Pinterest credits itself with more sales than your website analytics does. Start with “what does each report count?” rather than “which system is wrong?” Attribution windows, interaction types, account scope and values can differ.
Conversion Insights covers paid and organic content at business-account level, while ad metrics such as ROAS and CPA appear in Reporting. Record the report name and scope before copying figures into a client file. [1]
1. Specify the number being compared
An attribution window is the period allowing a conversion to connect with an earlier view or click. Record window, interaction type, period, currency and time zone. Define value: does it include discounts and shipping, and how do the business records account for returns?
Distinguish an event received by the platform, a conversion attributed to an interaction and a completed store order. Labeling all three “sales” does not make them the same unit. Show separate layers when explaining differences.
2. Understand modeled conversions
Pinterest explains that some reporting metrics may use modeled conversions to fill signal gaps. Metric tooltips in specified interfaces can indicate whether modeling may be used. [2]
This is not a reason to accept or reject every discrepancy. Do not expect an individual order match for every modeled unit; equally, do not excuse a verified incorrect event value as modeling. Test what can be checked at source and document what depends on reporting methodology.
Practical example: a reconciliation sheet
Complete this hypothetical template with account settings and client definitions.
| Comparison element | Business records | Action | |
|---|---|---|---|
| Unit | Attributed conversion under the settings | Completed order under the definition | Explain the difference before adding |
| Time | Report period and time convention | Order and update times | Align or document the difference |
| Value | Event value under source configuration | Agreed net value | Check discounts, shipping and returns |
| Modeling | Included in some reporting metrics | Not an individual order record | Separate what can be matched individually |
Start with one verifiable row, such as a test transaction’s value and currency, then investigate settings and totals. If value definitions differ, document reconciliation rather than modifying events to force dashboards to match.
3. Compare at a suitable follow-up age
If following a cohort of interactions from one period, review outcomes after equivalent follow-up. Do not compare a new group with another that had longer to buy, then label the difference a decline without checking.
Review product, market and offer mix too. More high-priced products can raise attributed revenue; order counts and margins are needed to understand the commercial result. No single number of days suits every buying cycle.
4. Separate explaining differences from proving additional sales
Event reconciliation tests measurement in specific cases. Attribution reports describe connections between outcomes and interactions. Establishing sales caused by spend beyond what would have happened anyway requires a different, appropriate evaluation.
Deliver each number’s source and definition, explained and unknown differences, next steps and owners. Use the business records for total sales: adding revenue attributed by multiple platforms may count one outcome repeatedly. Explain the limits before recommending more budget.
Official sources
Source documentation reviewed: 16 October 2026. Features and eligibility may change; check the linked documentation before implementation.
