Attention Is the New Currency

How algorithms turned time, exposure, and behavioral data into the infrastructure of modern consumption

The average person will spend roughly 6 hours and 37 minutes looking at screens today. Not working, not communicating with purpose. Just consuming. That figure, drawn from DataReportal’s 2024 Global Digital Report, is higher than the time most people spend eating, exercising, and engaging in meaningful social interaction combined. And embedded within those hours is a machinery of economic extraction that most people never think about, precisely because it was designed not to be thought about.

Almost none of those 6+ hours required a direct payment. The platforms that captured that time (YouTube, Instagram, TikTok, Netflix, news aggregators, recommendation engines) built systems so effective at retaining attention that the question of payment became almost beside the point. They found something more valuable than a subscription fee. They found a way to own a piece of your daily cognitive budget.

Take a 26-year-old software engineer in Bengaluru. One Sunday evening she opened Instagram Reels, not to shop, not to plan anything, just to unwind after a long week. She watched a cooking video. Then another. At some point she paused for eleven seconds on a reel about kitchen organisation. That was enough. Over the following two weeks her feed quietly shifted: more cooking content, small appliances, aesthetic home setups. She didn’t notice the drift. Three weeks after that Sunday, she opened Flipkart during a sale and bought a Rs. 3,400 spice rack she had never once thought about owning before. She assumed it was her own idea.

It wasn’t. The sequence was: eleven seconds of dwell time, behavioral signal, feed recalibration, manufactured familiarity, purchase intent, transaction. She never searched for a spice rack. The system didn’t need her to.

That process repeats itself billions of times every day. And it is the actual story of how modern consumption works, which is considerably stranger than the one most economics textbooks describe.

This isn’t a story about screen addiction or tech industry malfeasance, though those are adjacent conversations worth having. It’s a story about how the underlying economics of consumption quietly transformed over the past two decades, shifting from a model driven by purchasing power and scarcity to one driven by attention capture and behavioral influence. The transaction is still there, but it moved. It moved upstream, into your head, long before you open your wallet.

Reading this chart: Bars (left axis) show average daily screen time in minutes; the line with diamonds (right axis) shows annual digital advertising revenue per internet user in USD. Countries are sorted left to right by screen time. The stark gap between the Philippines, Indonesia, and Nigeria (high screen time, sub-$10 ARPU) and the United States and United Kingdom (somewhat lower screen time, $200-$312 ARPU) illustrates that attention capture and attention monetization are decoupled. The conversion infrastructure, not the raw hours, is what drives revenue per user.

How Consumption Used to Work

The industrial model of economics was, in retrospect, almost refreshingly legible. You had income, that income constrained your choices, and you spent within those constraints based on some combination of need, preference, and social pressure. The supply chain was physical, the shelf space was finite, and the advertisement was an interruption: a 30-second intrusion between the content you actually wanted and the content that came next.

Scarcity structured everything. A consumer goods company launching a new product in 1975 had to fight for shelf placement, invest in broadcast advertising at prices that reflected finite airtime, and wait for word of mouth to propagate at human speed. The bottleneck was distribution. Getting the product in front of people was expensive, and that cost served as a natural filter. Only products with sufficient margin could afford the exposure.

Marketing in this era was fundamentally interruptive and imprecise. You bought a television spot during a popular show and accepted that a large portion of your audience was uninterested. You paid for reach and tolerated waste. Economists could model this system reasonably well: households had income, income flowed to expenditures, and advertising was a tax on attention that funded content production. The accounting was messy but the causal arrows were clear.

Even early internet economics mostly replicated this model. Banner ads were digital billboards. Portals like Yahoo organized the web the way a shopping mall organized retail, a curated directory that charged for premium placement. The idea was that if you could get people to a page, you could monetize that page the same way you monetized a newspaper or a television broadcast.

What nobody fully anticipated was what would happen when the cost of distributing content effectively dropped to zero.

The Attention Economy Shift

The phrase “attention economy” was coined by psychologist and Nobel laureate Herbert Simon in 1971, and later developed by Michael Goldhaber in the late 1990s. Simon wrote that “a wealth of information creates a poverty of attention.” The insight was prescient, but it took another decade of internet infrastructure to make it economically structural rather than merely philosophical.

When distribution costs collapsed, content supply became effectively infinite. The constraint that had shaped media economics for centuries (that publishing was expensive, that broadcast spectrum was limited, that shelf space was physical) dissolved. And when supply becomes infinite, the economics change entirely. Suddenly, the scarce resource isn’t content. It’s the human attention required to consume it.

YouTube launched in 2005 and was acquired by Google for $1.65 billion in 2006, before it had generated meaningful revenue. The bet Google was making was not on YouTube’s current business, which barely existed, but on the trajectory of time spent. A platform that could aggregate enough human attention could monetize it through advertising. The specific form of monetization was secondary to the accumulation of the resource.

This logic now pervades every major consumer platform. TikTok’s core product innovation was not short video, which had existed before. It was a recommendation algorithm aggressive enough to keep users on the platform by surfacing genuinely personalized content with no manual curation required. The result was a median session length that alarmed researchers and competitors alike. By 2023, TikTok users in the United States were spending an average of 95 minutes per day on the app, according to data from data.ai (formerly App Annie). Instagram, which had existed for years before TikTok’s rise, subsequently rebuilt its feed around algorithmic recommendations and video in a direct competitive response.

The economic model that emerged from this isn’t simply “show ads to people.” It’s structurally different from broadcast advertising in ways that matter enormously. The old model paid for reach. The new model pays for behavioral predictability.

A platform that knows you watched 47 videos about running shoes in the past month, that you pause on certain price points but scroll past others, that you’re most likely to click through to a purchase on Tuesday evenings: that platform isn’t just selling ad space. It’s selling something closer to purchase intent, packaged and tradeable. Attention came first. Behavioral profiling followed naturally, and monetization emerged downstream of both.

Attention as a Measurable Economic Resource

One of the less-discussed shifts of the past decade is that attention became auditable. Not vaguely tracked, not roughly estimated. Quantified at a granularity that would have been impossible before ubiquitous mobile devices and persistent behavioral logging.

The metrics that drive platform economics today are not pageviews or unique visitors. They are daily active users (DAU), time-in-app per session, session frequency, video completion rate, scroll depth, and what Netflix calls “hours viewed”, a figure the company began disclosing publicly in its top-ten lists in 2021, though it remains cagey about full platform-level data. Facebook (Meta) publishes DAU and monthly active user counts in its earnings reports. Snap reports daily active users segmented by geography. YouTube reports that users watch over one billion hours of video per day. These are not marketing figures. They are the core operating metrics of attention-based businesses.

The reason these metrics replaced older web metrics is that they better predict revenue. A session duration increase of 10% on a social platform correlates reliably with ad impression volume and, further downstream, with behavioral data quality that improves ad targeting precision. A 1% increase in video completion rate on a streaming platform predicts subscription renewal likelihood. Platforms discovered that these engagement metrics led the advertising and subscription revenue figures by weeks or months. If engagement was rising, revenue would follow.

Reading this chart: Each cluster of bars represents one platform; each bar within the cluster is a geographic segment. The dotted vertical line separates ad-supported platforms (left) from subscription or hybrid platforms (right). For ad-supported platforms (Meta, Snap, Twitter/X, and YouTube) the US and Canada ARPU is 5x to 17x the Rest of World figure, reflecting both purchasing-power differences and the maturity of local ad-targeting infrastructure. For Netflix and Spotify, the regional gradient is far shallower because subscription pricing is tied more directly to local ability to pay, not to behavioral data quality. Note: Twitter/X 2022 figures are from the last audited annual filing (SEC EDGAR); post-acquisition financial data has not been made publicly available in audited form.

What makes attention uniquely valuable as an economic resource is its combination of scarcity and replaceability. Human attention is finite. There are only 24 hours in a day, and the share of those hours allocated to digital platforms has a ceiling set by biology. But unlike physical goods, attention doesn’t get consumed through use in a way that raises costs. Capturing one more hour of attention costs platforms almost nothing. But that extra hour still produces measurable value. That combination is what created the conditions for trillion-dollar platform valuations based on engagement metrics rather than earnings.

How Algorithms Shape What You Buy Before You Know You Want It

The consumer journey in classical marketing theory had a beginning: awareness. You became aware of a product, developed interest, formed intent, and made a purchase. Marketers tried to insert themselves at the awareness stage and shepherd consumers through the funnel.

What recommendation systems did was dissolve that boundary between awareness and exposure. The Bengaluru engineer wasn’t searching for a spice rack when she watched that cooking reel. She wasn’t shopping. But the system logging her session knows that kitchen content correlates with later searches for small appliances and home organisation products. It served her adjacent content that gently expanded her exposure surface. Eventually, the product appeared in her awareness, not as an advertisement, which her brain had learned to filter, but as something she had organically come to want.

That is the mechanism at scale. YouTube and TikTok use collaborative filtering (people who watched X also watched Y) layered with content-based signals from metadata and session behavior. Instagram’s algorithm weights how often you’ve interacted with an account, how long you pause on specific posts, and what content types your history suggests you prefer. What the system builds, through thousands of such micro-observations, is a working model of your attention, before any advertisement has even appeared.

Infinite scroll removed a natural friction point. Before it, clicking “next page” gave users a moment to decide to stop. Aza Raskin, who designed the feature and later expressed regret about its proliferation, estimated it generates approximately 200,000 additional hours of global scrolling per day. That number is almost certainly much larger now. The variable nature of a feed does the rest: when most posts are ordinary and a genuinely good one appears without warning, the unpredictability drives continued checking. Platforms don’t optimize for this explicitly. Their engagement metrics reward it automatically.

What this means for consumption is that modern purchasing behavior is increasingly being shaped by exposure events that happen in what feels like unrelated contexts. Someone who spends 40 minutes watching travel content on YouTube is not shopping for flights. But they are building a mental model of a destination, a travel style, a standard of accommodation. That model will influence a purchase decision weeks or months later. The platform’s ad system may then target them with flight deals, hotel promotions, or luggage ads based on those viewing patterns. The gap between exposure and transaction can be large, which is part of why the causal link between attention capture and eventual spending is underappreciated.

Reading this chart: Each row is one product. Three signals are plotted: red = social media attention (Google Trends proxy); blue = Google Search volume; teal = sales or purchase signal. All three are independently normalised so peak = 100, meaning only the timing relationship matters, not the absolute scale. The arc-arrow in each row marks the lag between the social attention peak and the eventual sales peak. Key finding: the lag ranges from 2 weeks (Stanley Cup, fast-purchase impulse) to 7 weeks (Air Fryer, higher-considered purchase). In every case, exposure precedes demand. The purchase did not trigger the search; the attention came first and manufactured the intent. Sources: Google Trends public data (product-level keyword series); Amazon BSR signals; Pop Mart Q2 2023 earnings; NPD Group consumer electronics retail data 2020. All series are modelled Gaussian pulses anchored to independently verified peak weeks.

The True Cost of “Free”

The proposition that internet platforms offer free content is technically accurate in the narrow sense that no payment changes hands at the point of consumption. It is misleading in almost every other sense.

The value transfer is not symmetric with what you might expect from a traditional transaction. When you pay for something, you exchange money for goods or services and both sides know what they’re getting. When you consume a “free” platform, you exchange attention, behavioral data, and a degree of predictability, and most people have little understanding of what that means in aggregate. Most never need to think about it. The transaction was designed to be invisible.

Consider what behavioral data actually represents. Every click, pause, rewind, and share on a platform is a behavioral signal. Aggregated across millions of users and run through modern machine learning pipelines, these signals reveal not just what you like but how you make decisions, what emotional states drive you toward or away from certain content, how long you take to make up your mind, and what marketing messages are most effective on you specifically. This is not data that you hand over consciously, in the way you might fill out a preference survey. It is extracted through your normal behavior and would take significant effort to avoid generating.

The economic value of this data is embedded in the advertising rates platforms charge. Google’s advertising revenue in 2023 was $237.8 billion. Meta’s was $131.9 billion. Both figures are primarily explained not by the number of people using the platforms, but by the quality of the targeting data those users have generated. An advertiser buying a Meta impression is buying access to someone whose behavioral profile suggests they are likely to respond to a particular message at a particular moment. That’s worth considerably more than a demographically-targeted broadcast impression.

The predictability dimension is less discussed but worth taking seriously. A platform that can predict what you will click on, how long you will stay, and what content will keep you engaged has, in effect, modeled your behavior. That model is a business asset. It improves with continued engagement, which is why retention metrics are so economically important. Every additional hour of your attention doesn’t just generate another ad impression. It refines the model that makes all future impressions more valuable.

This dynamic was partially described by Shoshana Zuboff in her 2019 book “The Age of Surveillance Capitalism,” though her framing was more political than economic. The core insight stands: the raw material of these businesses is human experience, processed into predictive behavioral products. Your life, disaggregated into signals, becomes a factor of production.

In high-income markets like North America, Western Europe, and Australia, revenue per user is high because users have purchasing power worth targeting and their data infrastructure is mature enough to monetize precisely. In lower-income markets across Southeast Asia, Sub-Saharan Africa, and Latin America, revenue per user is lower but growing rapidly as platforms build out regional ad infrastructure. DataReportal’s 2024 data shows that global digital advertising spend crossed $600 billion in 2023. The growth in emerging markets is where the next wave of attention monetization is being built.

The Numbers Behind the Economy of Time

Statistics in this domain are abundant, but they require careful interpretation because platforms control most of the granular data and release it strategically.

According to DataReportal’s January 2024 Global Digital Report, the global average daily internet use is 6 hours and 37 minutes. Social media use alone accounts for 2 hours and 23 minutes of that. Television viewing adds another 3+ hours in markets like the US. The overlap between categories is significant, since a large share of “television” viewing now happens on connected streaming platforms, and the aggregate screen time figure suggests that the waking hours not spent in front of a screen are becoming the minority.

Statista data shows that global digital advertising spend reached approximately $626 billion in 2023, up from roughly $200 billion in 2016. That seven-year growth trajectory reflects the maturation of attention-capture infrastructure: better targeting, better measurement, and more hours to monetize.

Reading this chart: The red line (left axis) tracks total global digital advertising spend in billions of USD. The blue line (right axis) tracks total global social media monthly active users in billions. The dashed teal line shows ad revenue per social media user per year. Between 2016 and 2019, both grew at roughly comparable rates. After the 2020 to 2021 post-COVID digital advertising surge, ad spend grew far faster than the user base, confirming that the same pool of attention became substantially more valuable as targeting infrastructure matured. By 2023, ad revenue per social media user had reached approximately $126, up from $84 in 2016.

The subscription economy, which represents a different but related model of monetizing engagement, has grown significantly alongside ad-based models. Zuora’s Subscription Economy Index, tracking revenue among its clients, reported compound annual growth rates in subscription revenue of roughly 15 to 20% between 2012 and 2022, compared to S&P 500 revenue growth of around 4% annually in the same period.

Reading this chart: Both series are indexed to 100 in 2012. By 2022, the Zuora Subscription Economy Index reached approximately 591, a 5.9x increase over the baseline. S&P 500 aggregate revenue grew to roughly 155, a 1.55x increase. The red shaded area is the compounding outperformance gap. The S&P 500 line captures the expected 2020 COVID revenue dip and subsequent recovery; subscription-economy companies show far less cyclical vulnerability, because their model is built around retained commitments rather than one-time transactions.

Netflix’s global subscriber base crossed 260 million in early 2024. Spotify has over 600 million monthly active users. Amazon Prime has over 200 million subscribers globally. These aren’t primarily content companies. They are machines for acquiring and retaining attention. The content is just the hook.

The relationship between recommendation exposure and actual purchasing is harder to find in clean public data, but Amazon provides the closest proxy. The company has disclosed that approximately 35% of its revenue is driven by its recommendation engine, surfacing products based on purchase history, browsing behavior, and collaborative filtering. That figure, from a 2013 McKinsey study that Amazon has referenced, may be dated, but the scale of personalization investment Amazon continues to make suggests the number hasn’t declined.

The Infrastructure Layer Nobody Talks About

Recommendation systems have become, in a meaningful sense, the hidden infrastructure of modern commerce. They are not just features of individual platforms. They are the connective tissue between attention and transactions at scale.

Reading this chart: The donut (left) shows each app’s share of total US mobile screen time in 2023. The bar chart (right) shows the same data with parent-company ownership labels. The central figure, 38.5%, is the combined share of just four apps: YouTube, TikTok, Facebook, and Instagram, owned by two companies (Alphabet and Meta). The top seven apps together account for nearly half of all mobile screen time, with everything else (maps, email, games, productivity, banking) competing for the remaining half. This is what attention oligopoly looks like structurally: a narrow concentration controlling the majority of cognitive real estate in the average day.

When Amazon surfaced its Marketplace in the early 2000s, it was a shopping search engine with limited personalization. By the 2010s, it had built one of the most sophisticated real-time recommendation systems in existence, one that factors in session context, historical purchase data, geographic location, price sensitivity signals, and time-of-day behavior simultaneously. The result is that the “first page” of an Amazon search for almost any product category is not an objective ranking. It is a personalized prediction of what you, specifically, are most likely to buy.

Spotify’s Discover Weekly playlist, launched in 2015, was an early example of recommendation-as-retention applied to audio. Users who engaged with Discover Weekly had higher subscription renewal rates than those who didn’t, a finding that drove Spotify to invest heavily in its recommendation infrastructure. The playlist now generates billions of streams per month. It is both a product feature and a retention mechanism, which is the same thing when your business model depends on continued subscription revenue.

Google’s Search, often framed as a neutral information retrieval system, has evolved into a personalized attention routing system. Search results are filtered by location, search history, account data, and real-time behavioral signals. The ads embedded in those results are priced through real-time auctions where the bid price reflects not just keyword competition but user-level predicted conversion rates. A search for “running shoes” from an account with a history of athletic purchases will surface different results at different ad prices than the same search from a first-time searcher. The platform is routing attention, and the routing decision reflects behavioral modeling.

These systems often deliver genuine value. The point is structural: recommendation infrastructure has become the primary mediator between supply and demand in large sectors of the consumer economy. Suppliers, whether content creators, product manufacturers, or brands, now optimize for algorithmic placement rather than shelf position or broadcast reach. The algorithm is the new distribution bottleneck.

For businesses, the implication is that building a product is no longer sufficient. You have to build for discovery within attention-capture systems. The growth of influencer marketing, a $21.1 billion industry in 2023 according to Influencer Marketing Hub, reflects this logic directly. Brands don’t just pay for impressions; they pay for content that can travel through social platforms’ recommendation systems, embedding product awareness in contexts where audiences are already primed for influence.

Where the Model Breaks Down

An honest assessment of the attention economy thesis has to acknowledge where the causal story is incomplete or overstated.

Income still matters, enormously. A person with no discretionary income cannot be influenced into purchasing by recommendation algorithms, regardless of exposure. The attention economy model assumes some baseline of purchasing power that can be activated through behavioral influence. For the roughly 700 million people globally living below the World Bank’s poverty line, attention capture is largely irrelevant to consumption patterns. Even in middle-income markets, large purchases like housing, healthcare, and education are driven primarily by financial constraints and institutional factors, not algorithmic influence.

The conversion rate between attention and spending is not uniform. Different demographic groups, different content categories, and different purchase types convert at radically different rates. A 70-year-old with high TV consumption and limited social media presence has a very different relationship to algorithmic influence than a 22-year-old living on TikTok. Cultural factors matter too. GWI’s consumer research consistently shows that the role of social media in purchase decisions varies significantly between Southeast Asian markets (where social commerce is deeply embedded) and European markets (where purchase behavior is more conservative).

Not all attention is created equal from a monetization standpoint. Passive background consumption (a podcast playing while someone cooks, music streaming at the gym) generates attention time statistics but limited behavioral signal. The monetization infrastructure works best with active, engaged consumption that generates dense behavioral data. An hour of TikTok scrolling is economically very different from an hour of background Spotify listening, even though both count toward screen and audio time statistics.

There’s also a genuine question about the long-term stability of this model. Apple’s App Tracking Transparency changes in 2021, which required apps to ask users for permission before tracking them across other apps and websites, materially reduced Meta’s ad targeting precision and contributed to the company reporting its first-ever revenue decline in the third quarter of 2022. Regulatory pressure, including GDPR in Europe and evolving FTC guidelines in the US, is creating friction for behavioral data collection. If the regulatory environment tightens significantly, the model that converts attention to behavioral data to targeted advertising becomes more expensive to operate and less precise in its outputs.

The System as It Really Operates

What emerges from this is a picture of an economic system operating on a different logic than the one most people carry in their heads.

The classical economic model of consumer behavior assumes rational agents with preferences and budget constraints, making purchase decisions that maximize utility. This model is useful at the macro level but increasingly inadequate for explaining individual consumer behavior in an environment saturated with algorithmic recommendation and attention-optimizing design.

The more accurate model looks something like this: platforms accumulate attention by offering content at zero monetary price. That attention generates behavioral data, which is processed into predictive models. Those models are sold to advertisers in the form of targeted impressions, and to other businesses in the form of Amazon-style recommendation placement. The behavioral models also feed back into the platform’s own recommendation systems, making future attention capture more efficient. Over time, the platform becomes increasingly accurate at predicting and then shaping consumer preferences, not through coercion, but through the gentler mechanism of controlling what people see and when they see it.

The consumer experiences this as convenience and personalization. The algorithm surfaces content that matches their interests, and to be fair, it often works remarkably well.

The economic extraction is real but invisible.

What is transferred in exchange for that personalization is not just ad impression revenue. It is a share of cognitive sovereignty. The platform increasingly determines what enters your awareness, which products you become familiar with, which ideas feel normal, which possibilities feel worth considering.

This is a form of influence that is structurally different from traditional advertising. Old advertising could persuade you to prefer one brand over another within a category you were already aware of. Recommendation systems can shape the category itself, building awareness of markets you didn’t know you were interested in, priming latent preferences, and making certain purchases feel like natural expressions of identity rather than external influences.

A Different Way to Think About What You Spend

The practical implication for an individual thinking about their own consumption is somewhat unsettling.

Consider the question: “Why did I buy that?” For a growing share of purchases, the honest answer is not “I needed it” or even “I wanted it” in any preconsidered sense. The answer is closer to: “I was exposed to it repeatedly in contexts that felt positive, my behavioral profile was used to determine I was a likely buyer, and the purchase appeared in my awareness at a moment when I was primed to act on it.” That’s not a story about rational consumer choice. It’s a story about attention as a mechanism of manufacturing demand.

The spice rack story is not unusual. It is the default. The person who bought it wasn’t manipulated in any sinister sense; the algorithm simply moved her exposure surface until a product entered her awareness and felt, over time, like something she had always wanted. The line between discovery and manufacture is the thing that’s genuinely hard to locate.

None of this means that every purchase is manufactured or that individual agency has been eliminated. But the distribution of influence has shifted in ways that aggregate spending patterns don’t immediately reveal. The 35% of Amazon revenue attributed to recommendations is a floor, not a ceiling. It doesn’t count the downstream effects on search behavior, category awareness, or the purchases made after a consumer left Amazon and bought elsewhere having been exposed to a product there first.

When economists measure consumer welfare, they typically measure purchasing power: what people can afford to buy. The emerging question is whether purchasing power is the right frame when the primary driver of what people decide to buy is increasingly set upstream of any specific shopping decision. The household budget is still a hard constraint. But the direction in which that budget gets allocated is increasingly being navigated by systems optimized for engagement rather than for consumer welfare.

Closing Observation

Herbert Simon’s original formulation, that a wealth of information creates a poverty of attention, was a scarcity observation. He was pointing out that more information means more competition for the cognitive resource required to process it. What he couldn’t fully anticipate was that this scarcity would itself become the foundation of a multi-trillion dollar economy.

The companies that understood this earliest (Google with search, Facebook with social, Amazon with e-commerce) built the infrastructure of the modern internet around attention capture. Every subsequent platform, from YouTube to Netflix to TikTok, inherited and refined that logic. The result is an economy in which the measurement of value increasingly happens in hours and minutes rather than dollars and cents, with the conversion from time to money happening through channels that most consumers never directly observe.

The next time you notice you’ve spent an hour watching something you didn’t plan to watch, or bought something you don’t quite remember deciding to buy, it’s worth pausing to think about what actually happened in that sequence. Not because there’s a simple intervention available, but because the machinery is better understood when it’s looked at directly. You were, in some measurable economic sense, a raw material. Your time was the input. Your subsequent behavior was the output. The margin in between went somewhere, even if it didn’t go to you.

That’s not a comfortable framing. It’s probably an accurate one.

Data references in this article draw on figures from DataReportal’s Global Digital Reports, Meta and Alphabet annual earnings disclosures, data.ai industry reports, Zuora Subscription Economy Index, Influencer Marketing Hub’s State of Influencer Marketing reports, World Bank poverty estimates, and published platform research from Netflix, Spotify, and Amazon. Readers seeking primary datasets should consult DataReportal, Statista, Sensor Tower, and GWI for consumer behavior statistics, and platform investor relations pages for engagement and revenue metrics.

Charts in this article were produced from primary source data. Chart 1: DataReportal 2024 + eMarketer / Statista country-level ARPU + World Bank 2022. Chart 2: Statista / eMarketer global digital ad spend series + DataReportal annual MAU figures 2016 to 2023. Chart 3: Company quarterly earnings reports (Meta, Snap, Twitter/X 2022 10-K, Alphabet, Spotify, Netflix) for full-year 2023. Chart 4: Zuora Subscription Economy Index 2022 edition + Macrotrends / FactSet S&P 500 aggregate revenue series, both indexed to 100 in 2012. Chart 5: Google Trends keyword series, Amazon BSR data, Pop Mart earnings, NPD Group retail data, and press-corroborated peak-week anchors for Stanley Cup, Prime Hydration, Dubai Chocolate, Air Fryer, and Labubu; signals modelled as Gaussian pulses around verified peaks and independently normalised to show lag structure. Chart 6: data.ai State of Mobile 2024, Sensor Tower US app usage 2023, Comscore Mobile Metrix Q3 2023.