Increase ad performance with creative data and AI

November 15, 2019

Very few organisations are actually getting real value from their marketing data

As our partner, RAPP’s Tania Feeley and I told the audience at this year’s MAD//Festvery few organisations are actually getting real value from their marketing data. So if you thought your business was the only one making a hash of it, you’re not alone. 

Back in the late 1800s – and long before the obsession with data began – John Wanamaker famously said: “Half of my marketing doesn’t work. I just don’t know which half”. 

This quote will still resonate with a lot of marketers today. But most will know  that expecting half of their marketing to work is incredibly ambitious.

Ad CTRs today hover below the 1% mark, meaning that it’s more like 99% of our marketing that isn’t working. And with advertising spend set to top $560bn this year in the US alone, there’s a lot of money that could be put to better use.

Have we only been looking at half the story?

In the midst of the ‘data boom’, we’ve been so myopically focused on what data tells us and the efficiency data brings that we’ve reduced creativity to the lowest common denominator

In Orlando Woods’ book Lemon (described as a repair manual to reverse the crisis in creative effectiveness) he analyses 620 ads that appeared during Coronation Street between 2004–18. This analysis revealed a decline in the use of characters, wordplay, double meaning and metaphors. This indicated that advertising is becoming less self-aware, and containing fewer cultural references and stories. In other words, we’re becoming less creative in our marketing campaigns because we’re afraid about what the data is going to say if we take those risks and they don’t pay off when it comes to ad performance. 

There needs to be a way for data to work intimately with creativity

Fear of risk-taking is the opposite of what marketing is supposed to be all about. There needs to be a way for data to work intimately with creativity.

Creative is data too

Because the creative is what makes us click. The images, the layout, the offers, the copy, the colours… these are all features that catch our attention. And despite what your Creative Director might tell you, there isn’t a perfect creative solution. If there was, we wouldn’t still be dealing in single figure response rates and increasingly plunging ad performance despite all our clever audience analytics and data-driven strategies. 

Those images, layouts, offers, copy and colours can all be boiled down to data too. And understanding the features present in each creative, and using them hand-in-hand with the data insights we can gather from campaigns, gives us access to the ‘secret’ to successful marketing.

Using data and AI to find what really makes us click

Combining behavioural insights from past campaigns with the creative data of each of our creatives allows us to predict what ads are going to have a higher performance. This is a process we at Datasine call ‘semantic content analysis’, and it’s something every marketer can do themselves by identifying the features in each ad and seeing how these correlate to the performance of those ads. 

Using semantic content analysis means putting creativity at the very heart of the marketing process. With data-backed insights into what types of content work and why marketers can take risks and innovate  – two factors that are behind the very best campaigns and the highest conversions.

But the more ads you have, the harder it is to scale this analysis. We need something that can perform these tasks in moments: AI

AI can step in to analyse the features in the image and relate them to performance much faster than a human can with much greater efficiency.  

Semantic content analysis is something our flagship product, Datasine Connect, is designed to do. The AI platform uses Facebook Ads engagement data to rank and rate the creatives marketers upload based on data-driven predictions about their success, saving time, money and resources for marketers while increasing conversions by 33% and more