So, in the name of research, I set about trying to understand this burgeoning technology.
The articles, video clips and websites I discovered all present DMPs through some very compelling features and benefits. And they make lofty (and jargon-riddled) promises, summed up as follows:
Using a DMP can increase the ROI of your PPC and display advertising
DMPs can optimise your customers’ experience on your digital touchpoints
A DMP will unlock insights about your existing customers!
With a DMP you can find new customers that 'look' just like your most valuable existing customers
When you pull it all together, DMPs seem to promise the world. The sceptic in me needed to do some digging; to discover the truth, I got my hands dirty at the coalface – working directly with the technology and asking a few questions along the way.
So – did the promises stack up?
ROI: “more for less”
Magically, DMPs do manage to reduce spend and increase income. The advent of trackable digital channels – and our ability to merge first, second and third-party data – means doing more of what works, and much less of what doesn’t.
Through a DMP, we are able to undertake media planning with far more certainty, across channels, and in real-time. With the ability to store many thousands of ads, the DMP selects and then displays the most relevant and appropriate message to each individual as they encounter the media ‘space’. In turn, the advertiser will merely pay for the time that their creative was on display.
OK. Bear with me here…
A publisher selling digital advertising space will decide which ad to display to a visitor after connecting with your DMP *[via a DSP]. If a client’s customer exists in the DMP (and can be matched), the platform will provide instruction on what message to display – or indeed, not to advertise and waste budget. Likewise, audiences with similar profiles and behaviours to a client’s existing customer base can be served appropriate messaging that is most likely to impact a sale.
*Note: Lesser-newbies will know that the serving of the ad and the bidding process is handled by a Demand Side Platform (DSP). But the rich segmentation and precision targeting is carried out by the DMP, as described above.
Personalisation: prove you know me
Imagine a world where a vegetarian can walk into a burger bar without the risk of being inundated by a meat-laden menu.
Thanks to DMPs, the online equivalent is here – and works very effectively. The burger restaurant’s website contacts its DMP for guidance on which experience is ideal for each particular user. And for our vegetarian visitor, the menu, design and nutritional content can be completely customised and made ‘appropriate’, based on previous behaviour (across channels) and intelligence from similar customers.
For one of our fintech clients, we ensure that their website references a specific credit card offering air miles for some customers and cash back equivalents to others. Different implementation – same approach. Ultimately, the DMP is curating the very best site experiences for individual users.
Enrichment: What don’t I know about my customers?
A few weeks ago I bought a Peppa Pig storybook for my niece. Ever the proud uncle, I announced the fact on my brother’s Facebook wall – and minutes later received a reply from his friend, informing me that she loves In the Night Garden too.
In a way, this exchange was a deconstructed DMP experience.
DMPs leverage, to great effect, the fact that other people (or companies) know information about your customers that you may not. It’s one of the key reasons that DMPs are constructed to ingest and process vast quantities of data. The richer our picture of a customer, the greater their value is to us – and others. One of the foundations of a DMP is the ‘data marketplace’, which underpins many of its features.
In my Peppa Pig experience, I represent the brand, my niece is my customer, Facebook the marketplace and the friend would be a third party data vendor. Through the DMP I can present my customers to the marketplace and purchase supplementary information about them from third party vendors. See my earlier blog post on cookies to understand how this is done anonymously.
Reach: Make me more of these…
We’ve all seen the film where a perfect human specimen is selected for cloning. Invariably someone wants to build a model army or super-race to support world domination. Perfectly Darwinian, the principle is simply to pick the perfect sample and replicate it at scale.
DMP appears to emulate this philosophy, potentially with that same dream of world domination…
Having identified a specific audience within the DMP, it’s possible to 'clone' it through the marketplace (marketers call these ‘lookalikes’). A typical approach might be to isolate the highest spending customers, or perhaps the most frequent converters.
The selection is presented to the marketplace’s pool of third party vendors. Each of the vendors considers our audience and identifies individuals with common traits. Some technical witchcraft takes place here, which basically prevents the need to transfer personal data (this is another instance where cookies come into play).
Vendors will then sell a set of appropriate customers (lookalikes, via cookies) from their base. The DMP can now deliver campaigns to those new cookies, effectively cloning the selection of ideal customers and increasing reach.
Result…I’m a DMP Convert!
DMPs undoubtedly promise some impressive features, and having looked under the bonnet, I’m now convinced that they can genuinely deliver on them too. It’s worth mentioning that the real magic sauce comes from the strategy underpinning your DMP and of course, the quality of the data you’re feeding in.
This technology is going to shake up the role of brands, data and media agencies – from how we work together to how we target. 2016 has been big, but I’ve got a funny feeling 2017 is going to be very exciting indeed.
Justin Watson is solutions consultant at Merkle.