Learning how to put your trust in data
Businesses still don’t trust digital. It’s fair enough, when you consider all the undisclosed rebates, the deliberate complexity, the lack of transparency, the ongoing problems of viewability, ad fraud, brand safety.
So when a brand is told by their partner that they should change their targeting strategy based on ‘what the data is saying’, it’s understandable why they might be reluctant. Especially if they aren’t able to view this data, or the agency can’t explain their reasoning in simple terms.
Still, this general distrust of digital can be counter-productive when it comes to running a campaign. The whole point of something like programmatic display, or AdWords, is that you can optimise a campaign over time, using performance data and micro A/B testing to gradually improve the effectiveness of your ads.
Examples from programmatic display
Programmatic is leading the way in terms of access to data, and having platforms that enable continual analysis and adaptation.
We had a finance client, for example, that was sure its audience were millennials with relatively low or modest incomes. Yet the data revealed that one of their highest performing audiences were actually people who were interested in high-end cars - very different to the type of person they expected.
A retail client believed their product appealed equally to both genders, but the evidence showed a considerable difference in favour of women. Based on this they adapted their previously unisex creatives to be aimed at women, and saw a major uplift in conversions.
The latest marketing news and insights straight to your inbox.
Get the best of The Drum by choosing from a series of great email briefings, whether that’s daily news, weekly recaps or deep dives into media or creativity.Sign up
Sometimes data tells marketers things they simply wouldn't have considered. A client of ours learned that serving ads on the first two weekdays of the month had a massive positive impact on CPA. Another client learned their best performing audience were people living in semi-detached houses in the north of England, for which there wasn’t really any logical explanation. Still, by building this audience into their targeting, there was a massive increase in conversions.
Not all data insights lead to recommended strategy change. If art lovers happen to also love hot desserts, this doesn't necessarily mean you should target for this. It could be more a reflection of their interest in Jamie Oliver, a characteristic anyone who has run programmatic campaigns before will know is pretty much true for any audience - don't know why, but everyone seems to love that guy.
How to use data to improve targeting
Analysing data accurately is a skill in itself, requiring creativity and a methodical approach.
Start with what you’ve got
On top of whatever 1st party data you have, you can add audiences from 3rd party data providers, selected according to the characteristics you might expect your customers to possess. DoubleClick allows you to also perform an audience composition report - which works out the match ratio between 1st party data with 3rd party audiences - to further enhance targeting.
Analyse the results
The audiences you initially target for are very rarely the most high performing. With a DSP such as DoubleClick, you’re able to create an audience performance report, which compares the targeted audiences with the ones you actually reached. Based on the millions of targeting options - in the market for, interests, recent browsing behaviour, demographics - you can identify the most suitable for your campaign.
Test and test again
It’s true that data can be misleading, which is why you need to apply a certain scientific rigour to how you use it. All the targeting types at your disposal – audience, keyword, category, website – can feed into each other, until you find the optimal targeting options. Thousands of different targeting strategies can be A/B tested rapidly using automation tools, the reliability of data can be measured according to its performance. By the end, there should be a high degree of certainty in the efficacy of the new targeting strategy.
Current DSPs make the practical aspect of this easy. The difficulty lies in gaining the assent of the marketing team. There is a lack of faith in the insights of data analysis because it is often counter-intuitive, or at least contradictory to the convictions of the brand. But so long as there is trust in the agency running the campaign, marketers ought to take a leap of faith here in the name of greater relevance, and of course greater performance.
GDPR and ISBA’s latest call for accountable media will help to rebuild the trust that is so lacking towards digital at the moment. Beyond greater regulation, though, there also needs to be greater understanding of the value of data-driven advertising. It is up to media agencies to be more willing to educate, more transparent about how they operate, and more open in general about how they use data - it’s cool, after all, and the main way they add value.
When we have a more transparent industry, the awesome results of digital will hopefully start to speak for themselves.
Dan Gilbert is founder and chief executive officer of Brainlabs
Content by The Drum Network member:
Brainlabs was founded in 2012 by former Googler Daniel Gilbert. After training some of the UK's top agencies, he realised they weren't very good. They lacked the kind of technical and mathematical expertise required to run successful PPC campaigns. So he left, and started recruiting the smartest people he could find to work for the most fun company he could create. And so Brainlabs was born.Find out more