top of page
  • Chris Boyd

Are you using Split Testing yet?

Facebook A/B testing, or 'split testing', can be your best pre-campaign strategist, so why don’t more marketing managers use it?


As a marketing manager, at some point you will no doubt have experienced the dilemma of what strategy to base your next campaign on. You may have your ad agency strategist with their extensive PowerPoint decks advising you on one direction. The agency creative director may favour another. You have you own research, your instincts and experience guiding you yet another way. It can be a tough call to commit your marketing budget on a hunch, creative instincts or a PowerPoint presentation, no matter how convincing the content. US General McMaster said of PowerPoint presentations: “It’s dangerous because it can create the illusion of understanding and the illusion of control”

The key message

Not much can replace the magic and power of a persuasive creative idea, but for many campaigns on social and digital channels, getting in and out quickly with a message that targets the main issue for consumers can the best approach. But what is the key message for your brand or product?


Thankfully, we no longer have to guess or commit based on our instincts or those of your ad agency. Facebook (and Google) provides us with a comparatively cheap, quick and direct way to gauge which consumer message is the most effective – split testing.


The mechanism

With split testing we can use the same imagery, varying only the headline and supporting text to work out which thought or product message appeals to consumers the most. Or vary the imagery and keep the headline the same, but for now I'm focussing this article on the text, as this often contains the core thought and message about your brand or products' unique selling proposition (USP).


In its most basic form, we could start with six different headlines for example, placed over the same image. Using Facebook’s automated delivery, their AI machine learning algorithm tests the ad-sets, eventually narrowing down the headline or key thought that generated most viewer engagement.


You can then take that headline (or headlines) which can also be the strategic thought, and use it as the basis for the main campaign on social or broadcast media using higher quality assets like video. The best part is there's no guessing or hunches required. So why don’t more marketing managers use it?

A example of a more detailed split testing road map:


What’s stopping us?

Well, it can legitimately be a time and money issue. You may only have budget for one burst of activity this quarter. Or you may have pressure from above to ‘get something out quickly’ and show results. Plus, once momentum starts, it can be difficult to say 'we’re just going to hold off for three weeks to run a pre-campaign test'. Some clients op for producing multiple high quality assets to begin with and use those to work out which one creates the best response, though often run out of budget to re-use even the successful assets. Those strategies can waste valuable time and money, and still not create the desired increase in sales or new customers. It can also increase the overall cost-per-thousand-impressions (CPM ) of the campaign when the less successful assets are factored in. Usually, once a campaign is completed, clients won’t revisit it for time or budget reasons.

Conclusion

My advice is to start just another two or three weeks earlier than your current planned campaign production start date, and allocate only around 7% of your budget to pre-campaign split testing. The test ad-sets can be created relatively cheaply using your existing still image assets, then working with a great copywriter and yourself to develop and select the headlines or strategic thoughts to be tested.


The results can be truly surprising and revelatory. Once you get into the mindset of split testing and adjusting your timelines appropriately, I doubt you’ll go back to the old way of developing strategy for online or perhaps even broadcast campaigns.



Comments


bottom of page