Getting The DMP Integration Right

dominic-saturData-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.

Today’s column is written by Dominic Satur, US platform director at Flashtalking.

Data-driven marketing is only as strong as the ability to ingest, model and apply the data at our fingertips. But as adoption of data-management platforms (DMP) becomes widespread among advertisers, one of the biggest challenges has been integrating this valuable, granular data into other platforms for media buying or creative personalization.

Early industry efforts have focused on server-to-server cookie mapping, where two platforms match their cookie pools through pixel firing. It’s not a perfect solution. The resulting match tables allow for segment data to be transferred between platforms, imported and onboarded to the partner. There are products on the market that can turbo charge this matching process through their broad matching reach, but they still only expect match rates of about 70%.

The other stumbling block to this approach is time lag. Many platforms offer daily log file transfer, but once onboarding times are factored, it still could be 36 hours before the data can be used. In today’s marketplace, speed to market is of the essence.

Given the demands of a fast-paced marketplace and the desire for the most precise possible data-driven marketing, I believe that it’s critical to explore alternative approaches to more quickly integrate and use as much DMP segment data as possible.

For perspective, you could think of the unmatched portions in terms of targeting waste: Since a third of an advertiser’s audience goes unmatched, this large portion of an advertiser’s audience gets zero message personalization. Add to that the lag time of waiting more than 24 hours to match the other two-thirds of their audience, and it’s clear that cookie mapping neglects an extraordinary amount of opportunity.

After much experimentation, I’ve discovered that today’s technology allows an earlier option in the process: a pre-impression request call to a DMP partner. In this scenario, user-level segment membership is passed back to the ad server, which essentially enables the identification of an audience member’s DMP profile in real time. This in turn allows for dynamic creative decisions to deliver a hypertargeted message using the most up-to-date information for 100% of a client’s DMP cookie universe.

The critical component is that nothing is being logged, just decisioned, leaving no residual drop-off from unmatched cookies. In the end, this means that every customer in the advertiser’s cookie pool will get a personalized message.

Sometimes it’s a matter of being willing to move beyond convention. This progressive systemic approach offers perfect matching rates in real time and reduces data leakage because there is no sensitive first-party data being stored by the ad server. In my experience, targeting is vastly improved, which enables more agile messaging, with practically no waste. And that is the optimal state of effectiveness and efficiency.

Follow Flashtalking (@flashtalking) and AdExchanger (@adexchanger) on Twitter.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!


  1. Regarding pre-impressions, does it mean tight integration between DMP and AdServer so that all information with metadata from DMP (segments) is available in AdServer for targeting, etc? Or approach is different?

  2. I think it’s important to call out that both the match rate and time lag issues are avoidable with current technologies in market. For example, Turn has an integrated DMP/DSP that has no cookie drop off and no time lag to execute on segments when used together. Technologies like this are built to avoid these traditional problems of a fragmented marketing/advertising technology stack.

  3. Mr. Satur seems to consider the current rate of 70 percent accuracy a drawback. But that means that reaching the right target viewer seven times out of ten isn’t good enough. How many marketers consider that unacceptable? Perhaps the age of the data overall is the problem to focus on now.

    • One third being lost, and the other two thirds being up to 36 hours old is the drawback I hear Dominic speaking to and he’s saying we can do better. I don’t know about you, but challenging the status quo and striving for excellence is what I love to see in any industry.

  4. What about combining mobile user data? About 52% of a consumer’s time is spent in-app and I would imagine that must add value being able to sync that for a more personalized message.

  5. Mr. Satur, Do any ad servers support the above paradigm? I’m wondering based on your comment “Sometimes it’s a matter of being willing to move beyond convention.”