Return on attention: what streaming's engagement problem tells us about the value of our time

Ted Gioia has published a characteristically provocative piece on Netflix, framing its recent results as evidence that what he calls audience capture is failing as a business strategy. His language of collapse deserves some caution, since Netflix remains a formidable business with industry-leading margins and well over 300 million subscribers, but beneath the polemic sits a question worth taking seriously: whether the relationship between platforms, audiences and value has begun to change in ways the industry's own metrics were never designed to detect.

The market clearly senses a shift. Netflix's second quarter revenue of $12.56 billion came in fractionally below consensus while growing 13 per cent year on year, and the shares still fell to a 52-week low, closing on 17 July close to half below their level a year earlier. When a company broadly delivers its numbers and gets marked down that heavily, investors are no longer debating the quarter. They are debating the durability of the model, and in Netflix's case the doubt centres on engagement.

Consider what the company chose to disclose, and what it chose to stop disclosing. Netflix will move its "What We Watched" viewership report from a twice-yearly publication to an annual one, a change announced shortly after Bloomberg reported that many of its most successful series shed between 30 and 70 per cent of their audience within four weeks of a second season arriving. Businesses seldom reduce the frequency of measures that flatter them. With Disney and Warner Bros. Discovery having already withdrawn quarterly subscriber counts, and now the sector's most transparent platform thinning its viewing data, the industry appears to be retiring the numbers it once used to narrate its growth.

The spend paradox

It would be easy to blame shrinking content budgets, but aggregate streaming investment is still rising. Ampere Analysis expects the major global platforms to spend $101 billion on content in 2026, passing $100 billion for the first time and accounting for roughly two-fifths of all content investment worldwide. Netflix is moving towards $20 billion of annual spend, and Disney has budgeted $24 billion across entertainment and sport.

At the same time, the volume of original programming has contracted sharply. Luminate counted 1,122 US series premieres across broadcast, cable and streaming in 2025, excluding news and live sport, an 11 per cent fall year on year and a third below the 2022 high of 1,695. Within that total, the scripted core has shrunk fastest: Ampere's commissioning data records global scripted orders falling from 759 at the 2021 peak to below 500, with the global streamers cutting scripted commissions in Western Europe by 44 per cent and in India by more than half during the first half of 2025. HBO Max reduced its original slate from 32 titles to 16 in a single year.

Figure 1. Global streamer content spending passes $100 billion in 2026 while US series premieres fall a third from the 2022 peak; the figures cover different markets but illustrate a changing content mix. Sources: Ampere Analysis; Luminate.

If total budgets are growing while output shrinks, the content mix is changing, and the published figures show the direction even if they cannot quantify a transfer. Streamers will commit $14.2 billion to sports rights in 2026, a 7 per cent increase, with the generalist platforms taking 44 per cent of that outlay against 31 per cent the previous year. Netflix's own shareholder letter revealed a telling asymmetry, with live programming absorbing just over 5 per cent of its content budget while producing only around 1 per cent of viewing hours. At the low-cost end of the catalogue, inexpensive formats are proliferating, and the "Other Shows" category into which Netflix folds its new video podcasts generated 757 million viewing hours in the first half of 2026. Sports rights inflation is claiming a growing share of budgets while cheaper formats fill the schedule, and the scripted originals that built these platforms sit in the squeezed middle.

Does spend still correlate with viewing and revenue?

This rotation matters because it severs a relationship the industry once relied upon. During the growth years, content investment, viewing and revenue rose together, with each new wave of originals bringing subscribers whose fees funded the next wave. The 2026 figures show those variables pulling apart. Netflix members watched 97 billion hours in the first six months of the year, an increase of just 2 per cent, while revenue grew 13 per cent. Consumption has barely grown while monetisation has accelerated. Netflix attributes the difference to membership growth, pricing and increased advertising revenue, with further price increases landing this year in the US, Mexico and Spain, and Paramount+ raising its own prices in January.

Revenue rising at 13 per cent against viewing hours rising at 2 per cent implies that Netflix now generates roughly a tenth more revenue from each hour watched than it did a year ago. Subscriber additions, advertising and currency mix all contribute to that figure alongside pricing, so it cannot be read as a pure price rise, but the trend is worth sitting with: the platform is monetising each hour of attention more heavily while the volume of attention it commands has barely grown. Netflix itself acknowledged the tension in its letter, noting that engagement means the quality and variety of viewing as well as the quantity of hours, as it develops a more refined view of how customers assign worth to the service.

Figure 2. Netflix revenue grew 13 per cent in the first half of 2026 against 2 per cent growth in viewing hours, implying roughly a tenth more revenue per hour watched. Source: Netflix Q2 2026 shareholder letter.

The returning-series data suggests customers may value those hours less. The Night Agent, among the most-watched titles in Netflix's history, drew 20.6 million views in its first season's opening week, 13.9 million for its second and 8.4 million for its third, a fall of roughly a third between the first two seasons and a further 40 per cent at the third, leaving the latest season 59 per cent below where the series began. One Piece came back 30 per cent lower, and that was the strongest result of any returning series in the Bloomberg analysis. Bridgerton, the platform's most dependable franchise, opened its fourth season below its third and became the first instalment in the series to miss the all-time top ten. Ted Sarandos countered that in aggregate the platform sees no material difference between first- and second-season viewing, and his rebuttal deserves recording, but the pattern across Netflix's flagship titles is hard to explain away, particularly when the company's parallel decision is to publish this category of data half as often.

Figure 3. The Night Agent's opening-week views fell 59 per cent across three seasons; across the wider Netflix sample, second seasons lost 30 to 70 per cent of their audience within four weeks. Sources: Netflix weekly Top 10; Bloomberg.

Holding the cadence as scripted output thins

The streamers appear to be preserving their release cadence while paring back what stands behind each scripted title. Netflix still premiered 133 series in 2025, only eight fewer than the year before, even as the wider industry cut scripted commissioning by a third and moved budget into sports rights whose inflation nobody controls. The available figures cover different scopes, mixing global budgets with US title counts, so the arithmetic cannot prove that the average production now receives less investment, though the weaker performance of returning series and the long gaps between them make it a reasonable suspicion. Series are funded a season at a time, renewal decisions wait upon a month of viewing data, and the two-year gaps that result ask viewers to stay attached to stories the platform itself treats as provisional. What the audience experiences is a library that grows every month while feeling increasingly interchangeable, franchises that return weaker than they left, and little that rewards genuine attachment. A customer who starts to feel that way brings a different frame of mind to the next price increase.

The affordability squeeze

These price rises are landing on household budgets already under sustained pressure from the cost of essentials. Deloitte's 2026 Digital Media Trends research found that around 40 per cent of American consumers had cut back on streaming within the previous three months for financial reasons, with respondents explicitly weighing entertainment against the persistent cost of food and housing. Parks Associates reports that affordability has overtaken content as the leading driver of streaming cancellations, with 30 per cent of US cancellations in 2025 attributed to household cost-cutting, up from 26 per cent in 2020, and its surveys find nearly a quarter of subscribers cancelling a service the moment they finish the specific programme they joined to watch. Netflix's own monthly churn remains remarkably low at around 2 per cent, genuine evidence of residual strength, but across the wider sector rotational behaviour is becoming normal, with a substantial minority of subscribers now timing sign-ups around release schedules and leaving once the show that drew them in has been watched. The picture in the UK is consistent, with Attest finding that value for money has become the single most important retention factor for 35 per cent of British subscribers across every age and income group, and a quarter now naming price stability itself, the simple absence of further increases, as a decisive consideration.

Figure 4. Cost has overtaken content as the leading reason subscribers cancel. Sources: Parks Associates; Deloitte; Attest.

When discretionary income tightens at the same moment that subscription prices rise, the perceived value of the entertainment experience must improve, or at the very least hold steady, for the exchange to survive scrutiny, because a household paying more from a smaller pool of disposable income applies a harsher test to every renewal. If the experience on offer is instead diluting, through thinner scripted investment, weaker returning seasons and a catalogue padded with cheap filler, the value equation deteriorates from both ends at once. Consumers rarely announce this calculation, and most could not express it as a formula, but the cancellation data shows them acting on it in growing numbers, treating streaming subscriptions less as household fixtures and more as short-term purchases to be switched off once the return has been collected.

Where the hours are going

The hours that premium streaming fails to hold are not disappearing. They are migrating to formats built on a different economics of attention. Nielsen's Gauge now records YouTube as the single largest distributor of television viewing in the United States, holding 13.5 per cent of all time spent on the TV screen in March 2026 against 10.5 per cent for Disney and 8.2 per cent for Netflix, which makes YouTube roughly two-thirds larger than Netflix on that measure, and the figure counts only viewing through a television set while excluding the phone, where the format's dominance is far greater. Short-form video has become the defining consumption habit of the decade, with TikTok users averaging around 95 minutes a day in the app, YouTube Shorts serving in the region of 200 billion views daily, and Reels accounting for half of all time spent on Instagram. Short-form video is free at the point of use, algorithmically frictionless, and consumed in the interstitial minutes that once belonged to nobody, and it has trained a generation to expect a continuous return on attention measured in seconds. A returning drama asking viewers to reinvest in a story they half remember from two years ago is competing with a format that never asks for commitment at all.

An even newer competitor for the same finite hours is generative AI, particularly the conversational applications that increasingly overlap with leisure. Sensor Tower projects that global time spent in generative AI applications will reach around 36 billion hours in the first half of 2026, more than double the 17.2 billion hours it recorded a year earlier. The methodology differs from Netflix's own reporting, so the comparison is indicative rather than exact, but the scale is striking: a category that barely existed three years ago is projected to consume more than a third as many hours as Netflix's entire global subscriber base watched over the same period, and it is doubling annually while streaming engagement grows at 2 per cent. Nor is this purely productivity time that leaves leisure untouched. Pew's 2026 research found that half of American adults now use AI chatbots, and that a quarter of them cite fun and entertainment as a primary use, ranking it third behind information search and work tasks. An evening spent in conversation with an AI is an evening not spent inside a streaming catalogue, and unlike the rivals the streamers have learned to fight, this competitor is interactive, personalised to a degree no recommendation engine can match, and improving on a cadence measured in months. The streaming industry spent a decade arguing over which subscription service would win the living room, and it may discover that the more consequential contest was for the attention of people who increasingly find their entertainment in a dialogue rather than a broadcast.

Figure 5. YouTube now leads all media companies on the US television screen, while time in generative AI apps doubles year on year. Sources: Nielsen; Sensor Tower; Netflix.

Live sport provides the clearest exception to this fragmentation, and its economics help explain why. Live sport is the only content that cannot be time-shifted, spoiler-proofed or algorithmically substituted, it is consumed communally and carries social currency into the following morning, and the market rewards those properties with a premium nobody applies to a mid-tier drama. The sports rights inflation described earlier, $14.2 billion of streamer commitments in 2026 with the generalist platforms' share rising from 31 to 44 per cent in a year, is in effect the industry pricing return on attention in the one arena where it can observe it directly, and Netflix's willingness to devote 5 per cent of its budget to programming that yields 1 per cent of hours but six of its ten biggest sign-up days makes the same calculation explicit. The platforms already understand that some hours are worth more than others; they have simply confined that understanding to sport while applying volume economics to everything else.

Figure 6. Netflix expects live programming to represent just over 5 per cent of content spending and around 1 per cent of viewing hours in 2026; live events have also accounted for six of its ten largest new-member sign-up days over the past five years. Sources: Ampere Analysis; Netflix.

The same repricing of liveness may help explain the current wave of broadcast consolidation. Sky's agreed acquisition of ITV's media and entertainment division for up to £1.6 billion, which remains subject to regulatory approval and would create the UK's largest commercial broadcaster under Comcast's ownership, looks like a declining-asset purchase on subscription economics but reads rather differently through a return-on-attention lens, because what ITV actually owns is Britain's deepest inventory of habitual live attention: daily soaps watched in something close to communal simultaneity, breakfast and daytime schedules embedded in household routine, live reality formats built on appointment mechanics, news, and event television that still assembles mass audiences in real time. Similar logic is visible in continental consolidation around free-to-air broadcasters such as ProSiebenSat.1. The broadcast schedule, long declared obsolete by the on-demand model, turns out to concentrate exactly the kind of hours that resist time-shifting and algorithmic substitution, and the acquirers appear to have recognised the value of that inventory while the prevailing narrative was still writing it off.

The final claimant on the evening is not a screen at all. Aggregate media consumption has not fallen. eMarketer forecasts 13 hours and 26 minutes of daily media use for the average US adult in 2026, although that total counts simultaneous use of different media separately, and the figures do not suggest that audiences are abandoning screens altogether. What has changed is where the marginal, discretionary hours of younger audiences are being contested. Eventbrite's 2026 research finds 79 per cent of 18-to-35s planning to attend more live events, 46 per cent of Gen Z actively limiting their screen time and nearly three-quarters saying in-person experiences matter more to them than digital ones, while global experiential-marketing spending reached $138.9 billion in 2025 and is forecast to grow by a further 10.3 per cent in 2026. An evening at a match, a gig or a dinner removes the viewer from every screen at once, and unlike a rival platform it cannot be won back with a price cut. The experience economy is bidding for the same finite hours with record money, and it is setting the benchmark for time well spent against which streaming is now silently measured.

The return on attention

Underneath all of this sits a question about what is actually being traded. The streaming economy measures itself through subscribers, hours, engagement and revenue, and these figures record how much gets watched without ever asking whether the watching was worthwhile. Attention carries a cost even where no invoice exists, because an evening spent inside one catalogue cannot be spent anywhere else. On this view, the monthly fee is merely the monetary component of a transaction whose larger element is paid in hours of a finite life. By that reckoning, the sector has pushed through something resembling a hidden price rise, collecting more revenue for each hour watched without clear evidence that the experience delivered in return has improved.

This is why I would argue the industry needs a measure of return on attention alongside the familiar consumption metrics. The useful question is not how long a viewer stayed but what the viewer received for staying: whether the hours left them genuinely entertained and glad of the time spent, or whether the system merely succeeded in preventing their departure. Platforms measure duration with extraordinary precision and measure satisfaction hardly at all, partly because satisfaction is difficult to quantify and partly because the honest answer might be unwelcome. Sport and live experiences show that at least some dimensions of return on attention are measurable, even if turning them into a single industry-wide metric remains difficult. Gioia's account of the platform lifecycle describes scale built through generosity, habits formed through convenience, and economics then optimised against an audience assumed to be locked in. My only amendment would be that no audience is ever truly locked in. Every subscriber runs a continuous, mostly unconscious audit of the exchange, and that audit turns hostile when prices outpace perceived value, when volume stands in for originality, when recommendation engines are tuned to prolong sessions rather than satisfy them, when the metrics that would settle the question quietly disappear from view, and when AI enters the production chain as a cost-reduction tool rather than a creative one. Each of these conditions is now visible somewhere in streaming, and several are visible everywhere.

Growth under hard limits

This is Growth Under Hard Limits applied to attention. Human attention is the binding constraint on the whole media system. Waking hours do not scale, and as the Nielsen, Sensor Tower and Pew figures make clear, streaming must now win each of them against YouTube, short-form video, gaming and a generative-AI category whose app usage is projected to more than double year on year. Where content is effectively unlimited and attention is strictly bounded, scale stops functioning as a durable advantage, because every additional hour has to be taken from a competing claim on the same non-renewable resource. A business that responds to a hard limit by extracting more revenue per unit of the constrained input, while degrading what it hands back in exchange, is drawing down its own franchise, and the second-season curves read like an early instalment of that bill.

None of this means Netflix is in immediate peril. It earned $3.4 billion of net profit last quarter, retains customers better than any rival, and reaches fewer than 45 per cent of its addressable households, so considerable headroom remains. The longer-term risk is more specific. The contest for raw attention is saturated, and the next competitive frontier is more likely to be return on attention, the demonstrable value a platform delivers on the hours entrusted to it. The services that endure will be those that treat viewing as a claim on the scarcest asset their customers possess. The signal worth taking from this quarter is probably not that Netflix is failing, but that the industry's founding assumption, that audiences will keep handing over growing amounts of their time and money for a flat or declining return, is quietly running out of road, and it is running out fastest among the households for whom every renewal now competes directly with the rising cost of everything else.

Sources: Netflix Q2 2026 shareholder letter and What We Watched report; Bloomberg (Lucas Shaw) reporting on second-season viewership; Ampere Analysis content spend and sports rights forecasts, January and February 2026; Luminate 2025 year-end TV report; MoffettNathanson and KPMG content spend estimates; Deloitte 2026 Digital Media Trends; eMarketer US time spent with media, 2026; Eventbrite Social Study, 2026; PQ Media experiential spend data; Parks Associates, Streaming Competition and Profitability, 2026; Attest UK streaming retention research, 2026; Nielsen The Gauge and Media Distributor Gauge, March 2026; Sensor Tower State of AI 2026; Pew Research, Americans and AI 2026; company reporting from Disney, Warner Bros. Discovery and Paramount; Ted Gioia, The Honest Broker.

illuminem Voices is a democratic space presenting the opinions of leading Sustainability Thought Leaders, their views do not necessarily represent those of illuminem.

The world needs sustainability knowledge. At illuminem, no interest group or shareholder can influence our work. Thank you for supporting our mission to make high-quality and independent sustainability information free for all. Every contribution helps. Thank you for donating today.