1. Creative testing: Is it actually delivering results?
Meta’s Creative Testing tool is designed to help advertisers compare different ads under more controlled conditions. Unlike standard delivery, where Meta can quickly favour the creatives it predicts will perform best, a creative test is intended to give selected ads a more equal, comparable opportunity to generate results. Meta Blueprint also highlights the importance of structured testing and evaluating whether results are conclusive. Meta Blueprint.
However, in my experience, creative testing has not consistently worked as well as we would have hoped for our clients.
We have seen tests run for around two weeks that generated only one or two purchases. That makes it difficult to confidently identify a winning creative, particularly when purchases are already low. Meanwhile, when we introduce several creatives into a standard campaign and allow Meta’s algorithm to optimise delivery, we can sometimes see stronger results without forcing a formal test.
Of course, this does not mean creative testing is ineffective. It means the testing method needs to suit the account’s budget, conversion volume and objectives.
For brands with limited budgets, splitting spend between several test ads can leave each creative with insufficient purchase data. A two-week testing period is not automatically enough to produce a reliable conclusion.
My approach would be to test with a clear hypothesis. Rather than comparing several unrelated ads, I would test one meaningful variable at a time: for example, founder-led video versus product imagery, a strong opening hook versus a softer introduction, or a collection ad versus a single-image ad.
I would also assess leading metrics such as click-through rate, cost per click and add-to-cart activity alongside purchases. These metrics can help explain why a creative is or is not attracting interest, although they should not be mistaken for proof that it will generate profitable sales.
2. Push Delivery to This Ad: Giving promising creatives a fair chance
One of the more interesting developments is Meta’s Push Delivery to This Ad feature.
The problem it addresses is familiar to anyone running campaigns with multiple ads, Meta can direct most of the budget towards one or two creatives, leaving the others with very little delivery. This is understandable from an optimisation perspective, but it can make it difficult to assess a new creative that has barely had an opportunity to perform.
The new feature allows advertisers to dedicate a portion of their budget to a selected ad for a defined period. I see potential here, particularly when we have a creative we genuinely want to evaluate but Meta is not allocating enough spend to it.
For example, a new founder video might communicate a brand’s story particularly well, or a different product angle might appeal to a customer segment that our existing ads are not reaching. If the algorithm consistently prioritises established creatives, Push Delivery could help us gather more evidence about the new ad.
However, I would not use it simply to force spend towards every under-delivered creative. There is usually a reason Meta favours certain ads, and directing budget towards another one does not guarantee better performance.
Instead, I would use this feature selectively, with a clear reason for testing the chosen ad and a defined period for reviewing the results. I would then compare its purchase performance, cost per acquisition and other relevant signals against the existing campaign.
The important distinction is that Push Delivery guarantees an opportunity for delivery, not a winning result.
3. Why I like grouping creatives by theme
Rather than relying entirely on formal creative tests, one approach that has worked well for us is separating ads by creative theme into different ad sets.
For example, an ecommerce account might have separate creative themes for founder storytelling, product-led imagery, customer reviews and collection-focused content. This allows us to structure campaigns around distinct messages and assess their performance more clearly.
I have seen great results with this approach because it gives us a more organised way to understand which creative direction is contributing to performance, rather than placing every concept into one ad set and allowing one or two ads to absorb most of the spend.
That said, splitting ad sets is not a universal solution. Too many ad sets can fragment budgets, duplicate audiences and make it harder for campaigns to accumulate sufficient conversion data. I would use this structure when the themes have a meaningful strategic difference and there is enough budget to support them.
My preference is to keep the structure as simple as possible, introduce genuinely different creative concepts and allow the algorithm to optimise within each group. Where a particular ad is still being overlooked, Push Delivery could become a useful additional tool.
4. Audience controls: More precision, but understand what you are selecting
Another development worth watching is the changing way Meta presents audience settings.
As shown in the Ads Manager interface, advertisers can see optional age and gender selections accompanied by a message explaining that media may also reach other people likely to respond. This distinction matters: an audience selection is not necessarily a strict restriction.
Meta’s Advantage+ audience approach distinguishes between audience controls and audience suggestions. Controls establish boundaries, while suggestions help guide the algorithm but can be expanded beyond when Meta predicts that doing so may improve performance.
In practice, this means selecting women as an audience suggestion does not necessarily mean that only women will see an ad. Similarly, suggested age ranges may guide delivery rather than impose a hard limit.
For ecommerce brands, this flexibility can be useful. A luxury occasionwear brand may want to guide delivery towards older women, while a younger fashion brand may have a different core customer profile. These signals can help Meta understand the intended audience without necessarily preventing it from finding customers outside the suggested demographic.
However, I would be careful not to assume that optional age and gender settings guarantee demographic exclusivity. Where a demographic restriction is commercially essential, advertisers should check the actual campaign and audience-control settings available in their account.
I would also test whether tighter targeting genuinely improves results before restricting delivery unnecessarily. The aim is to help Meta find valuable customers, not to narrow the audience simply because we can.
Final thoughts: Use the tools, but keep the strategy
My takeaway is simple: Meta’s new features are most useful when they solve a specific problem.
Creative testing can help answer focused questions, but low conversion volumes can make its results inconclusive. Push Delivery can give an overlooked creative more exposure, but it cannot make an unsuccessful ad profitable. Audience settings can provide useful guidance, but advertisers need to understand the difference between a suggestion and a genuine restriction.
For our clients, I would continue prioritising strong creative concepts, thoughtful campaign structures and enough conversion data to make informed decisions. I would use separate creative themes where they have demonstrated value, test new features selectively and avoid interfering with algorithmic delivery without a clear reason.
Ultimately, the goal is not to use every new Meta feature. It is to understand which tools help us make better creative decisions and generate more profitable results.
