Your browser is not supported. Please update it.

27 February 2020

AT Internet’s non-specialized analytics bad news for Comscore

A recent conversation with a major European broadcaster largely left Faultline’s notebook full of disappointing “no comment” responses, although one nugget of information we managed to prize out pointed us to French analytics firm AT Internet.

AT Internet supplies the broadcaster, who must remain anonymous for now, with a central analytics suite – monitoring and optimizing content across assets including its various websites, mobile apps and streaming services for both live and on-demand content. This is interesting on a number of levels, no less because AT Internet is not specialized in video, providing basic playback analytics such as which parts of a video were viewed.

Still, the company has tied up plenty of large media contracts, in the process even switching out analytics systems from US measurement giant Comscore.

But AT Internet has a few tricks up its sleeve to set it apart from the QoE video analytics experts such as Conviva and NPAW. It was initially assumed AT Internet was a relative fledgling but is in fact a veteran with 23 years’ experience in analytics. “A few years ago, there were maybe 15 major players in analytics and now you will find only Adobe, Google, and us,” claimed Declan Owens, Digital Analytics Expert at AT Internet.

“Our tool is a global digital analytics solution, not a tool specialized only in media analytics, which are more dedicated to technical teams. Nonetheless, our media analytics solution is very advanced for users looking to understand how media content is consumed. As a comparison on the market, Google Analytics does not include video analytics, Adobe had to buy Comscore to include in its platform (the solution has not been updated in 10 years), and other indirect competitors like MixPanel, Amplitude or Snowplow don’t have it either other than event based models,” Owens elaborated.

Numerous factors have kept AT Internet “resisting and progressing” in the cutthroat analytics market, namely the “reasoned vision” of the company’s CEO, as well as emphasis on data privacy and even – Owens claimed – keeping pace with the technical prowess of Adobe and Google. “We don’t have an army of developers like these main competitors, but we have recently been rebuilding the whole data factory. This now includes scalable technology with more data storage and faster analysis, based on flat and flexible data models, which is not the case yet with our competitors,” he said.

But while our brief chat with the unnamed major customer gave the impression that AT Internet is merely a supporting cog in its mammoth in-house data analytics engine, we got the impression that AT Internet’s tool is handling a larger bulk of analytics work than either company would be willing to reveal.

As mentioned, the 200-person firm is more of an omniscient business intelligence outfit. AT Internet has long been providing advertising analytics for broadcasters and most recently, two years ago, got into churn prediction. Here, AT Internet installed a module within its analytics tool for providing user insights, using RFM clustering (standing for recency, frequency and monetary value) to classify subscribers or viewers by loyalty. Loyalty of course can be thrown off its objective course by mitigating factors such as an extended vacation, which Owens admitted could interfere with an end user’s loyalty ranking, plummeting from as loyal as a Labrador to an imminent cord cutter. In a B2B environment, for example, there is a typical slowdown in July and August which the RFM clustering system would account for, Owens explained.

The French e-commerce division of Japanese giant Rakuten is another major client for AT Internet. With some 200 million references on its site, AT Internet optimizes product description pages to boost turnover, claiming a 5% increase in new client revenue and 2% boost in total sales turnover for Rakuten France.

Owens placed AT Internet in the middle of the pricing scale compared with the main competition. It starts at €355 a month which is significantly cheaper than the base prices for specialist QoE video analytics systems and it appears to get you a lot more bang for your buck with capabilities like data mining, although it lacks the ability to drill right down to the device level. AT Internet is therefore more focused on QoS than QoE, although not in the network delivery sense as it uses technology partners for CDN analytics including features like load balancing.

“We don’t really program things, we collect information to put on a roadmap,” added Owens, reiterating that AT Internet continues to see Adobe Analytics and Google Analytics as the two biggest threats going forward, undeterred by start-ups expert in machine learning or the potential commoditization of analytics.

Of course, at 23 years old, AT Internet has survived the transition from analytics in traditional broadcast and media to an IP world – so it looks well-equipped to survive whatever disruption comes next.

“To summarize, I’d say that with a benchmark, one would easily see that AT Internet is the best option for media analytics when analyzing in the prism of digital analytics, meaning having the need to include more types of interactions and aspects and just the media player in itself,” concluded Owens.