What Happens When You Actually Dispute Every Amazon DSP Billing Error for 90 Days

Most DSP owners dispute billing errors reactively — when they happen to notice one while checking their settlement, or when a deposit is obviously short. What would happen if you caught and disputed every single error, systematically, for 90 days?

That's the experiment. Track every billing category across every weekly settlement. Cross-reference every data source. File disputes for every discrepancy, no matter how small. Track the outcomes. See what the numbers actually look like when nothing slips through the cracks.

The Setup

A systematic 90-day dispute process requires four things running simultaneously:

1. Complete data capture: Every settlement report, every WST snapshot (captured before it refreshes), every Payee Central deposit record, and every fleet management record for the period. No gaps.
2. Cross-referencing across all 13 billing categories: Not just checking the big-ticket items (variable pay, incentive pay) but also verifying technology fees, damage charges, AFS charges, training costs, and every other deduction line item.
3. Dispute filing for every discrepancy: Regardless of size. A $50 technology fee overcharge gets the same treatment as a $250 dropped route. The goal is to understand the full landscape of errors, not just the obvious ones.
4. Outcome tracking: Every dispute logged with the date filed, category, dollar amount, and result — accepted, denied, pending, or expired.

What We're Measuring

Early Patterns

Based on what we're observing so far across DSP portfolios running this kind of systematic reconciliation, a few patterns are emerging:

Dropped routes are the highest-value individual errors

At $180-250 per route, a single dropped route recovery exceeds most other individual discrepancies combined. They also tend to have the highest dispute acceptance rate — when WST clearly shows a completed route that the settlement doesn't credit, the evidence is hard to argue with.

Deposit mismatches are the most frequent but lowest individual value

Small deposit-to-settlement differences ($50-500) appear nearly every period. Many turn out to be legitimate post-settlement adjustments. But a meaningful percentage are genuine errors — deductions applied without corresponding events, or rounding that consistently goes in one direction.

Small deduction errors compound significantly

A $30 technology fee overcharge doesn't feel worth disputing. But when it appears every period for 12 periods, that's $360. Across five deduction categories with similar small errors, you're looking at $1,500-2,000 per year in charges that individually seem insignificant.

Acceptance rates vary by category

Early indicators suggest that route-related disputes (dropped routes, variable pay miscalculations) have higher acceptance rates than deduction-related disputes (damage charges, AFS charges). This makes sense — route data is more black-and-white (the route was either completed or it wasn't), while deduction disputes involve more subjective judgment about fault or applicability.

The first 30 days surface the most errors

The initial period of systematic checking tends to uncover a backlog of errors from previous periods that were never caught. After the backlog clears, the per-week error discovery rate stabilizes at a lower (but still meaningful) level. This is the backlog effect — accumulated errors from months of spot-checking vs. systematic checking.

Why 90 Days

Shorter windows don't give you enough data to see patterns. A single week might have zero errors or five — you need multiple billing cycles to separate signal from noise.

• Days 1-30: Backlog clearing. This is when you catch accumulated errors from previous periods. Recovery numbers are inflated by the catch-up effect.
• Days 31-60: Normalization. The per-week error rate stabilizes. You start to see which categories are consistently problematic and which were one-time issues.
• Days 61-90: Steady state. This is the true run rate — the recoverable amount per period when you're catching errors in real time rather than retroactively.

The 90-day window also covers roughly one full quarter of Amazon's billing cycle, which captures seasonal variation, scorecard tier changes, fleet adjustments, and other factors that affect error rates.

What We're Learning

The clearest takeaway so far: systematic reconciliation finds money that spot-checking doesn't. That isn't surprising — it's what you'd expect when you go from checking 3-4 billing categories occasionally to checking all 13 every period.

What is surprising is where the money hides. It's not primarily in the large, obvious errors (though those exist). It's in the accumulation of small, persistent discrepancies across multiple categories — errors that are individually below the threshold of attention but collectively significant.

We'll update this article with the full 90-day results, including total recovery by category, dispute acceptance rates, and the normalized per-period recovery rate. In the meantime, the early data supports a simple conclusion: the errors are there, and they add up. The only question is whether you're catching them.