Unless you live under a rock (or have absolutely nothing to do with Meta), you have probably heard the phrase ‘Creative is the new targeting.’ It isn’t just a buzz phrase, though. Meta has rolled out a long list of updates to their algorithms and campaign architecture, so in most cases creative is now the primary targeting lever on Meta, as Advantage+ and algorithms handle audience matching.
So with such a big stake in the game, how should you handle creative testing? This blog explores why random testing wastes budget and why a repeatable framework is needed to give you the best chance at performance.
Move from Minor Tweaks to Testing Concepts
Testing a different font, or a slight iteration in copy, will barely give you the revelations you’re looking for. Meta’s system scans all of your ads and rates them based on how similar they are to each other – and if the difference really is as small as a change in font, it will classify these 2 creatives as identical and deprioritise one of them from showing in the auction, which is how you end up with ads getting pennies next to their name when it comes to delivery, not even giving your brand new assets a designer has spent time putting together a chance to perform.
What would be more sensible in this case would be to outline 3-4 distinct concept types to test first. For example, UGC vs. polished video, problem-focused vs. solution-focused, us vs. them comparisons:
- UGC vs. Polished Studio Video: Test raw, authentic, direct-to-camera UGC against high-production commercial video to see what level of brand polish your audience responds to.
- Problem-Focused vs. Solution-Focused Messaging: Lead with a specific pain point your target audience experiences versus leading directly with the end benefits and visual transformation your product provides.
- Us vs. Them Comparisons: Visually contrast your product or service directly against traditional market alternatives or competitor shortcomings.
- Static Graphics vs. Dynamic Carousels: Compare single-image feature callouts or press quotes against multi-card carousels that walk through a feature breakdown or customer story step-by-step.
The learnings from a test where the hypothesis is clear are miles more valuable than an ambiguous mix of creatives running side by side just for the sake of it.
Separate Your Testing from BAU Campaigns
What happens when new ads get introduced to an already existing campaign that already has a bank of learnings behind it? The new ads don’t have any learnings attached to them, but they are now in the bid alongside some of our top performers. This will once again make the chance of delivery to the newer ads less likely. Additionally, you run the risk of resetting the full campaign learnings, which would have an impact on overall account performance.
To prevent algorithmic bias, consider splitting your budget across your proven scaling campaigns like ASC and dedicating up to 30% to a dedicated testing campaign. If budget is really tight, you could also consider utilising the Meta creative testing feature at the ad level, where you can test up to 5/7 creatives in an already existing campaign without running them against the already top-performing creatives.
Track Creative Performance Across the Entire Funnel
Different users require a different message depending on what stage of the marketing funnel they are currently at, which is why there is a need for differentiation. It is also important that these different types of creatives are measured based on the appropriate metrics and not solely on immediate ROAS or CPA.
Break down how you evaluate your creative:
- Hook Rate (3-second view): Did the visual stop the scroll?
- Hold Rate/ ThruPlay: Did the message keep them watching?
- Outbound CTR: Did the offer drive action?
- CPA/ ROAS: Is it profitable enough to graduate?
For a deeper look into structuring assets across every touchpoint, read Embryo’s insights on matching creative to the customer journey.
Set Clear Graduation, Iteration, and Kill Rules
As already mentioned above, having a clear hypothesis is arguably the most important part of creative testing, so we need a strict protocol for winning, near-winning, and losing ads.
Before the creative brief is created and built into the account, define what qualifies an ad for “graduation” to the main scaling campaigns (e.g., hitting 10+ purchases at target CPA) – this is probably the simplest step. The part where most people get confused is when it comes to the near-winners. For example, ads that deliver a high hook rate but low CTR might just need a change in copy or landing page before you decide if the concept needs to be killed. Setting some guardrails and being fully involved in the creative production from the start is the way to go.
Frequently Asked Questions (FAQs)
- How many creative variations should I test at once? 5–10 per week depending on budget; don’t test too many variations at the same time to avoid budget fragmentation. Every asset needs enough budget to reach a reliable statistical threshold before you draw conclusions.
- How long should a test run before turning an ad off? Base decisions on spend and impression volume thresholds, not a fixed calendar timeframe. For example, reaching 1.5x target CPA or 2,000 impressions instead of simply 3 days. High budget accounts may reach data significance in a day, whereas lower budget accounts may take longer.
- Does native UGC beat high-production video? Unpolished, native content often outperforms polished studio ads because it blends naturally into feeds. It reduces instant ad fatigue and feels more like a genuine peer recommendation.
- How often should creative be refreshed in scaling campaigns? Metrics such as rising frequency, declining CTR, and increasing CPA point to ad fatigue. When you start noticing this trend, introduce newly graduated winners from your testing activity to maintain steady account performance.
To conclude, when structured creative testing is executed correctly, it has the power to turn paid social from a gamble into a predictable growth engine. Start by isolating your tests, defining clear hook-to-conversion metrics, running true concept-level variations, and maintaining a strict graduation framework.
Ready to level up your social performance? Explore how Embryo’s bespoke paid social media strategy can help your brand scale profitably.
