It is 9:14 a.m. on a Tuesday. Somewhere, an ad just worked.
A woman in Jakarta taps a TikTok video, opens an app, finishes a purchase. A man in Denver swipes past a banner, ignores it, then installs the same app from a YouTube short three hours later. Somewhere between those two taps, a pixel reports back to a server in Herzliya, gets normalized, deduplicated, run through a fraud check, stripped of personal identifiers, and folded into a number a marketing director in London will look at over coffee tomorrow. The marketing director will not think about that pipeline. That is the whole point.
AppsFlyer is the company that built it. It is a mobile measurement platform - the polite industry term for "the company that decides which ad got the credit" - and it has spent fourteen years becoming the layer of plumbing that the mobile economy now quietly runs on. Today it employs around 1,300 people across 25-ish offices, books more than half a billion in annual revenue, and reports that 80,000 brands send their app data through it. You have almost certainly never noticed the company. That is also the whole point.
Numbers AppsFlyer's PR team rarely puts on a t-shirt. We put them in a box anyway.
In 2011, mobile advertising had a small, embarrassing secret.
Nobody really knew if it was working.
Brands were lighting billions of dollars on fire to push apps onto the new shiny rectangles in everyone's pockets. The ad networks selling the inventory were also, conveniently, the ones counting the installs. This is a bit like asking the chef to grade the meal. The chef usually likes the meal.
What was missing was a referee - a neutral party, paid by the advertiser, that could look at every tap, every install, every in-app purchase, and say which channel actually caused it. The technical name for this is "attribution." The political name is "an inconvenient truth machine."
Oren Kaniel, a software engineer with a habit of staring at problems until they cracked, saw the gap. So did his co-founder Reshef Mann. They were not the first to notice it. They were just the first to be both stubborn enough and Israeli enough to build it as a piece of infrastructure rather than a feature inside an ad network.
An apartment in Herzliya. A judgment call about independence.
AppsFlyer was started in 2011 out of a small apartment north of Tel Aviv. The first money came from the Microsoft Ventures Accelerator - a few rooms, free coffee, and the gentle suggestion that the founders should probably go raise more. They did, but slowly, and only from investors who agreed with the central architectural decision: AppsFlyer would never sell ads. It would never run a media network. It would never own a piece of the inventory it was measuring.
This sounds reasonable. In 2011 it was nearly heretical. Every other promising ad-tech company in the world was racing to bolt measurement onto a buying platform, because measurement-plus-media is where the margins live. AppsFlyer's bet was that brands would eventually want a Switzerland - and that Switzerland, run patiently, would compound into a moat.
It took the better part of a decade to look obvious.
From one SDK to a measurement cloud.
What started as an attribution SDK - a small bit of code app developers drop into their builds - has grown, like most successful infrastructure, into a suite. Today the platform spans seven product families that all share the same backbone of independently-collected, privacy-treated mobile data.
Attribution
The original product. Tells brands which campaign actually caused an install or a purchase. Still the spine.
Audiences
Privacy-safe segmentation that activates across Meta, Google, TikTok and roughly seven thousand other partners.
Protect360
Real-time ad fraud detection. Because where there is ad money, there is someone in a basement faking installs.
OneLink
Deep linking that routes users to the right screen on the right device. Quietly responsible for fewer broken app links than you think.
Data Collaboration
Clean-room style environments where retailers and publishers share first-party data without exposing PII.
CTV Measurement
Attribution for the connected-TV ad you watched while half-doing the dishes.
Incrementality & SKAN
Tools for measuring whether ads cause behavior, or just notice behavior that was going to happen anyway.
AI Copilots
Plain-language data queries on top of the warehouse. Marketing analytics without a SQL prompt.
Fourteen years, told in headlines.
Eighty thousand customers walk into a measurement platform.
AppsFlyer's customer logos read like a department-store directory crossed with a Cannes Lions guest list: HBO, NBC, Macy's, Alibaba, Activision, eBay, Tencent, Nike. Most of them are not announced. The few that are tend to come up at conferences and in customer-story videos that have the unmistakable energy of a marketing director who has finally been allowed to sleep through the night.
Revenue, the boring upward line
Reported ARR/revenue, in USD millions. The kind of chart investment bankers like to print.
Estimates blended from press disclosures and CEO posts. AppsFlyer rarely discusses revenue in public; when it does, it tends to be on its own blog and in Hebrew first.
The partnership list is the second proof. AppsFlyer is a certified measurement partner for Meta, Google, TikTok, Snap, X, Pinterest and Reddit - which is to say, all the places people scroll while pretending they are not scrolling. Salesforce is also a strategic investor, which made a lot of partner integrations move surprisingly fast.
Privacy-first measurement, an unfashionable phrase finally in fashion.
For most of the 2010s, "privacy-first" in ad tech meant "we will get around to it." Apple's 2021 changes to the IDFA - the unique device identifier that powered a decade of mobile attribution - made it suddenly mean something else: a constraint. AppsFlyer had spent years preparing the ground, partly out of conviction and partly out of the entirely respectable instinct to not be on the wrong side of regulation. When SKAdNetwork landed, AppsFlyer was one of the first measurement platforms to ship a coherent answer.
The company's stated mission today is to deliver AI-powered, privacy-first measurement that turns marketing data into growth. This is the kind of sentence that gets written by a committee. The version on the engineering whiteboard is closer to: collect less, model more, never sell out the user. Both are true.
Connected TV. Retail media. AI copilots. Same job, harder math.
The mobile measurement war is, in most senses that matter, over. AppsFlyer won the part of it that scaled. The interesting question for the next decade is whether the same independent-referee model can travel - to connected TV, where the ad units are bigger and the identity is messier; to retail media networks, where Amazon, Walmart and Target want their own little ad businesses; to AI copilots that can answer a marketing question in English instead of SQL.
This is also where the company starts to look strange compared to its 2011 self. AppsFlyer now talks fluently about data clean rooms, incrementality modeling, audience activation, and zero-budget marketing analytics. It runs on a stack that mixes Kafka, Spark, BigQuery and Kubernetes. It ships models trained on its own pipeline. It has, in the unromantic way of all good infrastructure companies, become an enterprise software vendor that happens to have started in marketing.
Whether it stays independent or gets sold to a private-equity buyer at three-billion-and-change, the architecture is the architecture. Someone has to be the referee. AppsFlyer is, for now, still the one with the whistle.
9:14 a.m. on a Tuesday, one year later.
The woman in Jakarta opens a different app. The man in Denver, who finally installed, opens his for the third time this week. The pixel still flies to Herzliya. The marketing director in London still looks at her dashboard over coffee. None of them know each other. None of them think about the plumbing.
That is what AppsFlyer built. A piece of infrastructure so calm that an industry stopped arguing about who got credit and started arguing about how to grow. The ads are louder. The referee is quieter. That is how you know it is working.