually claw back 50% of the fraudulent payouts, they still profit. And they can repeat the process under a new name.

Detection catches the fraud. Structure prevents it from being profitable in the first place.

The goal is not to build an impenetrable system — that does not exist. The goal is to make your network an unattractive target compared to networks with weaker controls. Fraud flows toward the path of least resistance.

Publisher Onboarding: Your First Filter

The onboarding process is where you set the tone for the entire relationship. A rigorous onboarding process deters low-quality publishers before they send a single call.

Vetting before approval. Require basic due diligence: business registration, traffic source descriptions, references from other networks, and a conversation with a real person. This is not foolproof, but it filters out the most opportunistic bad actors. The publishers who cannot or will not provide basic information about their operation are telling you something.

Mandatory test period. Every new publisher starts with a 30-day test period at limited volume — typically 20-30% of their requested cap. During this period, every call gets reviewed. Not sampled. Reviewed. The test period establishes a quality baseline and catches problems before they scale.

Graduated volume increases. After the test period, volume increases are tied to quality metrics. A publisher who maintains a flag rate below 3% and a conversion rate above the campaign average earns more volume. A publisher whose metrics are borderline stays at their current cap until they improve. This is not punitive — it is rational resource allocation.

The publishers who balk at this process are rarely the ones you want in your network.

Setting Quality SLAs That Actually Work

Vague quality expectations produce vague results. Effective quality SLAs are specific, measurable, and tied to consequences.

Define clear thresholds for the metrics that matter:

Maximum flag rate. The percentage of a publisher's calls that trigger QC flags. A reasonable threshold varies by vertical — healthcare campaigns might tolerate a 2% flag rate while home services might accept 5% — but every publisher should have a defined ceiling.

Minimum conversion rate. The percentage of calls that result in a qualified outcome. Publishers whose traffic consistently fails to convert are either sending the wrong audience or sending fabricated calls. Either way, it is a problem.

Minimum average duration. Extremely short calls (under 30-60 seconds depending on the campaign) typically indicate hangups, wrong numbers, or callers who clearly do not match the offer. A publisher whose average duration is significantly below the network average warrants investigation.

Maximum duplicate rate. The same caller numbers appearing repeatedly across a publisher's traffic can indicate a recycled caller pool — a common tactic in fraud operations.

Put these thresholds in writing. Include them in your publisher agreement. Review them quarterly and adjust based on actual network data. The specific numbers matter less than the fact that they exist, are communicated, and are enforced.

Monitoring Cadence: Daily, Weekly, Monthly

Quality monitoring is not a monthly task. It is a continuous process with different activities at different intervals.

Daily: Dashboard review. Check flag counts, flag rates by publisher, and any critical flags (DNC violations, TCPA issues, confirmed coached calls). Critical flags get same-day response. This takes 15-20 minutes if your data is organized.

Weekly: Publisher-level review. Look at each active publisher's metrics for the week: flag rate trend, conversion rate, average duration, volume. Identify publishers whose metrics are moving in the wrong direction. A flag rate that went from 2% to 4% this week might be noise. Two consecutive weeks of increase is a trend that needs a conversation.

Monthly: Scorecard and network health. Compile publisher scorecards with 30-day rolling metrics. Review network-level trends. Identify your top performers and your bottom performers. Make volume allocation and rate decisions based on the data. This is also when you assess whether your quality thresholds need adjustment based on overall network performance.

The discipline is in the cadence. The operations that skip weekly reviews for three weeks and then scramble when a buyer complains are the ones that lose buyer relationships.

Quality-Based Routing and Volume Allocation

Your best publishers should get your best campaigns and your highest volume. This seems obvious, but many operations allocate volume based on relationships or capacity rather than quality data.

Implement a tiered system:

Tier 1 (proven quality): Flag rate consistently below threshold, conversion rate above average, 90+ day track record. These publishers get first access to new campaigns, highest volume caps, and premium rates.

Tier 2 (acceptable quality): Meets quality thresholds but does not exceed them. Standard volume allocation and rates. Monitored on normal cadence.

Tier 3 (on watch): Metrics approaching or occasionally breaching thresholds. Reduced volume cap, increased monitoring frequency, active conversation about improvement.

When publishers see that quality directly determines their volume and rates, quality becomes their problem too. This alignment of incentives is more powerful than any amount of post-hoc enforcement.

Know When to Pause: Your Safety Net

Even with strong onboarding and monitoring, situations arise where you need to stop traffic flow immediately. Having predefined pause triggers in your operations playbook provides that safety net.

Define the triggers ahead of time: flag rate exceeds a critical threshold (e.g., 15% in any 24-hour window), multiple DNC or TCPA flags in a short period, conversion rate drops below a minimum floor, or a burst of calls from the same caller number. When you see these patterns in your QC data, act immediately.

When a trigger fires, pause the publisher's traffic pending review. Not reduced. Paused. Notify the publisher with the specific reason and the data supporting it. Reinstatement should require review and, depending on the trigger, a documented corrective action plan.

Pause rules are not about being heavy-handed. They are about limiting damage. A publisher experiencing a legitimate traffic quality issue benefits from a pause too — better to stop and fix the problem than to keep sending bad calls and accumulate clawbacks.

The Economics of Quality

There is a persistent myth in pay-per-call that cheaper traffic is more profitable because the margins are higher per call. This ignores the true cost of low-quality traffic.

Factor in the full cost: clawbacks on fraudulent calls, buyer disputes that consume operations time, compliance liability from DNC and TCPA violations, lost buyer relationships when quality slips, and the opportunity cost of routing capacity wasted on bad calls.

A publisher sending traffic at $15 per call with a 12% flag rate and a 20% conversion rate is more expensive than a publisher at $22 per call with a 2% flag rate and a 35% conversion rate. The math is not close.

The networks that win in pay-per-call are not the ones with the cheapest traffic. They are the ones whose traffic converts reliably, generates minimal disputes, and keeps buyers coming back.

Building a Reputation That Compounds

A network known for quality standards attracts better publishers and better buyers. Good publishers want to work with networks that do not tolerate fraud because it means they are not competing against cheaters for volume. Good buyers want to work with networks that prove traffic quality because it reduces their risk and their QC overhead.

This reputation compounds over time. Better publishers send cleaner traffic, which improves buyer satisfaction, which attracts more buyers, which gives you more campaigns to offer good publishers. The flywheel works, but only if the quality standards are real and consistently enforced.

Every element of this strategy — onboarding test periods, quality SLAs, daily monitoring, publisher scorecards, pause decisions — depends on one thing: comprehensive, reliable call quality data. Tools like ConvoQC provide that foundation by analyzing every call individually, flagging coached calls and compliance issues, and tying results to the publisher source. The structure makes fraud unprofitable. The data makes the structure possible.