MarTech Landscape: What are Creative Management Platforms (CMPs) and Dynamic Creative Optimization (DCO)?

A screen from creative management platform Thunder

In digital advertising, ecosystems for programmatic bidding, targeting and delivery should mean that the days of one-ad-fits-all are over.

But, according to an AppNexus/Thunder study, 97 percent of campaigns don’t have a unique creative for each targeted placement. Apparently, it’s not unusual for a campaign to consist of a relatively few ad creatives, repurposed for different sizes.

Which seems like a complete waste. After all, if you can tell that a digital ad will be seen by a business traveler in a New York City high-end hotel on a weekday night at dinner time in winter, why not deliver an ad for that user’s type and needs?

What they are

To generate all the needed ad versions, two main platforms have evolved. One is called Dynamic Creative Optimization, known to friends as DCO. The other: Creative Management Platforms, aka CMP. In this installment of Marketing Land’s MarTech Landscape Series, we explain what those terms mean and what they can do for marketers.

In DCO, a variety of ad components — such as main body backgrounds, main images, overlay text, and so on — are dynamically assembled when they are served, according to the particular needs of the impression. Vendors include Adacado, Admotion, Adobe, Celtra and Sizmek.

So, an impression targeted at males 18–34 who live on the West Coast of the US and follow soccer might receive an ad built for that segment, in real time.

Here’s a demo of dynamically assembled ads from ad tech firm Sizmek, which offers a DCO. On that screen, click on the links in the upper left.

While DCOs can assemble versions that are built on the fly, its sibling, CMP, offers a computer-aided tool so creative staff can prebuild lots of ads for specific segments.

So, an ad impression targeted at males 18–34 on the West Coast of the US who follow soccer can be filled by an ad — out of dozens or hundreds created for the campaign — that was previously built for that audience segment with a CMP. A CMP — like Flite, Spongcell or Thunder — gives superpowers to ad creators, so they can churn them out far faster than a normal mortal.

The platform might employ machine learning, automated picture cropping, automatic size and design generation and other computer-assists to help a person generate a large variety of completed ads.

Their benefits and downsides

In a loose way, it’s the difference between a made-to-order meal and a pre-made one. Although both platforms allow for performance feedback to create a Darwinian selection of the fittest, where only the best-performing ad offspring survive for later generations, the differences point to different goals, just as a restaurant meal and prepared food have different goals.

DCO advocates point out that such a platform scales very well, a key requirement as the number of targeting conditions multiplies. Which it will, because of a dramatic increase in addressable devices and in details about users’ behaviors and situations.

DCOs, for instance, are great for those retargeted ads featuring a pair of sneakers that follows you for days from site to site, trailing you like a lost puppy, because you recently happened to look at a retailer’s product page on sneakers.

For that kind of remarketing, it’s hard to imagine any other way to pull a product shot and specs from hundreds of products in that one retailer’s catalog, immediately generating an ad to remind you of what you virtually walked away from.

What a DCO ad lacks in creative flair is made up by its immediate relevance to your recent shopping interest.

But CMP advocates, such as Thunder CEO Victor Wong, point out that social networks like Facebook don’t allow the use of DCOs in their closed environments because that would mean the ad is being assembled by an outsider. Facebook would lose control over, say, when the entire page is rendered.

Getting together?

Wong said it’s clear that targeted creative variations have big benefits. Thunder conducted case studies that found a three-fold increase in click-through rate and an 85-percent lift in conversions for ads whose creative is informed by the targeted data, compared to a repurposed generic creative. No stats yet, though, about DCO-generated ads versus CMP-powered ones.

Marketers get more control in a CMP, he said, not only over the creative for each ad, but also over the exact conditions or audience segment when each ad version is displayed. A DCO otherwise makes that decision, at the moment. But DCOs can generate far more variations than even the most empowered user of a CMP.

When asked if the two approaches could ever combine, Wong noted that his company’s CMP already allows ads to be built programmatically, based on something as simple as a spreadsheet. A key difference: They are not built at the moment of delivery.

And don’t be surprised to see DCOs evolve into having more intelligence about what works creatively and more output options that could, say, include pre-built ads designed for specific segments.