RTB Discrepancies
RTB Discrepancies & Reconciliation
A practical reference for understanding why RTB discrepancies happen, how to reduce them when integrating, and how to reconcile counts using billable-event tracking.
What a discrepancy is
A discrepancy is the gap between what one side counts and what the other counts or bills. Some level is normal in RTB, the goal is to keep it low enough for clean invoicing and clear reporting, not to eliminate it.
| Discrepancy | What it means |
|---|---|
| Under 5% | Acceptable |
| 5–10% | Worth monitoring |
| Over 10% | Worth investigating |
Why discrepancies happen
Some discrepancy is normal in RTB: each platform in the chain, supply partner, demand partner, optimiser, analytics counts independently, with its own filters, so several different numbers for the same activity is expected, particularly for clicks on display formats. The factors that drive the gap are largely client-side:
- Browser navigation optimisation. Modern browsers, especially Chrome on Android, optimise navigation aggressively. When the destination is known early, intermediate redirect hops can be collapsed, so a click or event may never reach the counting server — in which case it can't be counted. This is the single biggest driver of click discrepancies.
- Different invalid-click / multiclick filtering. Each side applies its own invalid-click logic, so totals only line up when compared like-for-like (valid + invalid together).
- Client-side measurement methods. Techniques used to enrich or fire events. Client Hints collection,
sendBeacon, and browser "heavy ad" interventions can each affect whether and how an event is recorded. - Format differences. Click discrepancies tend to run higher on video formats than on display. The most reliable events to reconcile on are impressions for display formats and clicks for click formats.
The structural way to reduce all of this is server-to-server (S2S) counting which removes the dependency on a client-side call firing and, for billable reconciliation, billable-event (BURL) counting.
How to integrate cleanly and reconcile
Detailed guidance lives on the pages below:
- Supply side (publishers): Integrating Publishing Supply best practices for reducing discrepancies when integrating supply.
- Demand side (buyers / DSPs): Integrating Advertising Demand signalling billable events so your counts align with the publisher's.
- Metrics & troubleshooting: Billable Event Counting (BURL) the Pixel / Billable / Served metric model, Served Ratio, and a step-by-step troubleshooting workflow.