When an A/B setup isn’t feasible — for example, when you want to test overlay impact on an existing high-performing campaign — a Pre vs. Post experiment is the ideal alternative.
In this approach, you keep everything the same and only introduce BigAtom Overlay in the “Post” period — helping measure performance uplift accurately.
Mandatory Pre-Requisites
These conditions ensure results are trustworthy and not influenced by other changes:
Requirement | Why |
Existing ad/campaign must have ≥100 conversions before test | Ensures Meta algorithm is stable (out of learning) |
No audience or budget changes across test window | Avoids delivery variation influencing results |
No pricing updates, discount changes, or major sales events | Prevents commercial impact bias |
Disable Meta Media Enhancements (images & video) | Prevents unwanted automatic creative changes |
Keep minimum 14 days for both Pre & Post periods | Provides statistically valid comparison |
Test Setup Overview
Component | Pre-Period (Before Overlay) | Post-Period (After Overlay) | Why It Matters |
Campaign | Existing active catalog campaign | Same campaign continues | Ensures consistency |
Creative | Default catalog images | Overlay-enabled images | Overlay becomes the only change |
Audience | No changes allowed | Same audience retained | Removes targeting bias |
Budget | No change allowed | Same budget retained | Keeps delivery stable |
Optimization Event | No update | Same event retained | Prevents algorithm shift |
Test Duration & Measurement Window
Stage | Duration | Notes |
Pre-Overlay Period | Minimum 7 days | Baseline data with existing native catalog images |
Post-Overlay Period | Minimum 7 days | Exact same campaign setup — only overlay added |
Total Experiment | 14 days recommended | Ensures clean comparison with no external influence |
For higher accuracy, avoid:
Bid strategy changes
Additional asset uploads
Catalog structural changes
Seasonal peak disruptions (unless comparable YoY)