04
Data Integrity
True Performance vs. Inflated Metrics
The account's "All Conversions" column included add-to-carts, checkout starts, directions, page views, and YouTube engagement — inflating reported ROAS by 2-4x. Here's what the real numbers look like.
| Month | Spend | True Revenue Confirmed Purchases Only | True ROAS | Inflated Revenue All Conv. Value (incl. ATC, engagement) | Inflated ROAS | Inflation |
| January |
$288K |
$539K |
187% |
$1.47M |
510% |
2.7x |
| February |
$208K |
$730K* |
351%* |
$1.94M |
931% |
2.7x |
| March |
$223K |
$614K |
276% |
$1.32M |
591% |
2.1x |
| Q1 |
$719K |
$1.88M |
262% |
$4.72M |
657% |
2.5x |
*February True ROAS of 351% still includes the data import anomaly (Feb 9-13). Corrected for that anomaly, February's estimated true ROAS is ~273%.
Why This Matters
Bid strategies were previously optimizing against inflated signals (up to 9.3x inflation in February). By recalibrating targets to real purchase values only, the algorithm now makes smarter decisions about where to allocate every dollar.