Your browser is not supported. Please update it.

22 June 2017

Conviva shines light on funneling $40m funding into AI, APAC

Video analytics company Conviva secured a $40 million funding round last week, so we were keen to delve into the company’s plans for putting the cash towards developing artificial intelligence (AI) technologies for its OTT intelligence platform – with initial skepticism about the use of the AI buzzword.

AI is not an area Conviva has traditionally blown its own trumpet for pioneering, so increased investment in this space, coupled with a road map towards reaping the rewards of a booming OTT video market in Asia Pacific, has the potential to further cement the company’s dominant position in the OTT UX market.

This is fitting of the tech industry as a whole today, not just video analytics, with the volume of companies claiming to be AI-driven soaring within the past year or so – in fact around 1,900 companies have taken AI investment amounting to $12 billion during the past 5 years. Faultline Online Reporter spoke to Conviva’s CMO, Ed Haslam, this week, who explained that Conviva has in fact been developing machine learning algorithms for more than four years, which have been available on its SDK to customers for some time.

We consistently find that companies struggle to explain what is really behind their AI and machine learning technologies, aside from algorithms that can predict behaviors, and there is also a mystery about what differentiates these companies’ in-house developments. Yet Conviva’s Haslam did a decent job of pinpointing where some of his company’s funding will be invested – in a specific form of machine learning called transfer learning algorithms.

Transfer learning is a method of machine learning with an additional source of information apart from the standard training data, whereby it removes the isolated tasks which come with traditional machine learning algorithms by developing methods to transfer knowledge learning in one or more source tasks, therefore improving learning in a related target task. Essentially this works in a similar way to how humans can apply learned knowledge to a variety of different tasks. For a more comprehensive description, see this paper from the University of Wisconsin.

Conviva is applying its existing machine learning algorithms in two ways. Firstly, for detecting patterns in video across devices, and secondly for helping content publishers identify important metrics, such as bit rate and buffer rate, which are used to predict engagement and set up appropriate proactive responses to any issues that may arise in the delivery of video, to avoid a customer hitting their disengagement threshold.

Conviva will be applying its newly developed algorithms to a fresh auto-diagnostic product for comparing CDNs and using a predictive process of elimination to identify which CDN is causing, or could potentially cause, any delivery issues. More on this will be revealed once the product launches, which Haslam said is expected to debut at IBC this year.

There is a mindset of some in the industry, as told to us by a handful of executives, that true AI cannot be achieved by a company unless it has built its own APIs and is not relying on third-party APIs. Conviva says it is one such company which has made this investment, developing log-based APIs which provide viewer level diagnostics to customers. This self-driven API approach allows a company’s own engineers to monitor and control exactly what a platform is learning and make the appropriate tweaks when necessary, rather than relying on external APIs such as those from Google, Microsoft or IBM.

Conviva’s $40 million funding, which included new investor Future Fund as well as existing investors Time Warner Investments, NEA and Foundation Capital, will also be funneled into expansion efforts in the Asia Pacific and Latin America regions. Haslam pointed out Australian telco Telstra and Hooq, an SVoD service owned by Singapore Telecom and Sony, as two of its existing major contracts in the Asia Pacific.

Asia Pacific and Latin America are both areas that our research arm has forecast to experience significant growth in OTT video over the coming years. Hooq is set to grow to around 1.5 million subscribers across its footprint in Asia by 2021, and Telstra’s TV Everywhere and Foxtel Play services are projected to see healthy uptake, despite a relatively slow start.

Conviva is in a comfortable position in the QoE OTT analytics market with over 50% of all major tier 1 players globally, including HBO, Fox, Sky, DirecTV, Time Warner Cable, Verizon, Dish, Liberty Global, BT and the BBC. It claims it receives reports from 2 billion devices a month, monitoring 4 billion streams across 180 countries.

The US firm has recently struck partnerships with Adobe and IneoQuest, which is also considered to be Conviva’s main rival along with S3 Group. Conviva is also in the early stages of discussing a partnership deal with CDN load balancing firm Cedexis.

Following the investment round, Conviva’s CEO Dr Zhang said, “We have consistently been growing faster than the overall market and have developed a strong, sustainable business model. We are excited to close this round of funding so we can accelerate our new product offerings and our global market expansion, extending our technology and market leadership in OTT measurement and analytics.”