What People Watch

Any business taking regular payments must deal with customers who stop buying its product. Understanding, and limiting, this churn is vital for those businesses, and even more so in the world of streaming services, where customers can dip in and out on a month-to-month basis. This month’s edition of What People Watch looks at what Barb data can reveal about these viewers.

Barb’s Establishment Survey, which reports on access to SVOD services, has shown household penetration of these services dropping back slightly in recent quarters. Netflix, for example, was present in 57.7% of UK homes in Q2 2023, compared to 59.7% in Q2 2022. Similarly, Amazon, Disney+ and Apple TV+ also saw at least one quarter-on-quarter decline in the last two years. This stands in contrast to 27 consecutive quarters of growth for Netflix from 2014 to mid-2021.

Many factors, including economic circumstances, cause these changes. Our data provide an insight into the volumes and types of SVOD content as well as the particular programmes people are watching that might affect their subscription choices.

Content is especially critical considering short-term subscription contracts. Although viewers can sign up to watch a particular show without being locked into a service for a year, this means they can easily cancel too. SVOD services need to offer a steady stream of great content to keep subscribers from hitting the cancel button.

So, what can our viewing data reveal about churn? We should acknowledge first that we aren’t analysing genuine churn – as in people who have cancelled a subscription. Our data don’t explicitly pick this out. Our data do, however, identify types of viewing behaviour that could be indicative of viewers susceptible to churn.

The red bars in chart 1 show the behaviour of a group of viewers who didn’t watch any Netflix in June 2023, despite having been Netflix viewers earlier in the year. The chart compares month-by-month viewing volumes for this group — let’s call them recently-lapsed viewers — alongside everyone who watched Netflix. And we see the recently-lapsed viewers perform increasingly worse than all Netflix viewers from February through to May. Looking further back, the stronger performance of those recently-lapsed viewers around the turn of the year could be explained, at least in part, by some high-profile shows such as Wednesday and Harry & Meghan.

Chart 1: Total Netflix monthly minutage indexed against November 2022

Source: Barb. Total Netflix minutes viewed per month indexed against total Netflix minutes for November 2022 for each target. November 2022 – June 2023.

By way of a word on method, the group we have called recently-lapsed viewers consists of anyone who watched Netflix between January and May 2023, but then didn’t in June 2023. This means our analysis isn’t skewed by people who haven’t watched Netflix at all since the beginning of the year.

The data in chart 1 reveal the challenge of identifying those likely to churn. In the six months prior to not watching Netflix at all in June, this group outperformed all Netflix viewers in December and January. The under-index in February was followed by a return to near parity in March. Both months were only fractionally behind the overall Netflix viewing pattern. It is difficult to say these presaged sharper declines in April and May before zero viewing in June.

Inconsistent usage followed by sharper declines prior to cancellation, or zero usage, makes sense in hindsight. A gradual decline to zero would be neater, and easier to spot, but human behaviour is seldom neat. Cancelling a subscription is more likely a result of lower viewing over time rather than a cliff-edge decision to move from regular to zero viewing of a service. One can also imagine members of a household threatened with the cancellation of a service using it more to prove to the billpayer that it’s worth continuing. Ultimately, this appears to leave us with a warning period of only two months – a narrow window in which to try to avert churn.

What can we learn by comparing the demography of people who watched Netflix in June and the group we’re calling recently-lapsed viewers? We can see dramatic differences in terms of age and presence of children in the home. 49% of our recently-lapsed viewers were aged 55+, compared to only 16% of those who viewed that month.

Given this age disparity, these non-viewers were also much more likely to have no children aged under 16 in their home. 72% of those non-viewers in June fell into this category, compared to 50% of people who did watch Netflix in June and 75% for UK households as a whole (Q2 2023 Establishment Survey).

Are these differences reinforced when we look at the types of content being watched by the recently-lapsed viewers in the months running up to June 2023?

Chart 2: Genre share Netflix viewers and non-viewers June 2023, March – May 2023

Source: Barb genre share for viewing from March-May 2023.

Chart two shows that this is the case. For the three months prior to recording zero viewing in June, drama and film made up almost three-quarters of these people’s viewing, compared to 61% for all Netflix viewers. We also see the impact of age and a lack of children in the home as our recently-lapsed viewers consumed less than half as much children’s content, proportionally, compared to all Netflix viewers. This reinforces the prevailing narrative that kids, overtly or not, heavily influence the retention of subscriptions.

We can also consider individual shows. At our Barb Briefing in April 2022 we discussed how certain SVOD shows acted as fireworks – they explode into life, representing a large amount of viewing to a service for a relatively short time. Other content was more like a bonfire – comforting and ever-present. The analogy, with due credit to Kevin Lygo at ITV who originated it, still holds.

Chart 3: Daily audience for top 100 shows on Netflix, all viewers, December 2022 – May 2023

Source: Barb. Sum audience on each day for the 100 most viewed programmes (any episode) over the period December 2022 – May 2023.

Chart 3 shows that from December 2022 to May 2023, 27% of viewing to Netflix’s top 100 shows was driven by five library titles: The Big Bang Theory, Brooklyn Nine-Nine, Friends, The Office US and Young Sheldon.

In chart 4, which repeats this analysis for those who viewed no Netflix in June 2023, we see a different picture. The bedrock is smaller: the share of viewing for the same titles was just under 18%.

Chart 4: Daily audience for top 100 shows on Netflix, those viewing no Netflix in June 2023, December 2022 – May 2023

Source: Barb. Sum audience on each day for the 100 most viewed programmes (any episode) over the period December 2022 – May 2023.

The fate of firework titles is less clear. Series four of the psychological thriller You, which launched in March, accounted for 2.3% of viewing to the top 100 programmes for all Netflix viewers over this period, but only 1.9% for those who viewed no Netflix in June. Conversely The Night Agent series 1 did marginally better among those who watched no Netflix in June after its launch in late March.

This pattern can be replicated on other services as well. Disney+, for instance, drew 25% of its audience over the same period from five library titles including The Simpsons and Grey’s Anatomy. That figure was just 19% for the same programmes from non-Disney+ viewers in June.

These data paint a picture of subscribers who may be at risk of churning away from Netflix. They are older, more likely to view drama and film and would have seen declining viewing for several weeks prior to becoming zero viewers.

And we can see what these recently-lapsed viewers were watching in June, with linear TV featuring prominently. This is the kind of insight Barb provides which has long been used by broadcasters — and is now available to streamers — to entice viewers to spend more time with their channel or service.

Unpicking churn is fiendishly difficult for those without direct access to a service’s data. Even for those who do have access to such data, Barb provides more insight on what lapsed viewers are watching on other services, thereby offering more understanding on how to combat churn.

And it’s worth noting our group of recently-lapsed viewers weren’t lost to Netflix forever. In July, those who hadn’t viewed the previous month were approaching a similar amount of time with Netflix as they had spent in May. Some of this viewing will be from new or returning subscribers. Some will be people who didn’t cancel, but just didn’t view for a month.

Yet this analysis reinforces the idea that — while fireworks are great for ten minutes — what keeps you at the party all evening is a warming bonfire.

Doug Whelpdale, Head of Insight, Barb