Write the learning question first

Before launch, state a specific question: does customer proof produce more qualified enquiries than a scenery-led opening? Does a vertical demonstration earn better downstream engagement than a static image? A campaign objective and a creative hypothesis are not the same thing; reach, traffic and leads reward different behaviors.

Specify the business outcome and the confidence needed to act. A video-play winner is not automatically a qualified-lead winner. If CRM outcomes are unavailable, acknowledge the limitation instead of implying that the platform's optimization event measures customer value.

Keep a stable campaign record

Save the campaign dates, audience context, objective, conversion definition, placements, landing page, creative IDs, offer and exclusions. Name assets by concept and format, not final-v6. Preserve original briefs and approvals so later interpretation cannot quietly rewrite the hypothesis.

Log budget changes, paused ads, landing-page edits, availability changes, holidays and tracking repairs. Without that context, a seasonal promotion may appear to prove a creative lesson when the real cause was a sold-out date, a changed offer or an unavailable booking form.

Distinguish experiments from observations

Where volume allows, use a controlled test of one meaningful variable, such as a proof-led versus problem-led opening. Keep other conditions as comparable as practical. Meta offers A/B testing tools, but test design and sufficient signal still matter.

If the audience or budget is too small, label results observational. The delivery system may allocate more impressions to one ad, so the winning asset was not necessarily exposed to equivalent people. A handful of conversions should lead to a follow-up hypothesis, not a universal rule.

Build an asset-level evidence table

For each asset record opening angle, proof, format, aspect ratio, duration, captions, call to action, destination and placement eligibility. Add objective-aligned outcomes: meaningful reach, landing-page views, qualified leads, purchases or CRM stages where available. Include frequency and creative fatigue as context.

Review whether the asset actually fit its placements. Meta's Reels guidance emphasizes vertical video, appropriate audio and key messages within safe zones. A landscape asset obscured by interface elements can underperform because of execution rather than because the core message is weak.

Investigate the whole conversion path

If an itinerary video attracts many visits but few bookings, check page relevance, date availability, form reliability and lead follow-up. If a testimonial produces fewer clicks but more qualified enquiries, the business may prefer it even when platform engagement metrics look less exciting.

Do not change every variable in response to one result. Classify possible explanations into creative, audience, offer, landing page, tracking and operations. The debrief should identify which explanation has evidence and which requires another test.

Example: shoulder-season hospitality

A resort tests three concepts: scenery, guest proof and a practical itinerary. The testimonial yields fewer clicks but a higher share of qualified enquiries. The defensible conclusion is that proof-led messaging deserves another controlled test for high-consideration audiences, not that testimonials always outperform scenery.

If the itinerary ad drives strong interest but booking starts collapse on mobile, the next action is a booking-flow review. Archive the creative with its original context so next season's team can compare like with like.

Turn findings into owned decisions

Use four columns: evidence, confidence, interpretation and next action. A clear production issue can be fixed immediately; a moderately supported creative insight should be retested; a low-volume observation should remain open. Assign owners and deadlines for creative templates, landing-page improvements and measurement changes.

Archive approved claims, usage rights, assets, IDs and test notes in one location. Connect this review to seasonal creative planning, Meta creative testing and CRM feedback. The purpose is a repeatable learning system, not a longer presentation.