Targeting on TikTok for app installs is fundamentally different from what most performance marketers learned on Meta. TikTok's algorithm is incredibly powerful at finding your ideal users, but only if you feed it the right signals.
The platform's recommendation engine was built to surface content to people who will engage with it, and that same engine powers ad delivery.
In this guide, you will learn exactly how to use TikTok’s targeting options (interest, behavior, custom audiences, lookalikes, and broad) strategically, when each one matters, and why creative is the most important targeting lever you have.
TikTok's targeting playbook looks nothing like what works on Meta, and marketers who treat the two platforms as interchangeable consistently underperform.
Prerequisites: You should have a TikTok Ads Manager account set up with the TikTok SDK or MMP (AppsFlyer, Adjust, etc.) integrated for app install tracking. You need at least a basic understanding of your app's core user persona.
Ideally, you have some install volume already (even from other channels) so you can seed custom audiences. A minimum test budget of $5,000 to $10,000 is recommended to meaningfully test multiple targeting approaches.
Page Contents
Step 1: Understand TikTok's Algorithm Before You Touch Targeting Settings
TikTok's ad delivery algorithm is content-first, not audience-first. Unlike Meta, where the system heavily indexes on user profile data and behavioral history, TikTok's algorithm learns primarily from how users interact with content in real time. This means your targeting settings are more like guardrails than GPS coordinates.
The algorithm will quickly expand beyond narrow targeting if it finds converting users elsewhere.
Recognize TikTok's signal differences from Meta
Meta has years of purchase and app install data per user. TikTok’s engagement signals drive performance differently, with less historical conversion data but incredibly rich watch time, replays, shares, and comments. This is why creative quality disproportionately drives performance on TikTok. The algorithm watches how people interact with your specific ad and uses that as targeting signal.
Start with the right optimization event
Always optimize for app installs or downstream events (like purchases or registrations), not clicks. TikTok's algorithm needs conversion signals to find the right users. If you optimize for clicks, you will get clickers, not installers. With iOS campaigns, make sure your MMP is properly passing postbacks so TikTok can optimize effectively.
In our experience, switching from click optimization to install optimization can drive dramatic CPI improvements overnight, with zero changes to targeting or creative. The optimization event is the single most impactful 'targeting' decision you make.
Step 2: Map Your Targeting Strategy to Your Funnel Stage
Not every targeting type works for every situation. The right approach depends on your budget, data maturity, and campaign objective.
Here is the framework we use at RocketShip HQ: interest and behavior targeting for cold prospecting with limited data, custom and lookalike audiences for scaling with existing user data, and broad targeting for maximum scale once you have strong creative and conversion signals.
Use interest targeting for early-stage prospecting
TikTok offers interest categories (like 'Gaming,' 'Finance,' 'Health & Fitness') that are useful when you are launching a new app or entering a new market with no existing user data. Select 3 to 5 relevant interests per ad group. Do not over-layer interests, as this shrinks your audience and starves the algorithm of learning signal.
Layer in behavior targeting for sharper intent
Behavior targeting lets you reach users based on actions they have taken on TikTok in the last 7 or 15 days: video interactions (liked, commented, shared), creator following patterns, and hashtag interactions. This is powerful for apps where user intent aligns with specific content consumption patterns.
For example, a fitness app can target users who recently engaged with fitness creator content.
Deploy custom audiences for retargeting and seeding
Upload your existing user lists (from your MMP or CRM) to create custom audiences. Use these primarily for two things: retargeting users who installed but did not convert on a downstream event, and as seed audiences for lookalikes. Exclude existing installers from prospecting campaigns to avoid wasting spend.
Build lookalike audiences from your best users
Create lookalikes from your highest-value users (purchasers, D7 retained users), not just all installers. TikTok offers narrow, balanced, and broad lookalike options. Start with balanced (the middle option) for the best trade-off between reach and quality. You need at least 10,000 users in your seed audience for reliable lookalikes.
Industry patterns suggest interest targeting CPIs tend to run meaningfully higher than lookalike or broad targeting once the algorithm has sufficient data. Interest targeting is a starting tool, not an endpoint.
Step 3: Embrace Broad Targeting as Your Scaling Engine
This is counterintuitive for marketers coming from other platforms, but broad targeting often outperforms interest-based approaches on TikTok at scale. Broad targeting means no interest, behavior, or audience selections—you let TikTok’s algorithm do what it does best: find users based on content engagement signals.
The catch is that broad only works when your creative is strong enough to act as your targeting.
Set the right conditions for broad targeting
Before going broad, ensure you have at least 50 conversions per week per ad group (TikTok's recommended threshold for exiting the learning phase). Your pixel or SDK needs to be firing reliably. And you need creatives that clearly communicate who your app is for.
Use age and gender as your only guardrails
When running broad, you can still set age and gender constraints. If your app skews heavily female 25 to 44, keep those parameters. But remove interest and behavior targeting entirely. Geographic targeting should obviously remain in place.
Across app categories like gaming, fintech, and health, broad targeting consistently outperforms stacked interest targeting once conversion volume is sufficient to move past the learning phase. The key is always creative quality. Understanding how to scale TikTok spend while maintaining performance requires respecting the algorithm’s learning phase.
Step 4: Treat Creative as Your Primary Targeting Mechanism
On TikTok, creative is targeting. The algorithm serves your ad to users who are likely to engage with that specific piece of content.
This means a casual gaming ad featuring puzzle gameplay will naturally reach puzzle enthusiasts, while the same app advertised with a competitive leaderboard angle will reach a completely different audience segment. This is the most important concept in TikTok app install marketing.
Design creatives for specific audience psychographics
Rather than relying solely on TikTok’s targeting settings to find your audience, encode your audience targeting into ad scripts. As Bastian Bergmann from Solsten explained on the Mobile User Acquisition Show, psychology-based creative changes can massively outperform algorithmic optimization alone.
For Solitaire Klondike, shifting ad copy from 'train your brain' to 'hardest solitaire game' based on psychological profiling improved IPM from 0.97 to 2.4. The targeting settings were identical. The creative did the targeting.
Test emotional angles competitors ignore
Most advertisers default to the same emotional territory. As Gonzalo Fasanella, CMO at Tactile Games, shared about story-driven ads for Lily's Garden, they found success exploring 'sadness, anger, anxiety' when 90% of competitive ads relied on 'funny or cute.' Different emotional hooks attract different audience segments, even with identical targeting settings.
Match creative format to audience behavior
The ideal TikTok ad length for app installs varies by audience. Younger audiences respond to faster 15 to 21 second creatives, while older demographics converting on higher-consideration apps may need 30 to 45 seconds. Your format choice is a targeting decision.
Run the same ad group with broad targeting but different creatives, and you will see completely different audience profiles in your analytics. We've observed that creative changes alone—with no targeting adjustments—can dramatically shift the gender and age composition of the audience the algorithm finds.
Step 5: Structure Your Ad Groups to Avoid Asset Stuffing
One of the most common mistakes on TikTok is dumping all your creatives into a single ad group. This is called 'asset stuffing,' and it prevents the algorithm from properly matching creatives to audience segments.
As discussed on the Mobile User Acquisition Show's episode on asset stuffing, placing all creatives in one ad set without thematic separation makes it impossible for the algorithm to identify appropriate audience segments.
Separate ad groups by creative theme or audience angle
Create distinct ad groups for each major creative concept or audience angle. For example: one ad group for 'competitive gameplay' creatives, another for 'relaxation/stress relief' creatives, and a third for 'social/community' creatives. Each ad group should contain 3 to 5 creatives that share a similar audience appeal.
Determine the right number of creatives per ad group
Overloading an ad group dilutes spend across too many creatives and delays learning. We have found that right number of creatives per ad group is typically 3 to 5 for optimal algorithmic learning. This gives the system enough variety to test while concentrating enough budget on each creative to generate statistically meaningful results.
Think of each ad group as a hypothesis about an audience segment. The creative theme defines the audience, and the targeting settings provide loose guardrails. This mental model prevents the asset stuffing trap.
Step 6: Set Up a Structured Testing Framework for Targeting
Do not test targeting types randomly. Use a structured approach where you run parallel ad groups with different targeting methods but identical creatives, so you can isolate the impact of targeting. Allocate roughly 70% of budget to your proven best-performing targeting approach and 30% to testing new ones.
Run a targeting ladder test
Launch three ad groups simultaneously with the same creative set: one with interest targeting, one with lookalike audiences, and one broad. Run each for at least 7 days with a minimum of $100/day per ad group. Compare CPI, Day 1 retention, and ROAS. In our experience, this test almost always reveals that broad or lookalike wins at scale.
Watch for hidden testing costs
As outlined in an episode about pitfalls of testing AI-powered creatives, more creative output requires proportionally larger test budgets. The same applies to targeting tests. If you split $500/day across 10 ad groups, none of them will exit the learning phase. Be disciplined about testing fewer things with adequate budget rather than many things with insufficient budget.
In our experience, TikTok ad groups need a meaningful minimum spend threshold before you can make any reliable judgment about targeting performance. Cutting a test at $50 in spend is one of the most expensive mistakes you can make, because you learn nothing and waste the entire amount.
Step 7: Optimize Based on Downstream Metrics, Not Just CPI
Cheap installs mean nothing if they do not convert downstream. Different targeting approaches often produce dramatically different user quality. Always evaluate targeting performance on Day 1 and Day 7 retention, ROAS (Day 7 and Day 30 if your payback window allows), and cost per key in-app event (registration, purchase, subscription trial).
Connect TikTok data to your MMP cohort reports
Break down user quality by ad group in your MMP. You will often find that the ad group with the lowest CPI has the worst retention, while a slightly more expensive targeting approach delivers 2x the ROAS. Mobile apps commonly see broad targeting deliver cheaper CPIs while lookalike targeting delivers meaningfully better downstream ROAS—making it essential to evaluate both dimensions before reallocating budget.
Shift budget based on efficiency, not volume
Once you have 7 to 14 days of downstream data, reallocate budget toward the targeting approach that delivers the best cost per quality event. If your app monetizes through subscriptions, optimize toward cost per trial start, not cost per install.
Limit the KPIs you show to creative teams. Tactile Games found that showing only 2 KPIs to their creative team prevented analytics-driven bias in creative development. The same principle applies to targeting decisions: pick your north star metric and optimize toward it.
Common Mistakes to Avoid
- Over-narrowing targeting on launch: Stacking multiple interest categories, behaviors, AND age/gender restrictions creates audiences too small for TikTok's algorithm to optimize. We've seen advertisers launch with very small audiences and wonder why CPIs are multiples above their benchmark. On TikTok, start broader than you think you should.
- Ignoring creative-as-targeting: Spending hours fine-tuning interest targeting while running generic creatives is backwards on TikTok. Your creative communicates to the algorithm who should see your ad. A mediocre creative with perfect targeting will always lose to a great creative with broad targeting.
- Asset stuffing across themes: Placing a UGC testimonial ad, a gameplay ad, and a story-driven emotional ad in the same ad group confuses the algorithm. It cannot find a coherent audience for three completely different messages. Separate creative themes into distinct ad groups.
- Killing ad groups before they exit the learning phase: TikTok needs approximately 50 conversions per ad group to exit the learning phase. If your CPI is $2, that means $100 of spend minimum. Many advertisers panic at volatile early CPIs and kill ad groups after $20 to $30 in spend, never giving the algorithm a chance to optimize.
- Copying Meta targeting strategies directly to TikTok: Meta's detailed targeting and interest stacking logic does not transfer. TikTok's interest categories are broader and less precise. The platforms have fundamentally different signal architectures. Treat TikTok as its own channel with its own rules.
Targeting on TikTok for app installs comes down to a simple hierarchy: get your optimization event right, structure your ad groups by creative theme rather than audience segments, and let your creative do the heavy lifting as your primary targeting mechanism.
Start with interest or lookalike targeting when you have limited data, then graduate to broad targeting as your conversion volume and creative library mature.
The campaigns that consistently win on TikTok are the ones where teams invest the majority of their effort into creative strategy rather than targeting settings. Test systematically, measure on downstream quality metrics, and remember that on TikTok, your ad is your audience signal.
Looking to scale your mobile app growth with performance creative that delivers results? Talk to RocketShip HQ to learn how our frameworks can work for your app.
Not ready yet? Get strategies and tips from the leading edge of mobile growth in a generative AI world: subscribe to our newsletter.
Related Reading
- TikTok Ads for app growth: the complete guide provides the comprehensive overview of the platform’s unique advantages for mobile marketers.
- Best call-to-action buttons for TikTok
- How Many Creatives Do You Need Per TikTok Ad Group?
- What Is the Ideal TikTok Ad Length for App Installs?
- TikTok Ads for app growth: the complete guide
