In the age of AI, the importance of testing cannot be overstated.
At TRGT, the team has spent the past few years developing a full-funnel testing framework that allows it to strategically plan tests that will have the biggest impact on KPIs such as new-customer growth.
Ecommerce brands such as Superga, AC Milan, Loop, K-way, and Charles Tyrwhitt have taken advantage of the data-driven agency’s creative testing at scale and its use of Meta AI tools like Advantage+ campaigns.
And, unlike the typical “one size fits all” approach of many other agencies, TRGT says its testing has consistently delivered remarkable results.
So what’s the secret to effective testing? TRGT shares its learnings here…
Top tips for testing AI in Meta
Structured testing is crucial - whether that’s new creative concepts, platform features or campaign settings.
Testing with impact
The tests within meta’s AI features which had the biggest impact were:
- Incrementality. It’s important to understand the incrementality of new campaign types such as Advantage+ on BAU (Business as Usual) activity. TRGT tested this extensively; across most ecommerce brands there was at least a 10% uplift in conversion volume.
- Upper funnel. The impact on conversion volume of running brand and direct response campaigns together was much larger than just running direct response campaigns alone. This was tested in a double cell format (Cell A: DR activity only vs. Cell B: DR activity + brand activity). Combining both types of activity led to 4% lower CPAs (cost per acquisition) overall than running DR activity alone.
- Budget caps for existing customers in Advantage+ campaigns. Measuring in-platform conversion-value data for new versus. existing customers was the basis for this test. In general, having a 5-15% cap on existing customers was the “sweet spot” for most clients, having a positive impact on new customer CPAs.
- Consolidated Advantage+ campaigns. These were tested against Advantage+ campaigns which were split by creative format. Results varied, but generally it was favourable to run a consolidated structure from a CPA and ROAS (return on ad spend) perspective. Clients which ran a consolidated structure versus one split by creative format generally saw 10% lower CPAs overall.
- Dynamic creative overlays. Dynamic creative overlays such as sitelinks and offers can help increase the CTR (click-through rate) and CVR (conversion rate) by giving potential customers several options to click through to from the website, as well as clearly showcasing offers outside of the conventional creative and copy fields.
The Advantage+ in scaling new creative
Many clients have used Advantage+ campaigns as a way to gain faster learnings from newer creatives at scale. One of the most important factors in the performance of TRGT’s most successful Advantage+ campaigns was regular creative refreshes.
Creative lifecycle usually depends on the level of spend, but it’s advisable to upload new creative at least weekly (or multiple times per week if monthly budgets are six figures or above) to fight creative fatigue.
The best creative testing should strike a good balance between large swings (new creative ideas) and iterations of winning concepts. The ratio between those two is a good indicator of the risk you are taking with creative testing, so it’s worth keeping an eye on this.
Creative diversity and testing at scale in the AI age
TRGT has analysed millions of data points across hundreds of accounts, proving a strong positive correlation between the number of creatives launched and the scale that can be achieved. Also creatives do experience fatigue and they provide their best chances of driving high volume of conversions efficiently within the first couple of weeks. With this in mind, it’s essential to have a creative strategy that includes always-on creative testing.
Achieving high-impact creative testing relies on three pillars:
- Research to fuel your creative ideas pipeline.
- Testing to find a good balance between large swings (new creative ideas) and iterations of winning concepts.
- Insights to analyse with data and infer why certain creatives work and others don’t.
AI represents an important layer to support all three pillars above - ranging from leveraging generative AI for briefing and idea generation to in-platform AI optimisations, plus an ever-growing range of tools that can support a cost-effective creative production.
Clients who employed the AI features above, along with a robust overall strategy and a test-and-learn approach saw the greatest success.
With an increasing number of AI features becoming available within the Meta platform, creative is one of the largest levers - if not the largest - that digital marketers have at their disposal.
A strong creative strategy with diverse creative output is more important than ever: it is often the defining factor for whether clients are able to scale up aggressively. Or not.