Look beyond the install
Estimate net contribution over an agreed target period, rather than treating every new user as equally valuable.
PAID ACQUISITION · 90-DAY NET VALUE
Understand user value before your next acquisition investment.
Use the first 24 hours of acquired users’ behavior to predict 90-day net value after acquisition and agreed service costs. Use that evidence in your next acquisition decision, then test the predictions against mature outcomes and advertising experiments.
$149 per month + 1% of effective registered Campaign spend
USER VALUE SNAPSHOTIllustrative data
D90 predicted net value
+$42.60
D0D30D60D90
Reference percentile
P82Channel reference coverage
Top 50%Tree leaf
L-07Core task completed → Repeat usage → High-value branch✳
Illustration only. Not a customer result or a guarantee of investment returns.
WHY USER VALUE
More installs do not automatically mean profitable growth.
Install cost and short-term revenue leave out refunds, acquisition cost and ongoing service costs. A user can pay and still contribute negative net value. Agree on one economic definition before deciding which users to acquire again.
Learn before D90
Use early behavior to estimate later value. Keep the original prediction so mature outcomes can test it honestly.
Act with evidence
Receive explainable labels and result APIs. Your team decides and executes advertising uploads; a registered experiment is needed to establish lift.
01 One revenue and cost definition02 Explainable user segments03 Verify against actual net value
FROM EARLY BEHAVIOR TO ACQUISITION DECISIONS
Turn long-term value into a signal for your next campaign.
Installs and first payments do not capture a user’s full long-term net contribution. Connect your customer data, estimate future net value, and use the results to guide later acquisition decisions.
Connect behavior with actual net outcomes
Connect attribution, product behavior, revenue and costs. Use mature users’ actual net value to define the model’s prediction target.
Predict 90-day net value from day-one behavior
After 24 hours from installation, deliver net value predictions, value segments and decision evidence for users with complete day-one data, within validated model, currency and cost scopes. Models must pass independent validation on real historical data; incomplete cases remain pending.
Use long-term value in your next campaign
Retrieve results through the API and dashboard. Your team maps and uploads signals under the advertising platform’s rules, then tests their impact through advertising experiments.
Your team supplies mature revenue and cost outcomes. We compare them with first predictions, review model suitability, and retrain and independently validate a replacement before publishing an update when needed.
User value=Revenue in the target period−Refunds, taxes & fees−Acquisition cost−Agreed service costs
01 / FROM BEHAVIOR TO VALUE
Understand the estimate. Keep the original prediction.
A validated regression tree maps early behavior to 90-day net value and unequal feature-based groups. Organic data is also received and stored, but is not billed as paid acquisition. Each response includes the leaf, decision path, reference percentile, threshold and value label. Negative values remain visible.
The first prediction is versioned and retained within the agreed data-retention window. Mature outcomes are compared with that prediction. A percentile is a relative position, not a payment probability.
A WORKING END-TO-END EXAMPLE
Follow the data. From API to dashboard.
Explore stored synthetic user records, value labels, channel comparisons, monthly usage and first-prediction validation in the live dashboard.
Open the live demo ↗The demo uses synthetic data and a separately tested model. It demonstrates system behavior, not real advertising lift. The console supports seven languages. The legacy demo uses Chinese; API documentation is available in English and Chinese.
ONE STANDARD SERVICE
Send the data. We handle the processing.
Your team integrates a documented API. We reconcile data and validate real-history models before scoring. Your team performs advertising uploads in its own accounts using the provided mapping.
Connect & validate
Align user identity, early behavior, attribution, revenue and cost definitions before the trial begins.
- Stable user and event IDs
- Timestamp and cost checks
- Duplicate protection
Predict & segment
Receive a predicted net value, a reference percentile and a clear high-value threshold.
- Customer-specific model
- Tree path and model version
- First prediction preserved
Results through API
Receive a durable batch receipt first. Warehouse loading and prediction status are separate; retrieve available results through read APIs and the dashboard.
- Batch and individual queries
- Separate read and write keys
- Private customer dashboard
Verify & maintain
Compare original predictions with mature net outcomes and review model suitability over time.
- Actual net value comparison
- Channel and cohort summaries
- Usage and billing transparency
Standard first-phase onboarding covers AppsFlyer / Adjust apps and S24/D90 scoring in validated USD scopes. Domestic tools and low-cost IAA require custom assessment; Web and mini-games are assessed before support is promised. Customers perform advertising uploads.
RESULTS YOU CAN CHECK
A prediction to inspect. An outcome to verify.
Start with user-level net value, a decision path and a frozen reference threshold. Follow channel and cohort summaries, compare original predictions with mature actual outcomes, and review customer-reported weekly campaign revenue.
User-level evidence
Predicted value, high-value label, eligibility boolean, model version and first prediction stay distinguishable.
Model validation
Independent historical validation tests the model. It does not establish advertising lift or guarantee an individual purchase.
Observed business results
Weekly revenue compares reported groups on the same timeline. Missing reports remain unknown; group differences alone do not establish causation.
The legacy demo contains synthetic data. It does not establish customer results or revenue lift and does not guarantee investment returns.
Inspect the synthetic demo ↗A CLEAR START
One conversation. A documented integration.
- 01
Leave your details
Share your email, product and approximate monthly volume.
- 02
Align data and value
Confirm the target period, costs, identity and early behavior inputs.
- 03
Start your 30-day trial
Receive authorized web accounts and separately scoped API tokens after onboarding.
- 04
Continue after confirmation
Paid service starts only after explicit acceptance. No card is charged.
Service pricing
Make the next decision with earlier evidence.
Leave your contact email. We will contact you to discuss your product, integration readiness and trial arrangements.
$149 per month + 1% of effective registered Campaign spend
One app, one advertising platform, USD and an agreed billing timezone. One base fee per contract.
Authorized data is analyzed internally; only registered Campaign results are delivered.
30-day company trial; no automatic paid conversion or debit.
Campaign service inquiry| Effective spend | Monthly service fee |
|---|---|
| $0 | $149 |
| $3,000 | $179 |
| $10,000 | $249 |
| $20,000 | $349 |
| $100,000 | $1,149 |
Pause stops new deliveries immediately; billing stops at the next local midnight.
Daily spend is customer reported and separately reviewed. Missing reports remain unknown.
Paid service starts only after explicit acceptance. No card is charged.
Legacy allowance contracts keep their agreed price and behavior.
Tell us about your product
30-day company trial; no automatic paid conversion or debit.
Payment method confirmed before activation
A FEW THINGS TO KNOW
Questions before you start
What does user value mean?
Net income in the agreed target period, minus acquisition and agreed non-advertising costs. Predicted values and mature actual values are stored separately.
Does high value guarantee a purchase?
No. It is an expected value and a position relative to a frozen threshold. It does not guarantee an individual payment, retention outcome or advertising lift.
Can I use my own historical data?
Yes. We align the input schema and train a tenant-specific model on mature historical outcomes. Production customers are not scored with the synthetic demo model.
Will negative values be uploaded to advertising platforms?
Negative net values remain available internally. Any platform signal requires a separately confirmed integration and an allowed event or value mapping; predictions are not reported as actual purchases.
Do I need a card to try the service?
30-day company trial; no automatic paid conversion or debit.
Paid service starts only after explicit acceptance. No card is charged.