BX3INTELLIGENCE
The decision systemGetting startedLibraryTrustPricingServices
Client login
The models

The whole shelf, not one model.

Mix modelling answers one question. Churn tooling answers another. We built a library that spans acquisition, measurement, customer, revenue and diagnostic work — so your second question doesn't need a second vendor.

The full shelf

Every model, and where it stands.

Framed for:

Explain what happened

4 models
Conversion / Rate Bridge
“Why did my conversion rate change?”
The number moved. This is how much of it was each thing.
Certified
Anomaly / Outlier Scan
“Was that move real or just noise?”
What broke pattern this month, before a quarterly review would have caught it.
Certified
Segment / Cluster Analysis
“Who are our real segments?”
Finds the natural groupings in your customers or accounts and what makes each distinct.
On the roadmap
Driver Analysis
“What actually moves this target?”
Which inputs move an outcome — an explanatory layer over a fitted model.
On the roadmap

Predict what's next

12 models
Revenue / Metric Forecast
“Where is this metric heading?”
Where a series is going, with a range around it.
Certified
Lead Scoring
“Which leads will convert?”
Every open lead, scored and ranked on how likely it is to convert within a defined window.
Also deployableCertified
Lapse Risk
“Who's about to stop coming back?”
Every recent buyer, scored and ranked on how likely they are to stop coming back within a defined window.
Also deployableCertified
Churn Risk
“Who's about to go quiet?”
Every active customer, scored and ranked on how likely they are to go quiet within a defined window.
Also deployableCertified
Contract Churn Risk
“Who won't be there at renewal?”
Which customers won't be there at their renewal date — each one scored against their own calendar.
Also deployableCertified
Contract Renewals
“Which customers are coming back?”
Which customers are most likely to renew at their date — each one scored against their own calendar.
Also deployableCertified
Conversion Propensity
“Which prospects or sessions are most likely to convert next?”
Every existing customer, scored and ranked on how likely they are to buy again within a defined window.
Also deployableCertified
Retention Curves
“How does each cohort actually decay?”
How long things last, and the exact point where you lose them.
Certified
Time-to-Event Hazard
“When does churn risk spike?”
The danger zones by tenure — h(t) — where risk accelerates.
In build
Windowed Churn Probability
“Will they churn between now and then?”
Churn risk for any future window you name, no new model needed.
In build
Competing-Risk Churn
“Why are they leaving — and which way?”
Voluntary, involuntary, and upgrade-out — told apart.
In build
Next-Best-Action
“What should we do for each customer?”
The action most likely to move each customer — multi-outcome propensity.
On the roadmap

Allocate where it pays

5 models
Channel / Touchpoint Attribution
“What deserves the credit?”
Which channels earn the credit, and whether the order they happen in matters.
Certified
Response Curves
“Where do returns run out?”
How much each extra dollar on a channel buys, and where it falls to a quarter.
Certified
Price Elasticity
“How sensitive is demand to price?”
Price and discount sensitivity from a fitted response curve.
On the roadmap
Media-Mix (MMM-Lite)
“What is my spend really doing?”
Channel response curves for decision support.
On the roadmap
Budget Optimisation
“Where's the next dollar best spent?”
Reallocation across the fitted response curves.
On the roadmap

Test what works

1 model
Incrementality
“Did it actually make a difference?”
Did the change cause the difference? Sized and sealed before you spend.
Design · CertifiedReadout · In build

Not a model on the shelf: Outcome Studio opens on any certified result, in every tier.

Three kinds of product, all priced in the open

Every model in the library runs as an analysis, included in every tier: you come with a question and a model answers it. When you want a model's scores kept fresh and ready to act on, it becomes a deployed model, operator-released on a cadence and priced on the page. And tests are a program: an always-on experimentation practice, priced per brand, where a design that can't answer is refused before it's billed. All three, priced in the open.

Every analysis, included.Deployed models, priced on the page.See pricing →
How trust is built into every result

Whatever the model, the result arrives the same way: trained on your past, tested on the future it hadn't seen, and graded in the open. When a result can't earn its grade, the run stops — it never ships a number it can't stand behind. Each model's page shows exactly what its results look like, and how it earns them.