Predicted LTV (pLTV)

A value you can read

No black box. The prediction is a decayed base value times a growth multiplier, and the multiplier is the sum of a handful of named drivers, each with a plain-English reason. Trajectory, breadth and headroom lift it; regression to the mean and an event-only pattern pull it back.

The drivers

Trajectory

Whether their value is accelerating, holding steady or cooling

Breadth

A wider footprint compounds; a single-product one does not

Tenure headroom

Newer relationships have room to grow; mature ones regress toward the mean

Subscription floor

An auto-renewing plan puts a floor under future value

Event elasticity

Value concentrated around big events is capped, not compounding

Drivers compose a growth multiplier that starts at 1.0 and is clamped to a sensible range, so no single relationship runs away. A modest-value relationship with a strong multiplier is promoted into Grow: the model's favourite is the one worth investing in, not just the one that is already big. Illustrative weights; the production model calibrates them on your data.

The bands

Invest

pLTV 1500+

Grow

pLTV 500 to 1499

Maintain

pLTV 150 to 499

Monitor

pLTV 0 to 149

Bands read on the predicted value, with one exception: a strong growth multiplier lifts a modest-value relationship into Grow, because forward potential is worth investing in before it shows up in the level.

What feeds it in your world

No new tracking to invent. These are events you already capture; the model reads them as they are.

·12 months of spend history per customer
·Purchase timestamps, so value can be recency-decayed
·Products used, the breadth driver
·Tenure and subscription status, the confidence and floor drivers

How it sharpens over time

Models are tuned, not installed. Each step is earned on evidence, and the previous step keeps running until the next one beats it.

v0
Heuristic, day one
Transparent rules on the data you already have. Live in Braze in weeks, and everyone can see why it fired.
v1
Calibrated to your history
Thresholds and weights re-fit to your own base and vertical, backtested against what actually happened.
v2
ML where it earns it
A trained model replaces the rules only when it beats them on holdout data. Explainability stays a requirement.
v3
Monitored and re-tuned
Drift watched, thresholds reviewed on a cadence, and every change proven with holdouts before it ships.

pLTV's confidence is honest by design: it reads 'early' on thin tenure and strengthens as history builds. The multiplier weights are re-fit to your own cohort curves, and the invest and grow thresholds are scaled to each vertical's economics rather than borrowed from someone else's.