Using Tenjin’s Predicted LTV to Optimize Campaigns
The N-Day All pLTV (Ad Mediation + IAP) metric in Tenjin allows you to evaluate campaign performance early, using our machine-learning forecasts of total user revenue. This helps you make informed decisions about scaling, maintaining, or pausing campaigns without waiting for cohort data.
- It combines IAP LTV and Ad Mediation LTV (ILRD) into a single predicted metric with ~90% average accuracy.
- Supported cohort sizes: Any day from 1 to 30 (e.g., 1, 2, 3, ..., 30 days).
- Predictions are based on 0-day metrics, matured cohort data and historical data.
- Available for all users on any paid Tenjin plan.
Here’s how to set it up and how to use it effectively.
Setting Up 1-Day pLTV (Ad Mediation + IAP) on the Tenjin Dashboard
To view 1-Day pLTV (Ad Mediation + IAP) for your campaigns:
- Go to your User Acquisition page in the Tenjin dashboard under the ANALYZE section.
- Click ‘Edit Metrics’ (top right).
Add the following columns to your table:
- Spend
- 1-Day All LTV (Ad Mediation + IAP)
- 1-Day pLTV (Ad Mediation + IAP)
This setup gives you both actual and predicted revenue for each campaign for 1-day cohort.
How to read the Data
Here’s an example of how your table will look after the above metrics are added, this example is when the data is broken down by campaign for a single date for your app:
- Spend: Total UA spend for the campaign.
- 1-Day All LTV (Ad Mediation + IAP): Actual day-1 total revenue from ad mediation revenue + IAP revenue.
- 1-Day pLTV (Ad Mediation + IAP): Predicted total revenue for the cohort of users after 1-day of acquisition.
Interpreting the Results
In this example you will notice:
-
Campaign 1 had a spend of $1,810.20 The actual 1-Day All LTV (Ad Mediation + IAP) was ~$1,270, and the predicted LTV was ~$1,307.
- The approx predicted ROAS would be around ~72% (Predicted LTV / Spend).
- This is a very strong early signal because users are only in their first 24 hours. If these users retain and monetize further in later cohorts (e.g., Day-3 or Day-7), the campaign is likely to exceed 100% ROAS over time or break-even.
- The recommendation would be to consider maintaining current spend to test the scale
-
Similarly, Campaign 3 spent $436.08 with actual 1-Day All LTV (Ad Mediation + IAP) of $356.49 and the predicted LTV of $366.28
- The predicted ROAS would be around ~84%, better than campaign 1.
- Again, this is a positive early signal. The cohort is already close to break-even on Day-1 and likely to become profitable in future cohorts.
- The recommendation would be to increase budget cautiously on this campaign and test scaling on later cohorts where the users mature leading to profitability.
Key Takeaways
- Use the Predicted LTV metrics as an early indicator to decide which campaigns to scale.
- Depending on your ROAS target, If the predicted ROAS is greater than the target you have on Day-1 itself, this is usually a great indicator for potential to scale. Users are not yet mature and will likely drive higher ROAS over time.
- If predicted ROAS is less than your ROAS target, reduce your spend or test new targeting or creatives.
- Focus budgets on campaigns that are showing strong predicted performance to maximize profitability.
FAQs
What is the difference between “1-Day All LTV (Ad Mediation + IAP)” and “1-Day pLTV (Ad Mediation + IAP)”?
The 1-Day All LTV (Ad Mediation + IAP) shows actual revenue (Ad Mediation + IAP) generated by users acquired on Day-1. The 1-Day pLTV (Ad Mediation + IAP) is a machine-learning forecast that predicts total revenue for the cohort based on 0-day metrics and historical cohort data.
Can I use Predicted LTV for all campaigns and ad networks?
Yes. Predicted LTV is available for all campaigns and ad networks in Tenjin, as long as you’re on a paid plan and have both Ad Mediation and IAP data integrated properly.
How accurate is the Predicted LTV metric?
The Predicted LTV metric has an average accuracy of ~90%. Its designed to give strong early signals for optimization but should be used in combination with other KPIs for campaign decisions.
Why doesn't predicted LTV always match actuals?
Our prediction model uses up to 78 input signals. Because it's learning from many variables (and their interactions), a 1:1 match isn't expected. Small gaps or delays in any inputs can affect accuracy, and actual LTV also shifts over time due to delayed events and data corrections. The model is retrained and recalibrated regularly, so discrepancies should narrow over time.