skip to main content

Optimising online advertising using your offline infrastructure

  • Jed Mole

    Jed Mole

Created at December 2nd, 2014

Optimising online advertising using your offline infrastructure

Like many people, I visited the poppy installation at the Tower of London. After explaining that each of the poppies represents one British or commonwealth soldier who died, my daughter thought about this and then asked (in the simplistic way that kids do) “How come there were 2 world wars, didn’t they learn from the first?” In a similar way, but on a completely different level, are we all guilty at times of ploughing into something new without either thinking about past lessons or making best use of what we already have? In the world of Data-Driven Marketing there is a whole raft of examples, such as optimising online advertising just using digital assets. But by focusing only on the digital elements, are we overlooking all the rich data that we have, and all that we know about our customers and target prospects?

Many organisations have a customer segmentation process or tool, either based on known customers or based on primary research. These segmentations, which have often been honed over the years at much cost, are widely used in product and marketing strategy as well as CRM activities. So far, not many people are able to successfully use them in their online advertising. Often the only way of integrating the two worlds is to produce a pen portrait for the media buyers, who then try to match those profiles to information provided by publishers. But how many Facebook ‘likes’ are there of the necessary brands, or expressions of interest in particular activities? This leads to the “scale vs targeting” trade-off; one that marketers are unfortunately all too familiar with.

However, it is possible to use your customer segments and prospect targeting in the online world without starting from scratch modelling on digital assets. The key is having a link between the online and offline worlds and then being able to build strong attribution models through that link in a compliant way. This is possible if the mapping and linking method use a safe haven containing sufficient descriptive characteristics to be able to identify your customer segments. A safe haven is a matching environment that links online identities to real individuals. Real names and addresses can then be matched to traditional lifestyle database.

The process is to take a sample of customers, or the research base, and match to the descriptive data (for example Acxiom’s InfoBase data) within the safe haven. The output data for modelling is anonymised for privacy compliance.

There are different ways you can build customer attribution models, so it is important to consider whether overlapping or separate segments are appropriate. Should individuals be in only one segment? Do you want people that best fit each segment, or the segment that best fits individuals? In some cases this is not important, but can be if you have multiple products across segments. The former is the basis for selecting people for campaigns. The latter is for selecting the offer for an individual.

In either case, the models are then applied to complete data asset in the safe haven. This will provide national, or international, coverage for your segments across major publishers. This accurately extends the reach of your segments to the required scale for online marketing. Implementation then becomes easier and more predictable. Media buyers can now simply select specifics segments on Facebook, Twitter and the like. Acxiom’s data assets and safe haven span countries across the globe making this approach suitable for national and global brands, enabling you to reuse the investment you’ve made in understanding your customers to drive better customer experiences and better marketing results.

Steve Kingston – Head of Analytics, Europe, Acxiom