Starting in mobile UA: your first 90 days

The first 90 days in mobile user acquisition should be split roughly into three phases: weeks one to four for learning the accounts and the data, and the vocabulary that goes with them, without touching spend decisions; weeks five to eight for running small supervised tests and building the first independent read of a report; and weeks nine to twelve for owning a defined slice of the budget and proposing one change based on evidence gathered in the first two phases.

The single most common mistake in that window is trying to skip the first phase.

What should you learn in the first four weeks?

A new UA hire's job in month one is comprehension, not action. Spend the first week getting read access to every platform the team touches: the ad network dashboards (AppLovin, Meta, Google, whichever mix the studio runs), the mobile measurement partner (AppsFlyer, Adjust, Singular), the mediation layer if the studio publishes as well as buys, and whatever business intelligence tool sits on top. Don't request write access yet; read access forces you to learn by tracing numbers rather than by changing them.

By week two you should be able to explain, without notes, the difference between cost per install and cost per action; between blended and paid return on ad spend; and between raw installs reported in a network dashboard and the attributed installs the MMP counts. These numbers rarely match and the gap between them is usually the first thing a junior misreads as a problem when it is actually definitional.

Weeks three and four are for the studio's specific playbook: which games in the portfolio drive most of the budget, which channels get most of that budget and why, what the current cost per install and payback targets are, and what the last quarter's biggest win and biggest miss were. Ask a senior teammate to walk you through one full campaign life cycle, from creative brief to launch to the first optimisation decision, using a real historical example rather than a hypothetical one.

What changes in weeks five to eight?

This is where supervised action starts. A good manager will hand a new hire a narrow, low-risk lane: one small geography, one creative slot, one modest daily budget, with a named person checking the account daily. Use this period to build fluency in the reporting cadence itself. Pull the same weekly report for four straight weeks and notice what actually moves week to week versus what looks like movement but is inside normal variance. This is the single fastest way to develop judgement about noise, and it matters more than any course or certification.

Weeks five to eight are also when a new hire should start reading the vendor and industry material the rest of the team already treats as background knowledge, the kind of state-of-gaming reports MMPs and market-intelligence firms publish, platform blog posts about attribution and bidding changes, and the studio's own historical campaign notes. Do not treat any single vendor report as gospel. Read enough of them to notice where they agree and where their numbers diverge, which is usually a sign of different methodologies rather than a wrong report.

What does the back half of the first 90 days look like?

By week nine, a new hire should own something end to end, even if it is small: one channel in one market, one creative refresh cycle, or one recurring report that a manager currently produces. Ownership here means being the person people ask first when something in that lane looks off, not the person who executes instructions without forming a view.

Weeks nine through twelve are the right point to propose a first test. It should be narrow enough to run cleanly and to explain in one paragraph: a new creative concept in one geography, a bid strategy change on one campaign, a change to the store listing screenshots. The test matters less for what it proves than for what it demonstrates about the new hire's habits: a clear hypothesis, a defined success metric agreed before launch, a control where possible, and an honest read of the result even if it does not confirm the hypothesis.

What mistakes do most new UA hires make?

The most common mistake is treating a dashboard number as a fact rather than as an estimate produced by a specific methodology, and reacting to daily noise as if it were a trend. A close second is asking for spend authority before demonstrating the ability to explain why a number moved, which understandably makes managers reluctant to hand over budget. A third is skipping the platform documentation and industry reports in favour of asking colleagues every question, which is faster in the short term and slower over a year, since it never builds independent judgement.

A subtler mistake is optimising a test for a metric that looks good in isolation but does not connect to the studio's actual payback model, chasing a lower cost per install in a channel that reliably brings in players who never convert to payers. Understanding the full funnel from impression to payer, not just the stage closest to spend, is what separates a UA hire who gets more responsibility quickly from one who plateaus at execution.

How do you earn trust with a first test?

You earn trust in UA by being right about small things repeatedly, not by being right about one big thing once. A first test that is modest in scope and clearly reasoned, that gets reported honestly and followed by a sensible next step, will do more for a new hire's standing than an ambitious test that produces an exciting but unreliable result. Managers remember the person who said "the result was inconclusive because the sample was too small, here is what I would run instead" far more favourably than the person who claimed a clean win off a noisy read.

The fastest route through the first 90 days is not speed. It is building, in order, the ability to read the data correctly, then to act on a narrow slice of it responsibly, then to earn the right to more.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• Programmatic DSPs for games, explained
• GTA 6 preview: viewership jumped 35%, plan CPMs