Why app installs change the TikTok setup
Web objectives measure through the pixel on pages the advertiser owns. App campaigns measure through an SDK inside the app and, usually, a mobile measurement partnerAn independent service that attributes app installs and in-app events across ad networks. that decides which network earned each install. That changes what has to be built before a single impression is bought, and what can be read afterwards. It also splits the campaign by operating system, because on iOS Apple's privacy framework limits what any network can see about individual users and returns results with a delay and in aggregate, as the app attribution, MMPs and SKAdNetwork guide explains in detail.
The second difference is economic. An install is not revenue; it is a chance at a user. TikTok's audience installs readily, which keeps the cost per installAverage ad cost of each app install. modest and tempts teams to treat installs as the goal. The app marketing hub frames the job as acquiring users whose lifetime value (LTVTotal value a customer brings over time.) exceeds what they cost, and the setup below is built to push TikTok towards those users rather than towards the cheapest downloads.
Recommended setup for app installs
| Setting | Recommended for app installs | Why it differs from other objectives | Feature page |
|---|---|---|---|
| Objective | App promotion, with the app selected and the store destination set | The result happens in an app store and an app, not on a web page | Campaign structure |
| Measurement | Measurement partner or TikTok SDK integration verified with test installs and events before spend | Delivery cannot optimise or report on events it never receives | App attribution and SKAdNetwork |
| iOS plan | Separate iOS campaigns built to the help centre's guidance for Apple's framework | iOS results arrive delayed and aggregated; web objectives have no equivalent | App attribution and SKAdNetwork |
| Event goal | Installs while learning; an in-app event that predicts retention once it has volume | The install is a means; the retained user is the outcome | Campaign structure |
| Creative | Creator-style screen recordings showing the app in use | Viewers decide on the app they can see working, not on a trailer | Spark Ads |
| Store listing | Screenshots and opening lines that repeat the promise the video made | The store page is the landing page and loses installs on any mismatch | App store optimisation |
Measurement partner or SDK events verified before spend
The help centre documents two routes for app measurement: integrating a mobile measurement partner that forwards installs and in-app events to TikTok, or TikTok's own SDK. Whichever route is used, check it end to end before launch. Install the app from a test device through a test link, complete the events that will be optimised, and confirm that each one appears in TikTok's app event view with the right name. Map the partner's event names to TikTok's standard app events where they fit, since delivery handles standard events more readily. The deep links and app analytics guide covers the in-app side, including making sure the analytics product, for instance Firebase, and the partner agree on what counts as each event.
iOS campaigns planned around Apple's framework
On iOS, TikTok reports campaign results through Apple's framework where users have not given permission for tracking, and the help centre documents how iOS app campaigns are set up for it, including dedicated campaigns and limits on how they are structured. Plan for three consequences. Results arrive later than on Android, so judge iOS campaigns over a longer window. Results are aggregated, so per-ad reporting is thinner. And the in-app events that can be reported depend on the conversion value scheme configured in the partner, which has to be agreed with the product team before launch. Keep iOS and Android in separate campaigns so their different reporting never blends into one misleading average.
Event goal chosen for the action that predicts a retained user
The help centre documents optimisation for installs and for in-app events. Start on installs, because a new app campaign needs volume to learn, but choose the target event at the same time: the action in the first days that most often separates users who stay from users who leave. Product analytics usually knows it: completing setup, a first saved item, a second session. The north star metric and OKRs for app installs page shows how to tie that event to the measure the business runs on. Move the ad group to the event once it is reported often enough to learn from, and expect the cost per install to rise as delivery becomes pickier. The campaign structure guide explains why the goal belongs at the ad group and why it should be changed by adding a new ad group rather than editing a learning one.
Creator-style screen recordings instead of polished trailers
On TikTok, people install apps they have seen working. A screen recording of someone using the app, with their voice over it explaining what they get from it, reads as a recommendation; a polished trailer reads as an advert and is scrolled past. Brief creators to show the moment of value early, the screen where the app does the thing it promises, and promote the posts that hold attention organically as Spark Ads, using the Spark Ads guide for authorisation.
Store listing aligned with the promise the video made
The store page is the landing page for an app campaign, and it has its own conversion rate. If the video promised one thing and the screenshots lead with another, viewers who tapped hesitate and leave. Align the first screenshots and the opening of the description with the strongest creative angle, as the app store optimisation guide describes, and revisit them when the winning creative changes.
A worked app-install setup
Streakly, an invented habit-tracking app
Streakly is a free habit tracker with a subscription. Its product team finds that users who complete a first streak of three days in their first week are four times as likely to be active at day 30. It spends £8,000 a month on TikTok, mostly on Android.
| Setting | Before | After |
|---|---|---|
| Objective | App promotion | App promotion |
| Measurement | Partner integration, installs only | Partner integration with a "first_streak_3d" event mapped and tested |
| iOS plan | iOS and Android in one campaign | Separate iOS campaign, judged over 14 days |
| Event goal | Installs | Installs for 2 weeks, then the first-streak event |
| Creative | Animated trailer | 6 creator screen recordings showing a streak being built |
| Store listing | Screenshots of settings | Screenshots leading with the streak calendar |
Before the change Android installs cost £0.90 and 7% of new users completed a first streak in week one, so each streak-completing user cost about £12.90. After the change installs cost £1.60, but 24% of new users completed a first streak, about £6.70 each, and active users at day 30 rose from 1,100 to 1,650 a month on the same budget.
The campaign looked more expensive on the install metric and became cheaper on the one that mattered. The event goal sent delivery after people who resemble retained users, and the screen recordings and store listing attracted people who already understood what the app was for.
Measuring app installs in TikTok
Ads Manager reports Installs and Cost per install for the install goal, and, for event goals, the in-app events and Cost per event for the chosen action, with other mapped events visible as columns. Read video watch metrics on the app creatives, average watch time and the share watching past the opening, beside install rate, because a creative that holds attention but converts poorly usually has a store listing mismatch. For iOS, read results through the reporting TikTok documents for Apple's framework, allowing for its delay.
Partner-reported installs and TikTok-reported installs will differ, and the business should report from one source, normally the measurement partner, while using TikTok's figures for in-platform decisions. The business measures are retention and value: users active after the first week and month, paying conversion, and cost per retained user beside customer acquisition cost (CACCost to acquire one new customer.). Write the chosen one down as the campaign's KPIA metric tied directly to an objective.. Review cost per install and per event twice a week on Android, iOS weekly once its results have settled, and retention by cohort monthly.
Where the app-install setup goes wrong
The commonest failure is launching before the integration is tested, so delivery learns from nothing and the first budget is spent blind. The second is moving to an event goal before the event has volume, which leaves the ad group stuck in learning. The third is mixing iOS and Android, so delayed iOS results drag the average down and the wrong campaign gets cut. And cheap installs flatter a campaign that brings users who never open the app twice; without retention in the review, the setup optimises for downloads nobody uses.
Common mistakes
- Spending before test events arrive. Verify the partner or SDK integration first.
- Installs as the permanent goal. Move to a retention-predicting event once it has volume.
- One campaign for both operating systems. Their reporting differs; keep them apart.
- Polished trailers. Show the app working, in a creator's hands and voice.
- A store listing that tells a different story. Match the first screenshots to the winning video.
