Separate a creative problem from a campaign problem
An unreadable offer, conflicting price or cropped product can be checked in the exported file. A weak campaign result can also come from audience fit, price, distribution, landing-page friction or measurement errors. Do not assign every loss to the creative after seeing a disappointing dashboard.
A pre-launch review is useful when it identifies something the team can verify and change. It is less useful when it produces a confident explanation with no source moment or suggests edits that the team cannot realistically make.
Build a cost ledger
| Cost or outcome | What to record | Common mistake |
|---|---|---|
| Review | Software/service cost and staff time | Calling review free |
| Revision | Editing, reshoot and approval time | Ignoring production cost |
| Live experiment | Spend, allocation, duration and measurement | Calling every impression on a weaker arm waste |
| Decision | What changed because of the review | Taking credit for changes already planned |
| Result | Comparable campaign evidence | Attributing all movement to the creative edit |
This ledger is more useful than a generic return-on-investment multiplier. It also exposes when a proposed review would cost more than the decision justifies.
Run a bounded pilot
Choose a defined set of upcoming creatives. Record the review before launch, including the recommendation, evidence and expected mechanism. Keep files and versions so you can identify what actually ran.
For each proposed edit, note whether the team acted on it, why, and what else changed in the campaign. A pilot can reveal useful operational outcomes such as fewer unclear briefs or faster agreement. It cannot, by itself, prove that the review caused a change in sales.
Where a causal performance answer matters, use an appropriate controlled experiment with comparable conditions. Read the scoring and A/B testing guide before turning a simple before/after into an effectiveness claim.
A worked decision without invented ROI
Imagine a hypothetical ad that advertises a bundle but shows only one item. The reviewer flags a mismatch. The team adds a clear bundle shot and confirms the final file matches the offer.
That is a completed quality correction. The review can document the mismatch, the edit and the cost. It cannot claim a conversion lift or euros saved without additional evidence. This example is an editorial scenario, not a customer result.
When the recommendation should be to keep the ad
A review should preserve strengths as well as identify problems. If a concern is generic, unsupported or inconsistent with good comparable campaign evidence, do not change the asset just to satisfy a score. Record the disagreement and the decision.
Once testing supplies enough useful evidence, select the creative to back. The objective is a better investment decision, not an endless sequence of rescoring.
Put the evidence in the brief
Use a source timestamp or frame, a clearly stated observation, a proposed change, an owner and a completion criterion. Keep predicted diagnostics, simulated buyer answers and measured outcomes separate.
The RoastIQ report guide explains how to read the diagnostics. Book a demo to discuss a review that fits your campaign and decision.