Cost Per Engagement (CPE) Guide: Pricing Model Explained

I was hemorrhaging money on installs that never opened the app twice.

Our CPI campaigns looked great on paper. Thousands of installs at $2 each. The CMO loved the numbers. Then someone finally asked the uncomfortable question: "How many of these users actually do anything?"

The answer was devastating. 73% of our installs never completed the tutorial. Never got past the first screen. Never experienced the product we'd spent two years building.

We weren't acquiring users. We were acquiring app icons that sat on phones for three days before being deleted.

That's when I discovered CPE—and it changed everything about how I think about user acquisition.

The Moment I Realized We Were Measuring the Wrong Thing

Cost Per Engagement flips the model. Instead of paying for installs—which tell you nothing about quality—you pay for specific actions. Tutorial completed. Level 5 reached. First item added to cart. Account verified.

The first time I ran a CPE campaign, my head of growth was skeptical. "We're going to pay $8 for what we're getting at $2?"

I showed him the math. Those $2 installs had a Day-7 retention of 4%. The $8 engaged users? 38%. Over 90 days, the "expensive" users generated 6x the revenue. They weren't more expensive—they were dramatically cheaper when you measured what actually mattered.

Why CPI Was Actually Costing Us More

Let me be blunt about what CPI campaigns actually buy you: volume. Not quality. Not intent. Not future revenue. Just volume.

With CPI, you pay for every install, regardless of whether the user opens the app once and deletes it, or becomes a loyal customer. The ad network gets paid the same either way. Think about what incentive structure that creates.

CPE changes the game. The network only gets paid when users complete a meaningful action. Suddenly, they care about quality too. They stop sending you click-farm traffic because that traffic doesn't convert to engagement—and they don't get paid for installs that don't engage.

But What About CPA?

You might be thinking: why not just go straight to Cost Per Acquisition? Pay only for purchases?

I tried that. The problem is scale. Purchase events are relatively rare—maybe 2-3% of users in the first week. Networks struggle to optimize for events that sparse. You get low volume, inconsistent delivery, and campaigns that never escape the learning phase.

CPE hits the sweet spot. Engagement events happen frequently enough for algorithms to optimize effectively, but they're still predictive of long-term value. It's the Goldilocks pricing model.

The Art of Choosing the Right Event

Here's where most people get CPE wrong: they pick the wrong event.

I watched one team set "app open" as their CPE event. Guess what? Their completion rate was 95%. Sounds great until you realize they were basically running CPI campaigns with extra steps. They weren't filtering for quality at all.

Another team went the opposite direction: "first purchase" as their CPE event. Completion rate: 3%. They couldn't get any scale, and the users they did acquire cost a fortune because the network had to burn through so many installs to find buyers.

The Four Criteria That Actually Matter

1. Predictive power. Does completing this event correlate with long-term retention and revenue? We found that users who completed our tutorial had 4.2x higher D30 retention. That's predictive.
2. Achievable difficulty. 20-50% of installers should complete your event. Too easy and you're not filtering. Too hard and you can't scale.
3. Time window. The event should happen within 24-48 hours. Longer windows make optimization nearly impossible and create fraud opportunities.
4. Fraud resistance. Bots can fake installs. They can fake app opens. They struggle to fake completing a 5-level tutorial or adding a payment method.

"The perfect CPE event is one that users must actually experience your product to complete. That's why it correlates with retention—you've selected for people who got past the first impression."

How I Found Our Winning Event

We tested three different CPE events on the same budget:

• Tutorial completion: $4 CPE, 45% completion rate, D7 retention 28%
• First core action: $6 CPE, 30% completion rate, D7 retention 41%
• Account verification: $12 CPE, 15% completion rate, D7 retention 52%

The math said account verification. Despite costing 3x more per event, those users had the highest LTV and the lowest churn. The effective cost per retained D30 user was actually lowest for the "expensive" option.

But here's the nuance: we couldn't scale account verification. The volume wasn't there. So we ran a hybrid approach—"first core action" as our primary CPE event at scale, supplemented by account verification for premium placements.

The Fraud Pattern That Almost Killed Our Campaign

Three weeks into our first CPE campaign, something looked wrong. One network was crushing it—40% event completion rate, way above our benchmarks. We scaled spend.

Then we looked at the time-to-event data.

Real users took an average of 8 minutes to complete our tutorial. Users from this network? 47 seconds. They were completing a 5-level tutorial in under a minute.

That's not possible without cheating. They'd found a way to simulate event completion without actually engaging with the app. We cut them immediately and implemented time-to-event monitoring across all sources.

How to Protect Yourself

• Set minimum time thresholds. If your event should take 5 minutes, flag any completions under 2 minutes.
• Monitor completion rate by source. Outliers are usually fraudulent.
• Track post-event behavior. Do event completers continue engaging, or do they disappear?
• Use your MMP's fraud detection. They see patterns across thousands of advertisers.

The Uncomfortable Truth About "Cheap" Installs

After that first successful CPE campaign, I couldn't go back to CPI. Every time someone pitched me cheap installs, I asked one question: "What percentage of those users are still active after 30 days?"

Usually, they didn't know. They hadn't measured it. They were selling vanity metrics—install counts that looked impressive in weekly reports but generated zero revenue.

CPE isn't more expensive. It's more honest. You're paying for what you actually want: users who engage with your product. Everything else is just noise that happens to live on phones temporarily.

The $8 engaged user who sticks around is infinitely more valuable than ten $2 installs who delete the app before lunch. Once you see that math, you can never unsee it.