Hiring a performance marketing agency for apps: the KPIs, reporting and fees to agree up front
Performance marketing for apps means paid users who pay you back on schedule. Here's how to set KPIs, what reports to demand, and how agency pricing models shape behavior.
- Judge app performance marketing on CPA, cohort ROAS and payback period, not on installs or CPI.
- Pick the primary KPI by business model and write the target and source of truth into the contract.
- Expect weekly channel and creative reports plus monthly cohort tables and a budget recommendation.
- Percentage-of-spend fees reward spending more; a retainer or hybrid with an outcome bonus aligns better.
"Performance marketing" gets used for anything with a dashboard. For an app, it should mean one thing: you pay to acquire users, and the money comes back within a time frame you agreed in advance. A performance marketing agency for apps should be judged on cost per paying user, return on ad spend and payback period, not on impressions, clicks or even installs.
This guide sets out the KPIs to put in the contract, how a good agency bids and reports, the pricing models you'll be offered and the incentives each one creates, so you can compare proposals on what actually matters.
What "performance" should mean for an app
Three numbers decide whether paid acquisition is working:
- CPA (cost per action): what you pay for the event that signals real value, such as a trial start, first purchase, subscription, funded account or completed booking. Not the install.
- ROAS (return on ad spend): revenue from a cohort of users divided by what you spent to acquire them, measured at fixed points such as day 7, day 30 and day 90.
- Payback period: how long until a cohort's cumulative revenue (or gross margin) equals its acquisition cost. This is the number your CFO or investors care about.
Installs and CPI are still useful diagnostics. They tell you whether creatives and store listings are working. But an agency that reports cheap installs while paying users stay flat is not doing performance marketing.
Pick KPIs that match your business model
| App model | Primary KPI | Secondary KPIs | Typical payback lens |
|---|---|---|---|
| Subscription | Cost per trial or cost per paid subscriber | Trial-to-paid rate, D30 ROAS, refund rate | Months to recover CAC from net subscription revenue |
| Ecommerce / D2C | Cost per first purchase | D30 ROAS, repeat purchase rate, average order value | First order margin plus expected repeat orders |
| Fintech / lending | Cost per funded account or disbursal | KYC completion, approval rate, early repayment | Margin per funded customer net of credit losses |
| Marketplace / on-demand | Cost per first completed order | Repeat order rate, order frequency, supply-side CPA | Contribution margin per order over 60–90 days |
| Ad-supported / games | Cost per retained user or D7 ROAS | D1/D7 retention, ARPDAU | Ad revenue curve by cohort |
Write the primary KPI, its target and the measurement source into the agreement. If you don't know your target yet, agree a test phase whose goal is to find it. Our guide to setting an app marketing budget shows how to work back from LTV.
How a good agency bids for outcomes
Every major network can optimize toward in-app events and value, but only if you send them clean events. A performance agency should set this up before scaling spend.
- Google App campaigns: target CPA optimizes toward one in-app action and target ROAS toward in-app action value. Google's App campaign bidding guidance recommends selecting a single action for target CPA rather than several, and starting a target ROAS from what comparable campaigns achieved.
- Meta: app campaigns can optimize for app events or for value once purchase values are flowing through the SDK or Conversions API. Our guide to Meta app install ads covers the setup.
- Apple Ads: keyword-level bids on App Store search, with performance judged on downstream events from your attribution data.
The agency's job is to choose which event each campaign optimizes for, move deeper in the funnel as volume allows, and protect the learning phase from constant edits. If your conversion events are too rare for the algorithm to learn from, they should say so and propose a proxy event that correlates with revenue.
Reporting cadence you should expect
Reporting is where you find out whether the agency thinks in outcomes.
Weekly
- Spend, installs, primary KPI cost and volume by channel and campaign.
- Early cohort ROAS against target.
- Creative performance: what launched, what won, what's fatiguing.
- What changed last week, and what they'll test next week.
Monthly
- Cohort tables: ROAS and retention by acquisition month and channel, at day 7, 30 and 90.
- Payback progress against plan.
- Budget recommendation for next month, with the reasoning.
- Blended view: total new paying users and revenue against total spend, including organic, so channel attribution doesn't hide cannibalization.
Agree one source of truth up front, usually your mobile measurement partner or your own backend data, rather than each ad network's dashboard. Networks claim credit generously. If you're choosing an MMP, see AppsFlyer vs Adjust vs Branch.
Share your app and goals. We'll send a channel plan, starting budget and the first creatives we'd test — free, within 48 hours.
Get my free growth plan WhatsApp usPricing models and the incentives they create
Ad spend goes to Google, Meta or Apple, ideally billed to your card on your accounts. The agency fee is separate. Here's how the common models compare:
| Model | How it works | Upside | Watch out for |
|---|---|---|---|
| Monthly retainer | Fixed fee for agreed channels and scope | Predictable; no incentive to inflate spend | Scope creep; effort may not scale with spend |
| Percentage of ad spend | Fee is a share of media spend, often with a minimum | Scales with work as budgets grow | Rewards spending more, not spending well |
| Hybrid | Base retainer plus a smaller percentage above a spend threshold | Balances predictability and scale | Needs clear thresholds in the contract |
| Performance bonus | Retainer plus a bonus for hitting CPA, ROAS or payback targets | Aligns the agency with your outcome | Targets must use your source of truth and mature cohorts |
| Pure pay-per-action | You pay per install or event, the agency buys the media | Low risk on paper | Often means opaque networks, low quality or fraud; you rarely own the data |
For most apps, a retainer or hybrid fee with an outcome bonus works best. Be cautious with pure pay-per-install or pay-per-lead deals: the supplier makes money by delivering volume at your price, not by finding users who pay.
When a performance marketing agency for apps is worth it
An agency isn't always the right answer. It tends to pay off when:
- You have product-market fit signals (users who stay and pay) and want to scale paid acquisition beyond what a founder can manage part time.
- You're spending, or ready to spend, enough per month that better bidding and creative testing will save more than the fee.
- You need several channels running at once and can't hire a specialist for each.
- Your in-house team is strong on product and brand but has never run app campaigns optimized for in-app events.
It's usually too early when retention is weak, when you can't yet track the events that matter, or when your test budget is so small that the fee would be a large share of total spend. In those cases, fix onboarding and tracking first, run small tests yourself, and bring in help when the numbers show users are worth buying. If you'd rather start with a single experienced person, compare the options in our guide to hiring an app growth marketer.
Whichever route you choose, make the first 60 to 90 days an explicit test phase with a written goal: find the channel, event and creative mix that hits your target cost per paying user, or learn why it can't yet.
Questions to ask before you sign
- Which KPI will you optimize toward in month one, and which by month three?
- What's your measurement source of truth, and how do you handle iOS attribution under SKAdNetwork and AdAttributionKit?
- How many new creatives will you ship per month, and who makes them?
- Will the ad accounts and data be in our name?
- When would you recommend cutting spend? Ask for an example of a decision rule.
- What do you need from our product and data teams to hit target?
Red flags
- Reports built around impressions, clicks and installs, with paying users buried or missing.
- Promises of a specific CPA or ROAS before they've seen your data.
- Agency-owned ad accounts you can't access.
- No creative plan, or the same ads running for months.
- Resistance to incrementality checks, such as pausing a channel in one region to see whether paying users actually drop.
For a broader view of evaluating agencies, see our guide to choosing a mobile user acquisition agency.
Working with Appryze on performance
Appryze runs app growth on Google App Campaigns, Meta and Apple Ads, led by a founder with 12+ years in user acquisition for consumer and fintech apps. We set up event tracking and attribution first, report weekly against the KPI you choose, and run everything on ad accounts you own. See how it works on our Growth page.
Frequently asked questions
What does a performance marketing agency for apps do?
It runs paid user acquisition on channels such as Google App Campaigns, Meta and Apple Ads, optimizing toward in-app outcomes like purchases, subscriptions or funded accounts instead of installs. A good one also sets up event tracking and attribution, produces and tests creatives, reports on cohort ROAS and payback, and recommends where to move budget each month.
How much do app performance marketing agencies charge?
Common models are a monthly retainer, a percentage of ad spend, a hybrid of the two, or a retainer plus a bonus for hitting agreed CPA or ROAS targets. Ad spend is paid separately to the networks. Fee levels vary widely by scope, channels and spend, so compare proposals on what's included and how the model aligns the agency with your outcomes.
What KPIs should I use for app marketing?
Use the cost of the event that signals real value for your model, such as cost per paid subscriber, first purchase, funded account or completed order, plus cohort ROAS at day 7, 30 and 90 and your payback period. Installs and cost per install remain useful diagnostics for creatives and store listings, but they shouldn't be the goal.
Is percentage of ad spend a good pricing model for an app agency?
It's common and scales with workload, but it rewards spending more rather than spending well. Many apps get better alignment from a fixed retainer, or a hybrid with a modest percentage above a spend threshold, combined with a bonus tied to CPA, ROAS or payback measured on your own data source.
Should the agency own my ad accounts?
No. Your ad accounts, attribution tools and analytics should be created under your company, with the agency added as users and ad spend billed to you. That keeps your historical data, pixel and SDK learnings, and audiences with you if you change agencies, and it makes spend fully transparent.
Share your app and goals. We'll send a channel plan, starting budget and the first creatives we'd test — free, within 48 hours.
Get my free growth plan WhatsApp usKeep reading
More in Performance Marketing Agency for Apps · All app marketing guides