Fuse Data Products

Reusable data models, made tangible

Each model is built once and re-skins to any business. RFM is the foundation you can open and play with today; the rest of the catalogue layers on top for compounding value, from Databricks into Braze.

One model is the start. The power compounds.

RFM is the deterministic foundation. Every other model layers on and adds a dimension. They are all standalone, and you turn each on only when it earns its place. Some are quick wins you can run today; others get better as purchase and spend history build. What you can switch on comes down to the data you have.

Deterministic

RFM

Recency, frequency and monetary segmentation, plus behavioural trends. It tells you who your customers are and what to do next, and it is explainable enough to use in Braze from day one.

Needs: Purchases

Adds: The explainable foundation the other models build on

When: The base model. Start here.

Open the model →

Deterministic

Engagement Score

Scores digital engagement on its own, separate from spend: logins, sessions, saved items, app opens.

Needs: App / web session + interaction events

Adds: The engagement axis RFM leaves out, for activation and experience-risk signals

When: As soon as session events flow

Open the model →

Braze AI

Intelligent Selection

Braze automatically shifts campaign traffic to the best-performing variant, so you optimise without manual A/B reviews.

Needs: Campaign + conversion events

When: Available now. A quick win.

Open the model →

Braze AI

Braze Recommendations

Suggests the most relevant product, complementary range or offer based on how someone engages. Personalised next-best-content.

Needs: Purchase history + product catalogue

Adds: Personalises what each segment is offered

When: Once purchase history builds

Braze AI

Send-Time Optimisation

Works out the best moment to reach each customer on each channel, then lets Braze deliver when they are most likely to engage.

Needs: Message + engagement history

Adds: The 'when' for every campaign, with no manual scheduling

When: Available in Braze as history builds

Open the model →

Churn Propensity

Predicts who is likely to lapse and pushes a risk score into Braze, so you can step in before they churn.

Needs: Behavioural / purchase / session events

Adds: A forward risk signal that recency only hints at

When: Starts on early data, sharpens over time

Open the model →

Predicted LTV (pLTV)

Estimates what a customer is likely to be worth over the next 12 months, so you know who to invest in.

Needs: Purchase + spend history

Adds: Turns a value tier into a forward-looking investment dial

When: Plan it alongside RFM

Open the model →

Next-Best-Action

A decisioning layer that picks the best action, offer and channel for each customer in the moment. The orchestration brain over the other models.

Needs: The feeder models + a decisioning framework

Adds: Turns many signals into one coordinated decision per customer

When: The apex, once the feeders exist

Open the model →

Channel Optimisation

Predicts the best channel for each customer, whether that is email, push, SMS or in-app, instead of blasting every channel.

Needs: Cross-channel engagement events

Adds: The 'which channel' for every message

When: As cross-channel history builds

Open the model →

Analytical

Incrementality / Uplift

Measures the true causal lift of a campaign with holdouts, and works out which customers are genuinely persuadable.

Needs: Holdout groups + conversion events

Adds: Proof of incremental value, not just correlation, plus smarter targeting

When: Layer it onto any always-on programme

Open the model →

Analytical

Contact-Frequency Optimisation

Finds the right contact cadence for each customer: enough to keep them engaged, not so much they fatigue or opt out.

Needs: Send + engagement + opt-out events

Adds: The 'how often', governing frequency caps by value

When: Once send and engagement history exists

Open the model →

Analytical

Closed-Loop Intelligence

Connects activity, behaviour and performance into one learning cycle, so you can see which channels, timings and frequencies actually change behaviour.

Needs: Currents + campaign performance

Adds: The Learn and Optimise step that makes every model smarter

When: Compounds over time

On their own, each model is useful. Together they compound: RFM tells you who and what, Recommendations pick the offer, and Send-Time, Channel and Frequency decide how it is delivered. Churn and pLTV add risk and worth, Next-Best-Action makes the call, and Incrementality and Closed-Loop prove what actually worked.