Choueiri Group’s digital arm, Digital Media Services (DMS), launched its audience targeting solution toward the end of February, making it one of the few media representation companies to offer such a solution. We caught up with the agency’s head of data science, Youmna Borghol, to learn more about everyone’s favorite not-just-buzzword, data.
Can you tell us a little about these audience data segments?
We represent more than 40 sites in the region across verticals, which gives us a tremendous amount of data. If you start stitching this data together, you can create audience segment profiles for better targeting. We reach 70 to 75 percent of the MENA population – 77 percent in the GCC alone – and roughly 85 percent in KSA, which is the key market. So, we have the scale to reach almost everyone.
There’s a certain association between data and programmatic, but aren’t these audience segments available even on direct media buys?
When you talk data and audience targeting, the thinking is that it’s all about performance and it’s mostly programmatic, simply because programmatic is fuelled by data. Also, DMPs were introduced in the programmatic ecosystem, so there is this correlation between data, DMPs and programmatic, but it doesn’t mean audience targeting can only happen programmatically. If you’re doing a branding campaign, you can still make sure you’re targeting the right audience with a direct buy. We create and activate those [audience] segments and push them to the ad server or to the SSP for programmatic or direct activation. So, instead of sending a booking order and asking for specific sites, brands can ask for a specific audience and buy that or access it programmatically.
We’ve often heard that there’s a lack of data in the region. Is it just a lack of data or that the data is not fresh or it isn’t the right kind?
It’s a combination. There was a lack of data overall, with only the big players, such as Google and Facebook, providing data, but only within their platform.
Today, even if there’s a lot of data, the question is more about whether that data has value or whether it drives efficiency [for the brand]. So, from a brand perspective, the two questions are: one, is it driving additional value to my targeting activities? And two, is it helping me better understand where the consumer is in the sales cycle? If it answers these questions, then it’s the right data.
So, while we have data and targeting, where we do we stand as the regional industry in terms of measurement?
When using audience targeting, it’s very important to have the right measurement. That consists of two things:
1. Measuring the right metrics. A lot of marketers measure CTR, conversion and leads when using an audience strategy for a branding campaign, instead of on-target impressions and other layers of intelligence, such as measuring incremental brand lift.
2. The right analytical framework. Attribution models are a big problem, because a lot of marketers are talking about last-click attribution models not being the right ones, but no one is creating any sophisticated models. If you’re attributing to last-click, that means your retargeting activities, for example, would take most of the credit, when everything before that is assisting in that conversion and you’re not tracking that.
What about measurement on mobile and across devices?
First, it’s about building the digital identity of the consumer and then adding a layer of data to stitch everything together and understand that it’s one person [across devices]. This is still challenging on a global level. However, you can still do cross-device identity management.
There are two ways of doing it: deterministic and probabilistic data. The percentage of deterministic data isn’t huge for us and most other publishers, and this is where probabilistic data comes into play. However, for probabilistic data, you need to have scale. If you’re a publisher with a small scale, probabilistic is not accurate.
What makes for a strong audience solution?
There are four core dimensions that need to be available:
1. Scale. If there’s no scale, the data won’t mean anything, because it won’t be accurate.
2. Breadth and depth of data. The more data you have around the same consumer, the more you can understand them.
3. Persistence. You have to track the consumer across different devices and consumers, or else you won’t have the full picture.
4. Precision. This is where declared versus passive and observed versus inferred data comes into play.
