Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Playbooks

App Marketing Operations Playbook

An operating playbook for running app marketing week to week: the store listing and reviews, releases, install campaigns, messaging and cohort reviews.
Playbookintermediate12 min readUpdated

What this playbook covers and when to run it

The playbook covers the steady state of marketing a live app: the activation event and the measurement chain, the store listing and its tests, ratings and review replies, marketing for each release, install campaigns, the push and in-app messaging programme, and the monthly reconciliation and cohort read that judge it all. It suits any team with an app in at least one store and more than one way of bringing users in.

It does not teach the methods. The mobile marketing pillar explains the field and orders the first quarter of work for a team with a live app and no programme; this playbook takes over once that first plan is done. The app marketing, messaging and measurement guides linked in each phase own their methods, and the pages under mobile marketing troubleshooting diagnose single symptoms, such as installs that never activate or notifications switched off, and add prevention routines. Email journeys for app users run through the lifecycle email playbook, and listing, permission and message tests through the experimentation programme playbook.

Run it when releases ship with no marketing attached, when reviews are read only when the rating falls, when install campaigns are judged on installs, when nobody can explain why the store console and the measurement partner disagree, or when push messages go out without anyone checking who has switched them off. Like every playbook in the playbooks library, it is a starting shape: a small team may hold several roles, and the rhythm stretches or tightens with the release schedule and the spend.

Roles and responsibilities

  • App marketing lead. Owns the programme: the activation event's definition, the rhythm, the monthly decisions on sources and budget, and the quarterly plan for listing tests and messaging.
  • Store listing owner. Owns the default listing and custom listings in both stores, keywords, screenshots and localised metadata, and runs listing experiments.
  • Release owner. Usually in product or engineering. Tells marketing what each release contains and when, owns release notes with the product marketer and confirms deep links and tracking survive each build.
  • User acquisition manager. Owns install campaigns across networks: campaign types, optimisation events, creative, budgets and the launch checks, and brings source-level results to the monthly read.
  • Messaging owner. Owns push and in-app messages, the permission requests and onboarding messages, topic preferences, caps and quiet hours, and re-engagement.
  • Analyst. Owns the event taxonomy, the measurement partner set-up, the monthly reconciliation and the cohort read by source, and keeps definitions the same across reports.
  • Support lead. Reads review themes alongside support contacts, supplies answers for review replies and routes bugs to product with the app version.

The AARRR framework is a useful way to check the split: acquisition sits with the user acquisition manager and the listing owner, activation and retention with the messaging owner and product, and the analyst reads them together. Write the owners down per phase, because the job most often orphaned is the review reply routine, which belongs to no team by default. Where an outside provider is to hold any of these roles, the guide to hiring and evaluating an app marketing agency covers which accounts and records stay in the business's name, which questions to put to a proposal and how to check the provider's reports.

Cadence, inputs and outputs

The rhythm below is a starting point. Tie the per-release checks to the actual release schedule, and set the weekly, monthly and quarterly reviews by how much the team ships and spends, not by a norm.

CadenceOwnerInputsOutputs
Per releaseRelease owner, listing ownerRelease contents and date, the route table for deep linksRelease notes, a listing checked against the release, deep links tested before campaigns use them
WeeklyListing owner, support lead, user acquisition managerNew reviews, installs and activations by sourceReplies sent, themes routed to product, campaigns with broken links or rejected ads fixed
MonthlyAnalyst, app marketing leadStore, partner, network and analytics counts; cohorts by sourceA reconciliation table with explained gaps; budget moved between sources
QuarterlyApp marketing lead, listing owner, messaging ownerTest results, opt-out and preference data, cohort trendsA listing test plan, messaging changes and a refreshed plan for the next quarter

The monthly decisions belong in the team's normal reporting, so app results sit beside web and other channels. The weekly view is operational; it catches breakages and replies to customers, and leaves judgement on sources to the monthly read, when the numbers have settled and been reconciled.

The playbook in phases

  1. Agree the activation event and the measurement chain

    The job here is one activation event, the analytics and attribution set-up, and deep links that land users where the message promised. The guide to finding the app activation event chooses one event from the app's own data by comparing retention for users who did and did not take each candidate action, checks its volume and timing against what campaigns and iOS postbacks can carry, and writes one shared definition into the tracking plan. The guide to app attribution, MMPs and SKAdNetwork explains what a measurement partner does, postbacks, privacy limits and modelled results. The deep links and app analytics guide covers universal and app links, deferred deep links for new installs and the event taxonomy. The gate: every team reports Activation Rate on the same definition, and the analyst signs off the chain.

  2. Run the listing and its tests

    Keywords, screenshots and custom listings are kept current and tested one change at a time. The app store optimisation guide covers store search and browse, keyword research per storefront, titles and screenshots, localised listings and the stores' own listing tests, which it describes as Apple's product page optimisation and Google Play's store listing experiments. The guide to custom store listings for campaigns builds variants matched to an ad's promise, and advises checking the current App Store and Play Console documentation before building, because the details change. The app store screenshot test designs a test read on install conversion per traffic source. Before any listing change is submitted, the app store listing checklist checks text, screenshots, privacy details and attribution links. The listing owner keeps a quarterly test plan with one hypothesis per test.

  3. Reply to reviews and route what they say

    Rating prompts are timed to the right moment, replies run on a routine and themes go to product and support. The guide to app ratings and review replies chooses the moment of completed value for the system rating prompt, sets who is asked and how often, and runs triage, replies, routing to product with the app version and tracking the rating by version and country. It records, as the ASO guide states from the stores, that both stores provide a system prompt and prohibit incentivising ratings or filtering who is asked by mood. The listing owner and support lead read new reviews weekly; recurring themes go to the release owner with the versions they affect.

  4. Ship each release with marketing

    Release notes users read, listing changes that match the release and links tested before campaigns use them. The guide to writing release notes and a changelog writes notes that lead with the change, make honest claims and go where users look. When a release changes what the app does, the listing owner checks the screenshots and description against it. The guide to testing app deep links runs a route table and a matrix by platform and context, including logged-out, missing-item and deferred cases, before any campaign or message relies on a link. The gate: the release owner confirms tracking and links passed on the release build; anything that failed is fixed before campaigns point at it. The route table passed to the release owner is the list those link tests are checked against.

  5. Run install campaigns

    Campaigns are judged on activated users, with launch checks and web-to-app routes where they fit. The app install campaigns guide covers campaign types by network, optimising to installs or in-app events, creative, budgets and pacing, the listing as the landing page, and install fraud and quality. The app install campaign launch checklist checks the measurement partner and SKAdNetwork set-up, the listing and deep links, the optimisation event, creative and payback before each launch. The web-to-app journeys guide moves existing web visitors and customers into the app with links to the matching screen, often a cheaper source than paid installs. The user acquisition manager reports CPI beside cost per activated user, never alone.

  6. Run the messaging programme

    Permissions are asked at a moment of value, topic preferences are honoured and re-engagement stays within caps. The push and in-app messaging guide covers earning the permission, message types and triggers, timing, frequency caps and quiet hours, deep links and testing. The onboarding and permission requests guide designs the first session, the path to first value and soft prompts that tell the truth; the push permission prompt test tests when and how to ask, read on Push Opt-In Rate with retention and uninstall guardrails. The notification preferences guide groups messages into topics people can switch off one by one, and the guide to re-engaging lapsed app users defines a lapse, picks the channel that can still reach each person and measures returns against a holdout. The messaging owner reviews opt-outs and switched-off topics each quarter and retires messages that cause them.

  7. Reconcile installs and read cohorts

    Store, attribution and analytics counts are reconciled, and money moves towards sources whose users stay. The guide to reconciling app install numbers explains why store consoles, the measurement partner, ad networks and analytics count different events, builds a monthly table, marks the gaps expected and sets a tolerance from the team's own history, then investigates when a gap moves, checking SDK releases and integrations first. The cohort analysis tutorial builds the Cohort table by install month and source with the retention or revenue measure the team chose. The analyst brings both to the monthly decision below.

Moving money between sources: the monthly decision

The monthly read exists to answer one question: which sources bring users who stay, and does the budget follow them? The decision is the app marketing lead's, on the analyst's reconciled numbers and the user acquisition manager's knowledge of each campaign.

What the cohort read showsUsual moveWho actsCheck first
A source with cheap installs and few activationsReduce budget or change the optimisation eventUser acquisition managerThe installs not activating diagnosis and install quality
A source whose users activate and stayHold or increase in steps, watching costUser acquisition managerThat its counts reconcile and are not duplicated
A source whose numbers moved without a campaign changeTreat as a measurement question before a budget oneAnalystSDK releases, integrations and the change log
Store conversion falling on one sourceFix the landing page before the budgetListing ownerThe store conversion diagnosis
Rising cost per install across sourcesRead the auction and the listing before cuttingUser acquisition managerThe rising cost per install diagnosis

Two rules keep the decision honest. A source is judged only on numbers that have been reconciled, because a gap that moved for a measurement reason looks exactly like a change in performance. And a source is judged on cohorts old enough to show the retention measure the team chose, so a new campaign is not cut before its users have had the chance to come back.

Illustrative example — not a real result

A home workout app settles into a rhythm

Gavrinko Fitness, an invented home workout app, spent its first year in launch mode: every release a scramble, reviews read when the rating dropped and campaigns judged on installs. The team of 5 adopts the playbook. Its activation event becomes "finished a first workout within 3 days of install", chosen because those users were far more likely to still be active at day 30. Each fortnightly release now ships with notes users read, a listing check and a deep link test of the 12 routes campaigns use. The weekly review routine tags around 60 reviews; a theme of complaints about workouts not syncing after one update goes to the release owner with the version number and is fixed in the next build. The monthly reconciliation shows the store console counting 14% more installs than the measurement partner, a gap the team explains as downloads never opened and re-installs, and records as its expected range. The cohort read then shows one network delivering installs at half the cost of the others but with only 9% of its users finishing a first workout, against 31% from search. The app marketing lead moves budget from that network to search and the web-to-app route, and the cost per activated user falls over the next quarter while total installs dip slightly.

Measuring the playbook

The playbook works when the app is marketed steadily between launches and money follows users who stay. Operating signals show whether the rhythm is holding: whether every release shipped with notes, a listing check and passed deep link tests, whether reviews were read and replied to each week, whether the monthly reconciliation ran and every gap outside the team's own tolerance was explained, and whether the quarterly listing test plan and messaging review happened.

Outcome signals come from the app's own data, read as trends against its own history rather than against figures from other apps: activation rate and early retention by source, cost per activated user, store conversion by traffic source, the rating by version and country, opt-in and opt-out by topic, and returns from re-engagement against a holdout. Report them in the team's monthly pack beside web and other channels, so app decisions are made on the same evidence as the rest of marketing.

Common mistakes

  • Judging campaigns on installs. An install is not a user. Report cost per activated user beside cost per install, and decide on the first.
  • Releases with no marketing. A release that changes the app without changing the listing, the notes or the links leaves the store describing a different product.
  • Untested deep links. A campaign that lands people on an error screen wastes every click. Test the routes on each release build before campaigns use them.
  • Reading reviews only when the rating falls. By then the problem has run for weeks. Read and tag reviews weekly and route themes with the version.
  • Treating counts that disagree as errors. The sources count different things. Reconcile monthly, expect a gap and investigate only when it moves.
  • Cutting a source on unreconciled numbers. A measurement change can look like a performance change. Reconcile first, then decide.
  • Messages that cost permissions. Push sent without caps, topics or a reason to read it teaches people to switch notifications off. Review opt-outs by topic and retire the messages that cause them.

Frequently asked questions

When should an app team move from a launch plan to an operating playbook?

Once the app is live and the first plan's work is done: the listing is built, measurement is in place and the first campaigns have run. From then on the work repeats with every release, every week and every month, and it needs owners and a rhythm rather than another one-off plan.

Why judge install campaigns on activated users rather than installs?

An install says someone downloaded the app; it does not say they got anything from it. The app's own activation event, chosen from its retention data, shows which installs became users. Sources that bring cheap installs and few activations cost more per real user than their install price suggests, and the monthly cohort read is where that shows.

How often should install numbers be reconciled?

This playbook suggests monthly, as the reconciliation guide does, but that is a starting rhythm. A team spending heavily across several networks, or one that has just changed its measurement set-up, may check more often; a small app with one source may need less. What matters is that the gap between sources is watched and investigated when it moves.

Terms used on this page

Frameworks

Where this page is used

Sources

  1. App Store — Apple Developer
  2. Google Play Console Help — Google

Related