On this page
PB shows fewer clicks than the provider dashboard. A creative appears under a title that was changed yesterday. A conversions field displays zero. None of those observations, by itself, establishes that the ad needs new copy.
First align the reporting scope and verify what was imported. A mismatch can concern dates, account selection, currency, missing snapshots or creative linkage. Diagnose that comparison before making a performance decision.
Match the account and reporting unit
Record the provider account, external ad identifiers and grouping used in both views. A workspace summary may include more than one connected account, while a provider screen may show a single campaign or ad group.
Check whether the report includes unlinked ads, archived items or only the displayed creative groups. A visible table is not automatically the entire reporting population.
State whether you are comparing impressions, clicks or spend. Those measures describe different events. Neither a click nor a recorded impression is an organic brand mention in an AI answer.
Align exact time windows
Preserve the start and end boundaries, the time zone and the time of comparison. “Last seven days” can mean a rolling cutoff in one view and calendar days in another.
Inspect the reporting windows attached to imported data where available. A snapshot can cover a period that crosses the selected cutoff. Do not assume a date filter splits an aggregate into exact hourly or daily portions.
If the comparison cannot be aligned precisely, describe the limitation and compare the strongest common scope. Do not manufacture missing daily counts by evenly distributing a larger total.
Verify import coverage
A connected account or recent connection-sync timestamp does not necessarily establish that performance data was imported. Check the actual import timestamp and the coverage of the relevant reporting window.
If the import is missing or failed, a displayed zero may represent unavailable evidence. Preserve the failure and investigate the import before calling it zero delivery.
In PB's reviewed flow, imported performance is aggregate account evidence. It does not identify the exact customer prompt that led to an impression or provide automatic attribution to an organic visibility observation.
Work through an unequal-window example
Illustrative example only, using invented counts:
| Source | Covered period | Clicks |
|---|---|---|
| PB imported snapshot | June 1–6 | 12 |
| Provider comparison | June 1–7 | 15 |
| Provider June 1–6 subset | June 1–6 | 12 |
The first two rows differ because the windows differ. Once the same period is compared, the observed click totals agree. This example establishes no general import delay or expected discrepancy size.
For a real mismatch, preserve the source export or view that supports the matched total. If the numbers still disagree, investigate the remaining scope and data questions rather than declaring the problem resolved.
Keep currency attached to spend
Confirm the currency of each account and reporting value. An unlabeled number should not automatically be described as dollars. Do not add values from different currencies without a defined, disclosed conversion method.
Use impressions and clicks only within their aligned population as well. Ratios derived from mismatched counts can look precise while describing no real reporting scope.
The practical first step is reconciliation, not optimization. Establish what was measured before comparing cost or performance.
Check the creative link and version
Compare external ad IDs and the content that was actually handed off. The current PB creative title may differ from the version used externally.
In the reviewed reporting implementation, results can be grouped by creative ID while displaying the current title. That does not prove that the current version earned the historical results. Text and destination matches can also be ambiguous when several drafts share the same values.
Preserve the handoff record and ask the authorized owner to resolve uncertain linkage. Avoid changing or rebuilding history simply to make a familiar name appear in the table.
Treat unavailable conversions as unavailable
The reviewed PB import flow collects impressions, clicks and spend but does not establish measured conversion data. A displayed zero in its conversion field should not be interpreted as evidence of no conversions.
Do not calculate conversion rate, acquisition cost or return on ad spend from that unsupported field. If verified conversion evidence exists elsewhere, report it separately with its own method and scope. Recheck the current integration before relying on newly available measurements.
Escalate a precise remaining discrepancy
A useful support packet contains the matched account, ad references, time boundaries, currency, import time, expected values and actual values. Redact credentials and private asset tokens.
Identify the unresolved question: missing import, overlapping windows, incomplete population or uncertain linkage. Any resync, mapping correction or account change should follow the authorized workflow; the investigation itself does not require altering campaign delivery.
Once the evidence is comparable, decide whether it supports a creative review. Until then, changing copy risks responding to a reporting problem as if it were a message problem.