From a value to an investment
The predicted value syncs into Braze as an attribute, and the band picks the play. Concentrate the spend where it pays back: protect the top, invest to grow the risers, sustain the steady, and keep a light touch on the rest.
Spend the budget where the value is going
One play per band, so the score turns into a message rather than a dashboard reading.
Protect and deepen
Your most valuable customer. Protect the relationship and deepen it: early access, genuine recognition, premium products. No discounting needed.
Invest to grow
Modest value today, strong upward trajectory. This is where investment compounds: build the purchase habit and widen breadth while momentum is on your side.
Sustain efficiently
Real value on a flat trajectory. Sustain the relationship efficiently: keep the habit warm without overspending, and lean into their seasonal sale moments.
Watch, low spend
Low predicted value for now. Keep contact light and cheap, and let the relationship prove itself before you invest in it.
Prove the investment paid back
Spending more on your best customers looks good until you ask what they would have done anyway. Hold a slice out, measure only the gap above them, and the model earns its budget on evidence.
Raw before-and-after would claim the whole targeted bar. True attribution counts only the gap above the control, the part that would not have happened without the model. Illustrative.
That is Predicted LTV (pLTV) working on sample data. The service page has the rest: what it needs from you, how long it takes to build, what it pairs with, and how the lift gets proven.