Your browser is not supported. Please update it.

12 May 2022

AI analytics space smells streaming blood

Video analytics vendors are using recent shifts in the streaming landscape to their advantage, twisting a narrative of ‘peak streaming’ whereby only OTT content providers with the largest datasets and strongest AI-based platforms will survive.

While the industry should do its best to ignore such sensationalist headlines related to Netflix’s subscriber blip and the implosion of CNN+, it is true these two incidents have contributed to a new analytics landgrab as service providers scramble to harness everything there is to know about user behaviors, out of fear they might be next.

Symphony MediaAI, for instance, is spinning up a marketing drive based on these high-profile turning points for global streaming. The US analytics firm’s CTO, S.V. Vasudaven, has undergone a drastic mindset change as a direct result of the recent weaknesses of Netflix and CNN+.

Speaking on a webinar panel on AI and analytics, hosted by IBC365, Vasudaven claims his message would have been very different if the event was taking place a month earlier. Now, he is waxing lyrical about a surge in competition for consumer attention, particularly in acquiring and maintaining that attention.

On one hand, this may be an opportunistic PR stunt on Symphony MediaAI’s behalf. On the other, it might be a genuine reflection of customer demand that we should be taking seriously.

What’s clear is that Symphony MediaAI believes the next frontier in video analytics lies in delivering AI-based tools to data teams to predict and execute on game-changing business decisions well ahead of the competition.

Holistically, Symphony MediaAI’s software platform helps content provider partners manage, track, predict and grow content monetization. The manage and track parts of the platform are what Vasudaven describes as the fundamental nuts and bolts of accounting – such as tracking where revenue is coming from, which distributors are providing it, and who’s performing best.

Once you are able to capture the financial nuts and bolts, you are in a position to be able to predict with what he calls ‘classic analytics,’ which relates to questions such as why are things trending? Or which distributors are working better for me than others?

This classic approach to analytics can tell content providers where they are heading, but the difference with AI-based analytics is that you can discover where you should go. AI in this context can answer questions like which piece of content should I license? Which content will perform well with which audience? Are there signals I can pick up on to learn where strategically I should head?

AI is good for helping with tasks such as classification, predicting, detecting outliers, and decisioning. This issue is that there’s so much data that has come into the situation that you need machine learning to help out.

Vasudaven cited a use case of using AI/ML to mitigate churn – working with an unnamed SVoD provider to understand and predict when subscribers will churn.

In this case, the client provided Symphony MediaAI with a wealth of signals, which it ran through ML algorithms and found behavioral characteristics that indicated tendency to churn, including binge watching certain titles before churning.

When a user starts bingeing one of these titles, this should then be a trigger to call your best churn-prevention mechanisms into action.

One data grouping even found a cluster of subscribers that had been on perpetual free trials due to a software bug, which was an embarrassingly obvious missed revenue opportunity that Symphony MediaAI helped to monetize. No wonder this customer doesn’t want to be publicly named.

This ties into how the platform is becoming more proactive in nature, such as by identifying subsets of viewers with a high probability of churn and offering a number of actions to circumvent this.

“In an unsupervised ML environment, you can capture the shape of data,” Vasudaven explains. “But ML knows nothing about video, just data; grouping data together which is similar and allowing us to go in and inspect these groupings. This surfaces what is referred to in the industry as unknown unknowns.”

Symphony MediaAI serves both subscription-based and ad-supported clients with very different needs. Its tools light up insights into how consumers are interacting with content, but with a real focus on where the money is coming from, reducing expenses around customer acquisition, and focusing on maximizing existing customers, thanks to its parent company’s roots in financial services.

What impresses us about Symphony MediaAI is how it melds two data reservoirs that traditionally are accessed separately. On one side, it works with content providers on accounting, where subscriber churn burns a gaping hole on the balance sheet, most of the time. But by only analyzing accounting, you miss insights into why consumers have churned.

Symphony MediaAI is still in the early stages of marrying money tracking with behavioral data, to answer questions such as what is a consumer actually doing? What are they watching? How much are they watching?

It’s only when you bring these two datasets together that you can see what decisions to make on content procurement and what effect it has on viewership. Did I make the right choices? How can I close the loop on the decision process, to see the cause and effect relationship between business choices and consumer response?

For Vasudaven, one of his favorite aspects of AI is that once you have the history, Symphony MediaAI can provide tools to content executives to make growth predictions. He has a prediction of his own, foreseeing increasing demand over the next several months for AI and ML capabilities in addition to the company’s base product, to surface insights that are humanly impossible to visualize without computational horsepower.

“It’s going to benefit consumers to access content they’re interested in and will benefit content providers to make the right investments to maximize enjoyment of subscriber populations,” voiced Vasudaven.

All this talk of AI and ML wouldn’t be complete without an obligatory contingency comment from Vasudaven on how these tools are designed to support data science teams, not represent the threat of replacement.

Symphony MediaAI has seemingly come from nowhere, really as two companies in one that has burst onto the media scene from the financial services side where it has operated for around 30 years. As the name suggests, the focus is on media, but it has sister companies focusing on areas such as healthcare AI and retail AI.