AI Business recently got the chance to speak to Dr. Boris Mouzykantskii, the Founder, CEO & Chief Scientist of IPONWEB. We begin by discussing IPONWEB position in the bulging solution-provider space. Boris is assured of his company’s capabilities:
“We are very much at the forefront of applying AI in the form of machine learning to real-time trading and decisioning in digital advertising”.
He elaborates on this by explaining IPONWEB’s key proposition in the enterprise:
“Very simply, we empower businesses to leverage their valuable data assets to better meet specific digital advertising goals using a highly tailored approach to technology. For each customer, we effectively customize a unique machine-learning technology platform built around their business needs and requirements that enables better real-time advertising predictions and decisions at huge scale and creates significant competitive advantage”.
So in which specific industry verticals are we seeing IPONWEB? Boris outlines the company’s “base use case”:
“For digital businesses (think Uber or Netflix or Amazon, and their smaller rivals), the cost of user acquisition is a significant business expense, and one of the most scalable channels for user acquisition is digital advertising. It has now become possible to ‘in-house’ the decision on which ads should be shown to which user (and at what cost) on a case by case basis – assessing billions of opportunities to show an ad to a user every day and making a custom decision every time using the ‘Real Time Bidding’ infrastructure. In making those split-second decisions, businesses need to leverage as much internal data on their customers as possible; the scale of this task at speed drives the need for purpose-built machine-learning solutions to generate the best possible results”.
But these beginnings in digital advertising are quickly evolving, with IPONWEB able to tackle more complex use cases further enabled by machine learning. Boris details a couple of these opportunities:
“The first is Attribution – the process of accurately quantifying which advertising most influenced and drove a user’s decision or action. Much of the attribution used in advertising today is flawed because it inherently assumes that for a conversion to occur, a consumer must be exposed to an ad. In reality, many consumers make purchase decisions every day, regardless of advertising exposure, perhaps because they read a glowing review of your product or because they’re already a loyal customer. The real challenge for marketers is: how do you find those customers who require messaging from your brand to convert, and target only them? This is the concept behind ‘incrementality’ – driving incremental value from your marketing efforts, and it is a difficult challenge to solve. It takes very powerful and programmable machine learning technologies to create and execute ongoing audience sampling that identifies and messages only to those users who ‘need’ to see your ads. But this is where we believe the future of marketing is heading, because this is the only way to truly measure and evaluate ROAS (return on ad spend).
“A second example is in the TV space. In TV, 80% of the ad buying and selling takes place in what’s called the “upfronts,” a process that occurs months before any ad or TV program will hit the air. Pricing, negotiations, and demand are all driven by a complicated system of estimated ratings, paneled audience metrics, and so on. But consumption of TV, as we all know, is fragmenting across screens and devices. People are watching TV online, via over-the-top (OTT) streaming devices, like Hulu, on their smartphones, on-demand, and yes, still linearly.
So, Boris says, the big question is: How does a major broadcaster optimize upfront buys, executed months in advance against specific target audiences and program ratings in this increasingly complex and fragmented landscape to ensure delivery against advertiser goals, while still maximizing yield for every available TV slot?
“Solving a problem like this requires evaluating vast amounts of data and communication across several different legacy systems and measurement tools. But the opportunity and payoff for both brands and advertisers is huge. For advertisers, it means reaching your exact target audience across devices, which is proven to drive brand metrics; and for broadcasters, it means extracting more value from every available impression across all channels while still delivering against contracted campaign goals. This is another area where machine learning can make a big impact, consuming mass amounts of data instantaneously to make highly accurate predictions and recommendations for optimizing upfront linear TV buys across channels and devices”.
Ultimately, in the longer term, Boris says that IPONWEB sees opportunities for companies across all verticals to deploy custom AI and machine-learning technologies to create significant competitive advantage in their advertising and marketing efforts. But there are currently two types of company – those ready and actively looking to embrace the technology, and some which aren’t quite there yet:
“Some organizations are ‘more ready’ today than others. And for now, we are largely focused on servicing those segments that are most actively looking to embrace the technology and have a real need for the competitive advantages that it can provide”.
“Today”, Boris continues, “this broadly falls into three key areas”. He briefly outlines these:
“Digital-first marketers: Think eCommerce companies, travel companies, major app or internet-based businesses and services where digital advertising and technology are completely central to their business.
“Data-rich brands and advertisers: Companies with huge first-party consumer data assets who are looking to leverage that data to maximum effect in a secure way. (Think Telco’s and Financial companies.)
“Media companies who are looking to better control technology as their business become more digital. (TV broadcasters, traditional publishers, radio etc.)”
And what about within the company? Boris explains the nature of IPONWEB’s research efforts:
“We’re continually looking to see where machine learning and AI technologies can be used to solve major challenges within the overall advertising ecosystem and we have a constant R&D program looking at and tackling these problems”.
Their approach, then, is creating to solve problems, rather than creating to find them:
“We’re hesitant to productize our offerings, as we believe when it comes to machine learning, data, and creating true business value, a one-size-fits-all solution simply doesn’t work. So we don’t think of ourselves in terms of products, but more in terms of solving problems in a highly customized way”.
This personalized methodology enables closer business relationships, Boris concludes:
“What we aim to do is partner closely with a company to understand their business requirements, goals, data sets, infrastructure, and challenges – then design bespoke solutions to solve their problems in way that fits seamlessly into their organizational structure and existing workflows to create real value”.
You can access the original interview here