CPP Experiments feature in SplitMetrics Acquire helps you turn Custom Product Page testing into a structured workflow, from setup through rollout and analysis.
What CPP A/B Experiments Are
CPP A/B experiments let you test multiple Custom Product Pages against each other in Apple Ads to identify which page delivers the best performance for your goals.
SplitMetrics Acquire handles setup, traffic distribution, statistical analysis, and ongoing monitoring, allowing you to focus on making decisions rather than managing manual processes.
Why CPP A/B Experiments Matter
Most UA teams run CPP tests, but very few can turn the results into actionable insights. At scale, several common challenges come up:
🔹 Manual setup is error-prone
Duplicating ad groups, switching CPPs, and keeping names consistent involve multiple steps where mistakes can occur and impact test validity.
🔹 Traffic distribution is difficult to control
Without proper monitoring, experiments may run for weeks without producing reliable or usable results.
🔹 It is difficult to identify a clear winner
Without predefined duration and decision criteria, tests may end too early or run longer than necessary.
🔹 Surface metrics can be misleading
Metrics such as tap-through rate alone do not fully reflect whether a CPP drives installs, conversions, or downstream revenue.
🔹 Rollout is disconnected from testing
Even after identifying a winning CPP, applying it to live campaigns is often a separate manual step.
🔹 Post-rollout impact is unclear
After rollout, it is not always obvious whether the winning CPP actually improved performance without incremental measurement.
The core challenge is managing the full CPP testing lifecycle, including setup, monitoring, analysis, decision making, rollout, and impact measurement, in a consistent and scalable way.
How CPP A/B Experiments Work
Each experiment goes through five stages, with continuous health monitoring running in the background:
- Setup defines the experiment method and parameters.
- Data collection runs with automated health checks.
- Win signals indicate which variation is leading and how strong the signal is.
- Full-funnel validation helps evaluate performance beyond the primary KPI.
- Scale allows you to apply the winning CPP across campaigns.
Each stage is covered in detail in the 📚 CPP Experiments section of our Help Center.
What CPP A/B Experiments Unlock
CPP Experiments tool automates the key parts of the testing workflow:
✅ Setup automation
Automatically creates and configures test ad groups based on your selected method.
Copies targeting, bids, keywords, and audiences from the original ad group.
Freezes automations such as Rules, Bid Strategies, Budget Allocation, Keyword Expansion Strategies, and CPP Scheduler during the experiment to keep results clean.
✅ Traffic handling
Dayparting: rotates variations on a fixed intra-day schedule.
Parallel: runs all variations simultaneously while monitoring and balancing traffic distribution.
Switch: rotates variations one at a time on a daily or weekly schedule.
✅ Health monitoring
Continuously tracks changes that could affect data quality.
Sends alerts when attention is needed, including what changed, who made the change, and recommended actions.
Maintains a full audit trail of all activity.
✅ Duration estimation
Estimates the expected duration of the experiment before launch based on historical data.
Applies minimum time and data thresholds to avoid premature conclusions.
✅ Statistical analysis
Uses a Bayesian model to estimate the probability of each variation winning based on your primary metric.
Verifies that performance differences are stable rather than short-term spikes.
Produces clear outcomes such as Winner, Inconclusive, or Not enough data.
✅ Safe environment and reversibility
The original ad group is paused during the experiment and automatically reactivated when it ends.
All automations are restored after completion.
If the experiment is stopped early, the system automatically reverts changes without requiring manual cleanup.
🏆 Key capabilities at a glance
- Automated experiment setup using Dayparting, Parallel, or Switch methods.
- Data-driven duration planning before launch.
- Centralized experiment management in one place.
- Clear, probability-based win signals rather than raw comparisons.
- Primary KPI-focused analysis such as TTR, DR, CPT, CPA, and more.
- Full-funnel visibility from impressions to downstream conversions.
- Built-in health monitoring to ensure data reliability.
- Seamless transition from insights to rollout.
What to Expect from an Experiment
Not every experiment will produce a clear winner. If variations perform similarly, that is still a valuable outcome for decision making.
Some traffic fluctuations are expected after launch as Apple Ads adjusts to new ad groups. This typically stabilizes within the first day.
CPP assets should remain unchanged during the experiment. Updating creatives, such as screenshots or promotional text, may invalidate the results.
Experiments require time. Minimum thresholds ensure results are statistically reliable, and an estimated duration is shown during setup.
Who This Is For
This feature is designed for UA managers and Apple Ads practitioners who run campaigns with Custom Product Pages and need reliable, data-driven CPP decisions at scale.
Prerequisites
- An active SplitMetrics Acquire account with CPP Experiments enabled.
- At least one Apple Ads campaign with Custom Product Pages configured in App Store Connect.
Where to Start
Open CPP Experiments from the main navigation under the Optimization Hub. The Experiments Hub provides an overview of all experiments, including active, completed, and scheduled ones.
Guide Series
# | Article | What it answers |
1 | What are CPP A/B experiments? (you are here) | What this is and why it matters |
2 | How to choose a method, set your KPI, and launch an experiment | |
3 | How to interpret signals and validate impact | |
4 | How to roll out the winning CPP | |
5 | How to know whether your data is reliable | |
6 | How to manage multiple experiments at scale | |
7 | Common questions and answers |