Solutions / Segmentation

Customer segmentation (RFM/ML)

Your whole base clustered by behaviour and value, recalculated daily, with each segment described in plain language and pushed into the CRM and the ESP as a field you can target.

Marketing CRM ML
100% of the base
segmented, refreshed daily
Segments described
in words, not cluster numbers
In your tools
CRM, ESP and BI, not a slide
Who it is for

CRM and marketing teams personalising at a scale where sending everyone the same message has clearly stopped working.

Short answer

RFM and behavioural clustering are run over your transaction and engagement history, producing segments that are stable enough to plan campaigns around and specific enough to be worth targeting. Each segment comes with a plain-language description, its size, its value and its typical behaviour, and every customer’s segment is written into the CRM and the ESP daily — including the movement between segments, which is often the more useful signal.

The problem

The base is split into new, active and lapsed, and every campaign goes to whichever of the three is largest.

01

Three segments is not segmentation

Active covers a customer who buys weekly and one who bought twice last spring. Any message written for that group is written for nobody in particular.

02

The segmentation lives in a slide

A consultant produced beautiful personas that exist as a PDF. Nothing writes them onto a customer record, so no campaign can actually use them.

03

Segments are recalculated once a year

A customer who lapsed in March is still marked loyal in October. Acting on a stale segment is worse than acting on none, because it carries false confidence.

How it works

From trigger to result, step by step.

01

Start with RFM, because it works

Recency, frequency and monetary value on clean transaction data give a strong, explainable base that everyone in marketing already understands. Anything more elaborate has to beat it.

02

Add behavioural clustering where it earns its place

Category mix, discount sensitivity, channel preference, basket size, return rate. Clusters are checked for stability across periods — a segmentation that reshuffles every month cannot be planned around.

03

Describe each segment in words

Size, value, typical behaviour, what distinguishes it, and what to do with it. A segment nobody can describe out loud will never be used, whatever its silhouette score.

04

Write segments into the working tools

A segment field per customer in the CRM and the ESP, refreshed daily, plus the previous segment so movement is visible. Campaigns then target segments without anyone exporting a list.

05

Track migration and value

Movement between segments each month is the early signal: loyal customers drifting toward lapsed is visible long before churn shows up in revenue. Segment value is reported monthly.

Before / after

What changes on the ground.

Today, by hand
×New, active and lapsed as the whole model
×Personas that live in a slide deck
×Segments refreshed once a year
×Campaigns aimed at whichever group is biggest
With the automation running
›Behaviour and value segments across 100% of the base
›A segment field on every CRM and ESP record
›Daily recalculation, with movement tracked
›Campaigns planned per segment with known value
What you get

Delivered, not demoed.

›An RFM baseline and behavioural clusters validated for stability
›A written description, size and value for every segment
›Daily segment and previous-segment fields in your CRM and ESP
›A monthly migration and segment-value report
›A working session so the marketing team owns the segments
Built with

We build in your stack rather than moving you onto ours. The list below is what this solution most often connects to — other systems are a scoping question, not a blocker.

Python scikit-learn pandas BigQuery dbt Airflow HubSpot Klaviyo Power BI
Time to production4–6 weeks
Build priceFixed quote
First stepFree mini-audit
Honest limits

When this is not the right solution.

·With a few hundred customers you should be talking to them, not clustering them. The insight is available directly and it is better.
·If your channels cannot target a segment field, the segmentation cannot be acted on, and connecting the channels is the first project.
·If you sell to a handful of large accounts, account planning beats statistical segmentation and costs a great deal less.

Questions we get about this one

Usually five to nine. Fewer and they are too coarse to write for; more and the marketing team cannot hold them in their head, so half go unused.

Bring us the process that hurts.

The mini-audit is free: we take your version of this process apart and tell you plainly whether automating it pays. If it does, you get a scope and a fixed price.