Your browser is not supported. Please update it.

19 May 2022

Recommendations roll-up imminent, vendors agree

Content recommendation engines are in a state of flux. While they are increasingly essential among a fragmenting content landscape, the underlying technology is becoming increasingly run-of-the-mill.

As the finishing touches are put on the modelling for the forthcoming forecast from our research arm, Rethink TV, now seemed like a good time to reflect on what we have learned so far from those at the coal face.

Content recommendation engines have certainly climbed down from the pedestal that they were placed on around ten years ago. The technology first emerged around the turn of the millennium, but it took a decade for the industry to fully understand the value of using techniques like collaborative filtering to engage, sustain and retain your user base.

From the start of the 2010s, content recommendation engines became the new, shiny, must-have technology for any media publisher that wanted to effectively respond to the endless consumer choice posed by the digital age.

The AI/ML hype bubble that rose around five years ago only served to exacerbate this, with many media publishers having grand visions of knowing exactly what their customers would want to see, even before they did. Content recommendation was an invaluable necessity.

It is not that content recommendation has become any less valuable in more recent years, but that the technological demands for improving UX now require more than just a recommendation engine. Content recommendation is now placed inside the ‘personalization’ suite of products, alongside targeted ad tech, UX, content management systems and metadata analytics tools.

As such, our hypothesis has been that on a per-sub basis, the price of content recommendation engines is set to decline within the next five to ten years. From the multiple vendor discussions we have had, there seems to be a reluctant acceptance of this fact.

However, vendors are fighting this commoditization by expanding the suite of tools on offer – either through in-house R&D, or via M&A.

For the first response, just look at the huge suite of tools on offer by market frontrunner ThinkAnalytics. The company is the last-man standing from the original cohort of recommendation engines, largely because it has been able to diversify into metadata enrichment, churn detection and UX services.

It has risen to the challenge of facilitating the satellite demands of its core technology.

Of course, not all companies have the resources to scale alone fast enough, so we can only imagine scenes like 24i Media’s recent purchase of The Filter becoming all the more regular. The period of consolidation is well underway.

Despite this accepted commoditization of the technology, content recommendation engines are still considered essential. In our surveys of Faultline and Rethink TV’s subscriber bases, it consistently crops up as a key area of intrigue for our readers.

This is to do with the black box nature of the industry. Not only is each vendor incredibly secretive about the inner workings of their algorithms (understandably so), but none have a clue as to where their own pricing stands compared to their competitors.

The precious nuggets of pricing information that we have extracted have left other vendors gob smacked, debunking their assumed ranking as a top-tier vendor with premium pricing to match.

We have also heard several vendors cry wolf on the vast user footprints reported by the likes of ThinkAnalytics and XroadMedia. Both report numbers in the hundreds of millions, but one fellow recommendations firm is adamant that “the real number is somewhere between what they said and zero” – in what reads as an unhelpful smear attempt.

Another reason for the sustained interest is that the core reason for the technology still exists. As the content landscape fragments, users are still incredibly overwhelmed by choosing what to watch, and operators are worried that this choice paralysis will lead to churn unless they become an integral guide in the content discovery process.

There are many more talking points that have proved controversial among vendors – not least the value proposition of AI/ML – but we will save those deep dives for the report itself. This market has surprised us multiple times on this investigation, and we look forward to fleshing out these discoveries in full.