Campaigns comprising display and mobile ads should be developed together so retargeting can be carried out seamlessly; but differences in consumer experience must be addressed.

Yann Aïtbachir, PocketMath's director of product development, said the mobile RTB ecosystem needed to consider user experience where consumer attention was shorter compared to the desktop user. The delivery of data also was reliant on mobile networks, which typically would be slower than wifi connections, depending on a country's infrastructure.

These factors narrow the window of opportunity to capture consumer attention.

Hence, Aïtbachir said, ads would need to be optimised for the mobile network so that they consumed lower bandwidth, allowing for faster loading, and to be served much faster to the mobile user.

In addition, the landscape in Asia was vastly different, with mobile markets and mobile consumer behaviour very dissimilar between India and Singapore, for instance, he said. Display ads also might not be as relevant in India or Indonesia, where consumers were likely to have skipped the desktop or laptop and headed straight for mobile platforms.

He added that CPMs were likely much lower in emerging economies where smartphone penetration rates and purchasing power were lower. Populations in these markets also spent less time online.

The disparity has brought challenges for attribution, as there are now multiple channels that influence how a consumer makes a purchase, be it on mobile or desktop, he noted. "So, targeting and retargeting your audience have to be adjusted depending on the lifecycle of the various consumer groups and how their mobile behaviour fits into that life cycle," Aïtbachir said.

PocketMath CTO Andrew Alcock

PocketMath CTO Andrew Alcock noted the challenge of associating user profiles across channels, particularly as consumers moved from desktops, where cookies were typically used to track their behaviour, to mobile, where cookies could not be deployed. "Right now, consumers are viewed as different people as they move across platforms," he said.

No legacy, fewer problems

Attribution also could prove challenging because it might be some weeks after consumers viewed an ad before making their final purchase decision. They would have accessed different channels during this time and it could be difficult to attribute these impressions to the final buying behaviour, Alcock said.

Such challenges have always existed, but new tools and access to data analytics were now emerging to enable advertisers to better track their returns, he said.

"Asia is in a better place because we now have the advantage of marrying big data with programmatic [without worrying about legacy systems], which the US and other matured markets now have to rework and integrate into their existing systems," he noted.

PocketMath's co-founder and COO Eric Tucker added that matured markets also had to deal with handling cumbersome, big systems. "So, somebody coming into this space now has the advantage of more fine-tuned systems," he added.

Tucker further noted that programmatic had been perceived as unapproachable because market players had not made it easy to comprehend, adding that this was why PocketMath aimed to make its services as transparent and user-friendly as possible.

Alcock said: "Programmatic means you have to interface with another system, you have to write your own software, and run your own software. All of that is expensive, especially if your business is not related to technology."

The good news is, adopting programmatic will get increasingly simpler as the technology matures, similar to how other technology segments have.

"If you wanted to set up an email previously, you would need a Unix server and know how to set it up. Now you can get a mailbox for free online without any setup," he said. "We see ad tech evolving in the same way, where you can access [ad support tools from] Facebook just by going online. In our case, you can access billions of impressions per day through an interface."

PocketMath tracks impressions and integrates with third-party conversion tools that track various metrics based on the customer's requirements and definition of conversion, Aïtbachir said, adding that hyperlocal GPS targeting can further improve attribution.

"Direct attribution is potentially higher because you are able to attribute any conversion to the ad that would have been served hyperlocal," he said, adding that this is usually used to support events or promotions.

Asked how Facebook's move to end support for reporting at the device level would impact PocketMath's ability to do retargeting, he noted that there were always other ways to access data.

He added that PocketPath's role in the RTB ecosystem was "one of a data conduit", where the mobile DSP facilitated retargeting for its clients, but did not own any personal data.

"There is always going to be a huge ecosystem outside of Facebook, just as how one exists outside of Google now," Aïtbachir said. "Ending support doesn't mean the data is completely inaccessible – the inventory exists anyway, so it's just a question of Facebook providing access to that data and, if necessary, how we can integrate with them as a data provider as we now do with DMPs."

Tapping the cloud for scalability & speed

Reiterating the need to look at campaigns from a singular viewpoint regardless of platforms, Alcock noted that many brands still managed the different channels separately. "Brands need to stop thinking about advertising like they did five or ten years ago. Technology has now enabled a lot of possibilities and they now really need to understand that," he said.

To address the high data response requirements of RTB, PocketMath runs its infrastructure in the cloud on Amazon Web Services (AWS). The Singapore-based startup ingests more than one petabyte of data each month from ad exchanges and processes the data to generate insights to support its retargeting capabilities, Alcock said. Running its operations on a cloud model was the only way it could efficiently scale and manage the data churn as well as ensure fast data transmission.

"With RTB, you have to respond within 60 milliseconds or you won't be able to submit a bid," he said, stressing the importance of establishing close proximity to where ad exchanges were located. PocketMath runs on AWS data centres in the US, Europe, and Asia. "We constantly look to improve our understanding of the data we're getting and integrate that with various data sources, including our customers' data and third-party data, and be able to target and retarget more effectively."