A green transfer can contain incomplete data
Illustrative scenario: a client report should show the whole week. The import finishes, but Wednesday is missing. The chart still looks convincing. Coupler documents imports that finish successfully with truncated data because of a limit. Transfer status does not replace result checks.
We use Google Sheets and one fictional advertising account with one row per day. First define what a row means in your actual report; a different structure needs different checks. The example’s advertising events are not unique people or sales.
1. Define the expected report
Download the complete example data, a version missing Wednesday, and the review record. Import each dataset into a separate sheet or spreadsheet using File → Import → Upload.
Expect client DEMO-A, account ADS-01, September 21–27, 2026, one currency USD, seven rows, $700 in spend, and 56 events. These are constructed test values. For your actual report, obtain control totals separately from the source; summing the same import is not independent evidence of completeness.
2. Check identity and the full period
Compare client, account, currency, and date on every row. Sort by date and inspect the seven expected days. The faulty dataset contains six: Wednesday is absent. A label saying “current export” does not resolve that gap.
For larger data, use an expected list of account, date, and campaign combinations. Seven rows is valid only for this simple model. If the source legitimately omits inactive days, verify that rule. Do not automatically interpret absence as zero.
3. Identify duplicates before calculating
The example’s key is client + account + date. Use Data → Data cleanup → Remove duplicates on a working copy only, selecting those columns for comparison. Record repeated keys first. Do not silently fix a client report by deleting duplicates: different rows for one date may indicate an incomplete key, such as an omitted campaign field.
A duplicate in this model should hold publication while you investigate. Retain the original import in restricted internal storage.
4. Calculate totals and the ratio
In the complete dataset, spend is column E and events column F, rows 2–8. Enter =SUM(E2:E8) and =SUM(F2:F8) into empty cells. Expect 700 and 56. The faulty dataset totals 600 and 48.
The formula uses English function names and commas for a US spreadsheet locale. Other locales may use semicolons. Check File → Settings for locale and function-language settings in Google’s instructions. IF selects a result based on a condition.
Calculate the ratio of totals. For example, two days spending $100 each with one and ten events give 200 ÷ 11 ≈ $18.18 per event. Averaging their daily ratios gives (100 + 10) ÷ 2 = $55, which answers a different question and is not the cost per event over the whole period.
5. Decide whether to release
Send the report after identity, period, keys, and independent controls agree. Hold unknown results, assign a responsible person, and set the next review. After repairing the export, recheck the entire result, not just one changed cell.
Give the client a separate file with their own data. Open it using the recipient’s account or a verified recipient role. Google warns that protected cells are not a security measure; filters and hidden tabs do not separate clients.
Documentation reviewed October 1, 2026. The datasets and calculations are our illustrative example, not real advertising performance. Add independent deadline monitoring for an entirely missing report.