The Pocket-Sized Market

You tap **Allow** because it is raining.

You are on a pavement you do not know especially well, with a bag that has become heavier in the last six minutes and a train delay behind you. The app is new. It wants your location, precisely rather than approximately, and it would also like to send notifications. There is a little blue dot pulsing on the map where you are standing. You give it what it asks for. A car appears in three minutes, which is to say the proposition proves itself immediately and in full view. You get in. The driver glances at the phone fixed to the dash, then at you, then back at the phone.

The permissions remain long after the rain.

That was the cleanest bargain the smartphone ever offered. You granted a company the right to know where you were and the right to get your attention later, in exchange for help now. Two taps, and the market closed around you.

This mattered less as a cultural anecdote than as a business development. A television set could tell a broadcaster only so much, even with an Audimeter attached. A web browser could tell a search engine a great deal, but mostly while you were at a desk and doing something that already resembled a task. The phone knew when you left the house, how fast you were moving, whether you stopped outside a shop, which app you opened instead of replying to the message you had just read, how long you paused before tapping, what time you usually lost patience, and, later, with a little help from the watch on your wrist, how fast your heart was beating while all this occurred.

It was the first fully-instrumented attention surface in history.

By “fully instrumented” I mean a medium in which almost every action relevant to attention can be logged automatically and linked to a persistent identity: opens, taps, scrolls, pauses, returns, purchases, shares, swipes, location, motion, time of day, device model, operating system, battery level, network quality, referral source, and the increasingly intimate category known in product meetings as “context,” which sounds considerate until you inspect the fields. A web page could record a click. The phone could record the click, the previous three screens, the elapsed hesitation, the place where it happened, and whether you came back after the push notification at 8:12 p.m. on Tuesday performed better than the one at 7:47 p.m. on Thursday.

The convenience story was not wrong. It was merely missing the most important line item.

Mobile did not simply add another screen to the pile. It changed the unit economics of interruption. Where television sold thirty-second claims on a programme, the phone sold tiny claims on the day itself. The crucial innovation was not merely that the device was personal. Radios had been personal; Walkmans intimate; laptops already knew a fair amount about their owners. The innovation was persistence. The phone stayed logged in. It stayed near your body. It stayed reachable by software written by parties you had never met and, in many cases, never paid. It converted the old problem of “How do we reach people often enough to change behaviour?” into a much more modern one: “How often can we reach you before you notice the reaching?”

There was no single villain here, which is one reason the arrangement became durable. There were a series of entirely legible business decisions. Apple sold a premium device and then a software store to make that device indispensable. Google made sure the mobile web and Android would protect search’s tributaries. App developers discovered that charging up front was hard, that subscriptions were harder than they looked, and that advertising remained the old reliable method for extracting money from attention without presenting an awkward invoice to the person supplying it. Carriers wanted data usage. Venture investors wanted growth. Advertisers wanted more certainty than television ever gave them and more scale than search could provide on its own. The phone satisfied everyone except the party whose available mental bandwidth was being carved into smaller, more profitable pieces.

You were not asked to think of yourself in those terms. You were asked whether you wanted turn-by-turn directions, weather alerts, your boarding pass, the package update, the ride home, the digital ticket, the discount code, the family group chat, the reassuring pulse of read receipts, the map pin hovering over a new restaurant, the camera that was there when something worth keeping occurred. Of course you did. The miracle was real. The invoice was simply routed elsewhere.

Smartphones did not begin as advertising machines. They became the ideal substrate for advertising because they began as everything machines. The same slab of glass could be camera, diary, map, newspaper, jukebox, boarding pass, game console, wallet, mirror, and alibi. Once the device became universal, every business wanted a square inch inside it. Once every business wanted a square inch, the old media logic returned on very contemporary terms: inventory, yield, targeting, delivery, attribution. The screen was small. The market surrounding it was not.

You can date the closing of this market with boring precision if you like. The iPhone arrives in 2007. The App Store follows in 2008, turning software distribution into a consumer habit rather than a technical pastime. Then came push. Apple’s push notification service appears soon after, partly as an elegant answer to a practical engineering constraint: early iPhone apps were not supposed to run in the background, and developers still needed a way to tell you something had happened. A message from a friend. A move in a game. A bid on an auction. The phone would deliver the alert; the app could wake only when summoned. Technical compromise has launched stranger empires.

Android ensured scale. Cheaper smartphones spread the habits downward and outward, across classes and countries, until the category stopped meaning “expensive computer in your pocket” and started meaning “the thing you carry by default.” App analytics firms flourished. Mobile ad networks flourished. Attribution firms promised to tell marketers which install came from which campaign and whether the campaign had been worth buying in the first place. Software development kits — SDKs, the little packets of someone else’s code embedded inside an app — multiplied quietly through the ecosystem, collecting events, sending them home, and helping turn each cheerful utility into a reporting surface for several businesses at once.

This is the part of the story that tends to get discussed as privacy, which is fair as far as it goes and insufficient for our purposes. Privacy is the language of personal rights. The phone’s deeper significance sits in the language of market design. Once behaviour can be measured this continuously, attention stops being sold mainly in rough demographic batches or search-defined bursts. It can be sold according to situation. A commuter at 8:17 a.m. near a station is one product. A parent in a supermarket at 5:42 p.m. is another. A teenager awake at 1:13 a.m. on a Friday looks like a third. Time of day becomes a pricing variable. Location becomes a pricing variable. Motion becomes a pricing variable. Later, as watches, earbuds, health apps, and insurance products begin passing along inferences about sleep, activity, or stress, biometrics edge toward the same role. A medium once sold against age and programme environment now begins to sell against circumstance itself.

If this sounds slightly excessive, consider the humble weather app, one of mobile’s earliest and purest commercial forms. Weather is useful. Weather is local. Weather is time-sensitive. Weather produces repeat visits because the sky, unlike your cable provider, changes with some frequency. A phone carrying a weather app therefore knows where you are, gives you a reason to check it multiple times a day, and can surround that glance with advertising for coffee, coats, pollen tablets, umbrellas, flights, groceries, patio furniture, or sinus relief depending on season, place, and hour. Add alert permissions and the app can also decide when to come and fetch you. A storm warning is plainly useful. “Rain starting soon” sits nearby. “Weekend getaway deals for sunnier counties” was always waiting outside.

The lock screen became a genre because it became inventory.

A generation earlier, advertisers bought programmes and hoped the surrounding attention would hold. Search let them buy intent with much greater accuracy, but the format was still constrained by the fact that you had to ask for something first. Mobile opened a third route. A business no longer had to wait for your question or for a broadcaster’s schedule. It could create a reason to appear, then pay for the data and software needed to improve its timing. An app asking to send notifications is making a request that sounds administrative and is actually commercial. It is asking for the right to interrupt you later at near-zero marginal cost.

The oddity is worth lingering over. You bought the device. You pay, directly or through your employer or family, for the data plan that keeps it connected. You charge it with electricity that is not free, despite what technology advertising sometimes implies about physics. Then software producers ask whether they may use this privately funded channel to reach you whenever it suits their retention curve, revenue target, or campaign calendar. If television had mailed a salesman into your sitting room twice an evening and sent you the bill for the petrol, the arrangement might have drawn more notice.

Yet the permissions dialog rarely feels like a negotiation. It appears while you are trying to accomplish something else. You want the ride. You want the photos to back up. You want the boarding gate changes. You want the parcel alert because the parcel contains the replacement charger for the phone that now handles parcels. The request is folded into utility so neatly that refusing can feel churlish, even incompetent. Many people hit **Allow** on the first pass because the demand is bundled with the function. Many more do it because the consequences of delay are immediate and the consequences of consent arrive later, dispersed through the week in manageable doses: a badge here, a buzz there, a little red count glowing like an unresolved obligation.

That dispersal was a commercial masterpiece. Interruptions are easier to resent when they arrive as obvious interruptions. A break in a television programme announces itself. A billboard intrudes by standing where your eyes are likely to land. An email subject line is at least honest enough to queue with other email subject lines. The smartphone learned to style interruption as service. A notification could resemble a favour, a reminder, a social update, a warning, or a token of care. Sometimes it was one. The market thrives when the distinction has to be judged each time.

Product teams noticed the economics quickly. Mobile acquisition is expensive. If you are building an app and paying to install it through app-store ads, social campaigns, influencer sponsorships, or cross-promotion, you are effectively buying the right to be forgotten in a more respectable font. Many people install an app once, open it twice, and never return. The cost of reacquiring them through paid channels can exceed what they are worth. Push notifications offer a cheaper route back. Once you have permission, you can summon them without bidding again in an external market. A message goes out. Some proportion open. Some proportion purchase, subscribe, scroll, watch, or simply restore the habit. The open rate becomes a proxy for success. The reactivation curve moves. Retention inches upward. The quarterly story improves.

You can watch the logic spread from category to category. News organisations, facing the cheerful arithmetic of declining print revenue and low mobile display CPMs, discover that alerts pull readers into the app. First come the serious notices: election result, court ruling, storm warning, market shock. Then the urgent soft things. Then the merely available things with exclamation marks fitted to the front. Retailers send flash-sale pushes because dead stock has always made people inventive. Food-delivery apps discover lunch and late evening as rhythmic opportunities, then refine by postcode, weather, local events, and your previous orders. Language-learning apps shame you gently about streaks. Dating apps inform you that someone may or may not be interested, a message calibrated to exploit the human difficulty of leaving an ambiguous possibility alone.

The lock screen holds all of them in one place, a tiny municipal noticeboard for competing claims on your nervous system.

None of this required malice. It required only one observation: if attention on mobile is measurable and recallable, it can be managed the way inventory was managed in older media. Frequency is no longer a planning estimate. It is a dashboard number. Delivery can be monitored hourly. The creative can be A/B tested. The landing page can be optimised. The notification copy can be personalised down to your first name, your nearest branch, your abandoned basket, or the fact that you usually order on Thursdays after 7 p.m. It is direct marketing with the manners removed.

And it worked often enough to become doctrine.

A small scene from the era will tell you more than a dozen abstractions. In countless product teams, in businesses glamorous and squalid and many in between, some version of the same weekly meeting occurred. The growth chart sags after day three. Paid acquisition costs have risen again. The finance team wants better lifetime value from each install. Someone proposes a win-back campaign. Someone else shows data suggesting that people who enable notifications return at a much higher rate. Because the operating system may give you only one clean shot at the permission, they debate the pre-permission screen first — not the real system prompt yet, but a soft rehearsal in warmer language, designed to explain what you will supposedly miss. The copy changes. The button colour changes. The opt-in rate climbs.

The screen on your phone changes a week later. You see nothing of the meeting.

This is why mobile deserves its own place in the ledger. Search had already taught the market to price declared intent. Social had already taught it to harvest behaviour and make the audience produce the supply. The phone did something more enclosing. It reduced the amount of attention that was economically invisible. Unmonitored time, which used to include queues, commutes, waiting rooms, lifts, ad breaks, toilet cubicles, restaurant pauses, supermarket lines, and the not entirely trivial period between waking and becoming a person, entered the system as addressable inventory.

The death of boredom was a commercial event.

Boredom, before the phone, was simply a condition in which no one was billing against the interval. You looked out of windows. You stared at cereal boxes. You read the back of the shampoo in a bathroom not your own. Adults were left alone with their thoughts for seven minutes at a bus stop. Some of this time was annoying. Some of it was restorative. All of it was difficult to monetise at scale.

Mobile solved the monetisation problem with a mixture of engineering and behavioural luck. The phone was already in your pocket. The network was already on. Feeds became endless, which removed the stopping cues built into pages, schedules, and physical objects. Infinite scroll ensured there was always one more item available without requiring the tiny but meaningful decision to click “next.” Pull-to-refresh added the variable-reward mechanism almost cartoonishly. Swipe down. Let go. See what appears. The motion resembles a slot machine handle closely enough that even product people grew embarrassed.

The psychology is familiar: variable reward does its work at near-zero friction. If sometimes the refresh reveals something unusually pleasing, interesting, enraging, flattering, lucrative, or socially dangerous, the uncertainty itself becomes motivating. You pull again. A feed that updated on a fixed schedule would be easier to ignore. A feed that might contain something decisive at any moment recruits you into checking. Mobile made the check frictionless. Notifications made it unnecessary to remember to check at all.

Suddenly the small fragments mattered. A ten-second glance in a queue, multiplied by a few billion people over a few dozen occasions per day, turns into a media market of ridiculous size. Time spent rises. Opens rise. Ad impressions rise. Purchases happen in app. The ad-supported free game prospers because it can occupy the little ragged edges of the day. The social feed prospers because it can absorb every waiting period not already claimed by work or childcare. The video platform prospers because the cost of filling silence collapses. Entire product categories become viable because boredom has been converted from empty calendar space into salable inventory.

You can see the strategic consequences all over the landscape. Media once fought over evening hours, commute radio, weekend supplements, Sunday night tentpoles. Mobile did not replace those battles so much as annex the territory between them. It taught businesses to think in micro-moments: near me, right now, while waiting, after school pickup, before bed, during the match, when the parcel is due, while the kettle boils. Each small window could be forecast, tested, and sold. The phone turned life’s connective tissue into a schedule.

There is a darker elegance to this than is usually admitted. The market did not need you to spend dramatically more total time with media all at once. It only needed to make previously idle intervals legible and tradable. The day remained twenty-four hours. Mobile found more shelf space inside it.

Once this infrastructure existed, the old distinction between “media company” and “other kind of company” began to blur. A retailer with an app and repeat traffic has audience. A delivery platform with search and ranking has inventory. A marketplace with sellers competing for visibility has an auction. A maps product with local business listings has paid placement waiting impatiently in the wings. A grocery app with sponsored results looks less like software and more like a trade publication with a checkout function. The phone did not merely give existing media firms a new channel. It taught every category with sufficient traffic to imitate media economics.

This is where the chapter’s least glamorous but most important claim comes into view: every app eventually becomes an ad business.

The phrase is deliberately rude. It compresses distinctions people care about. Some apps sell display advertising. Some sell sponsored listings. Some sell promoted placement within search results. Some sell affiliate links, paid recommendation slots, brand partnerships, or “featured” results whose relation to relevance is best described as entrepreneurial. Some do not sell formal advertising at all and yet adopt the governing habits of the ad business: relentless measurement of attention proxies, constant experimentation on prompts and timing, monetisation of visibility, and pressure to increase returns whether or not extra returns correspond to extra value. The labels vary. The gravity does not.

Consider the grocery-delivery app. Gross margins are tight. The search bar is already there because people need to find yoghurt. Suppliers care very much where their products appear. One quarter, under pressure, the platform introduces sponsored placement. You type “cereal.” Familiar boxes rise toward the top with tiny disclosures. The app remains a grocery app. It has also become a media property selling access to intention at the point of decision.

Once you see the grocery example, the rest scarcely needs a guided tour. Maps has sponsored pins waiting just off-stage. Commerce platforms build internal ad markets because merchants will pay richly for visibility at the moment of purchase. Ride-hailing, food delivery, streaming, travel, dating, and fitness all discover, sooner or later, that repeated access to attention makes promotional inventory hard to resist.

This is not some aftermarket defect. The app that solves a recurring problem earns repeated visits. Repeated visits generate data. Data improves targeting. Higher-value placement invites a sales force, or at least a dashboard. A second business grows inside the first, trading on the attention it attracted.

Amazon provides the clearest mature example, though its ambition extends far beyond the phone. A retail site, then an app, becomes one of the largest advertising businesses in the world because product search is a form of intent, seller competition is fierce, and the platform controls the first page. Sponsored products proliferate. Brand budgets migrate. The line between helping you find a battery and selling someone priority in the battery aisle becomes administratively clear and experientially blurry. Mobile intensifies this because the screen is smaller and the ranking positions matter more. Scarcity does wonderful things for yield.

You can see why investors liked the arrangement. Ads are scalable. Software is leverage. Attention is generated by the utility or content people already wanted, then monetised repeatedly. If direct payment proves difficult — and on mobile it often does, because app-store norms trained people to expect low prices or “free” — then advertising offers a familiar release valve. “We’ll monetise later” is easier to say when later can always include selling access to the audience you accumulated without charging them directly.

Meanwhile the operating systems themselves became governors of the market. Apple and Google controlled app distribution, permissions, defaults, and the technical affordances of tracking. Their decisions shaped who could measure what. When Apple later made a public performance of restricting cross-app tracking, it altered the flow of data and damaged parts of the advertising ecosystem, particularly among firms that had grown fat on cheap targeting. The gesture was partly principled, partly competitive, and entirely revealing. A gatekeeper only gets to reprice a market this large if a market is what sits beneath the rhetoric. The phone did not stop being an attention machine because some pipes were rerouted.

A second scene. Somewhere in a newsroom, also somewhere around the same decade, editors argue about push alerts. Traffic from social has become unreliable. Search is useful but episodic. Mobile direct traffic is precious. The app can reach readers without a platform intermediary if they have enabled notifications. One editor wants fewer alerts because overuse cheapens the channel and annoys readers. Another points to the numbers: pushes drive immediate spikes; spikes support advertising; the business is under pressure. They test more alerts. Open rates hold. The threshold shifts.

The next month you receive a notification informing you that a celebrity has “broken silence” about something whose silence had previously improved your life.

Absurd, obviously. Also instructive.

It would be comforting to imagine these as abuses piled on top of an otherwise healthy model. The harder truth is that they are often the healthy model. Product teams are evaluated on retention, time spent, open rate, conversion — proxies that reward more contact. A feature that increases engagement by bringing you back five extra times may be celebrated even if the extra visits leave you feeling vaguely pickpocketed. The dashboard records the return, not the sensation.

You can hear the old television language mutating. Reach survives, though it becomes more individual. Frequency grows more exact. Audience composition becomes cohorts, segments, lookalikes, predicted purchasers, stressed commuters, sports fans in a betting mood, and whatever else the data brokers can stitch together. Recall gives way to attribution. The brand manager's dream of knowing which half of the spend worked inches closer to reality, or at least a serviceable simulation of it.

Once a business learns to think of attention as a continuously measurable input, whole classes of design converge. Streaks, badges, unread counts, autoplay, infinite scroll, recommendation feeds, "don't miss," "still thinking about this?" — the entire soft-pedalled industry of re-engagement belongs to the same family. Their shared feature is that they convert absence into a problem to be solved by renewed contact.

And because the phone is always there, renewed contact can happen almost anywhere. Bedrooms became admissible terrain. So did dinner tables, school runs, funerals, traffic lights, and the little silent customs of daily life that older media had to approach more cautiously. The phone entered places where publishers never had distribution before, then normalised its own presence there by being genuinely indispensable. This is why moral panic misses the mechanics. The problem is not that screens are seductive in the abstract. The problem is that companies discovered a device with high trust, high intimacy, rich measurement, and low marginal cost of interruption, then built businesses around expanding its claim on idle moments.

You can usefully translate the standard permission requests into business English. When the dialog says **Allow notifications**, it means: permit this company to seek re-entry into your attention without paying an external platform for reacquisition each time. **Allow precise location** means: let this company improve the timing, relevance, and commercial value of what it shows or sells to you, including offers tied to place and movement. **Allow tracking** means: help this company connect your behaviour across contexts so attribution improves and future targeting becomes easier to price. The language on the screen is softer. The underlying request is plainer.

Literacy begins there. You do not need to throw the phone into a lake. You need to see the transaction as a transaction. A boarding pass app may deserve notification rights for gate changes. A weather app may deserve location access while in use. A family messaging app occupies a different moral category from a shopping app that wishes to announce “drops” at 10:04 p.m. with the urgency once reserved for air-raid sirens. The point is not purity. The point is to stop reading every prompt as if it were a neutral feature unlock rather than a proposed commercial relationship.

That shift in reading changes more than settings. It clarifies why the phone feels different from earlier media, even when the content looks familiar. Television interrupted programmes. Search monetised needs you declared. Social industrialised participation. Mobile wrapped the whole sequence in a device that sits within arm’s reach for most of your waking life, measures nearly everything relevant to response, and can speak first. Once software learns to reach out rather than wait, attention stops behaving like something you allocate in blocks. It starts to feel like something others spend on your behalf.

The practical consequence is unpleasantly concrete. Every enabled notification, every granted location permission, every silent concession to “always allow” enlarges the set of moments in which your day can be priced by circumstance. That does not make each permission a mistake. It does mean each one has to be read as a budget line. Some interruptions earn their keep. Some are merely cheap. The phone is designed to make the cheap ones look helpful until volume reveals their true character.

By now the dashboard logic is no longer confined to obvious ad businesses. It appears in retailers, banks, airlines, newspapers, schools, health apps, grocery services, meditation products, and productivity tools. Everywhere there is repeat traffic and a quarterly target, the temptation is the same: send one more alert, collect one more data point, turn one more surface into inventory, fill one more patch of nothing. The people making those decisions rarely sound sinister. They sound practical, which is how most durable business models announce themselves.

That should leave you with a slightly altered view the next time a phone lights up on the table. The question is no longer “Is this technology good or bad for me?”, which is too foggy to be useful and too moral to be accurate. The better question is: who is buying the right to arrive here, with what evidence, at whose cost, and by what measure of success? On mobile, those answers are often sitting in plain sight. They are just written in the gentle administrative prose of permissions, preferences, reminders, and updates.

The market has become small enough to fit in your hand. That is why it feels so large.

paid out: broadcast entertainment and shared culture; searchable utility at the moment of need; free publishing, distribution, and affiliation; portable maps, messages, cameras, payments, and coordination carried everywhere took in: audiences sold in bulk; commercially valuable intent sold by the click; unpaid content production, behavioural data, and social graphs; persistent location traces, device identifiers, notification permissions, app-session logs, and the idle intervals turned into inventory measured: ratings, share, and recall; clicks, conversion rate, and quality score; engagement in all its fertile aliases; opens, installs, retention, time spent, movement, and the probability that a person in a place at a time will respond left behind: a culture trained to accept interruption as the price of media; a web organised around commercial legibility; a social sphere ranked by engagement; and the practical disappearance of unmonitored time