Segmentation & targeting

A segmentation model of your customers, not the whole country.

Fifty One Degrees builds bespoke customer segmentation and predictive targeting on your own first-party data, modelling the people who actually convert for you.

Off-the-shelf tools classify the UK population. Fifty One Degrees models the people who actually convert for you.
Book a Discovery Call →
Built on your first-party dataOwned by youPoC in ~7 weeks
Living model
app.51d.ai / segments / live

Living Segmentation Model

Live
Your customer base · 6 segments, refreshed from behaviour
High-value loyalists24%
Growth prospects31%
Occasional buyers22%
At-risk / lapsing12%
Owned by you · synced to CRM & ad platforms · model AUC 0.88
51.4°N · 0.1°W
Trusted by growing UK businesses
Heatable
Freddie's Flowers
Stiltz
Resi
Equals Group
Panmure Liberum
The problem

Why doesn’t off-the-shelf segmentation work for you?

National, not yours
“We license Acorn or Mosaic, but the segments describe the whole UK, not our buyers.”
A generic national label, not a model of who converts for you.
Census-aged
“Our segmentation refreshes slowly and leans on the 2021 Census.”
Your market moves faster than a census cycle.
Never activated
“The data is stuck in the CRM and the warehouse.”
Insight that cannot be activated is decoration.
Rented, not owned
“We pay an annual licence and still own nothing.”
With a third-party tool, the model and the IP belong to the vendor.
The Living Segmentation Model

What is the Living Segmentation Model?

The Living Segmentation Model is Fifty One Degrees’ approach to customer segmentation. Rather than licensing a fixed, national, census-based classification, Fifty One Degrees builds a segmentation model on your own first-party data, enriched with third-party signals where they add value, that refreshes as your customers behave and is wired into the systems your teams use every day. Owned by you, specific to your business, and predictive, not just descriptive.

01
Owned, not rented
Built on your data; the model and IP transfer to you.
02
Living, not static
Refreshes continuously from real behaviour, not on a census cycle.
03
Activated, not shelved
Pushed into your CRM, ad platforms and operational systems so it drives spend.
What it includes

What does a segmentation engagement include?

Bespoke segmentation (the core).
Unsupervised clustering of your customer base into commercially meaningful segments: k-means (the industry standard the commercial tools are built on), plus self-organising maps (SOMs) and neural-network approaches for large, high-dimensional data, and deep-learning segmentation to refine the driver variables. Output: a segment scheme specific to your business, with pen-portraits, segment sizing, and every customer record tagged.
Geodemographic and property enrichment.
Third-party data is a useful input, not the answer. Enrichment from Census 2021, the ONS Output Area Classification (the free open alternative), Land Registry, EPC and property data, indices of deprivation, and licensed postcode-level classifications. For home-centric businesses, the property-intelligence angle.
AI for Home Services →
Propensity and predictive targeting.
From who they are to what they will do: propensity-to-buy, conversion and lead-scoring, customer lifetime value, churn, next-best-action and campaign-response models. The Data Science & ML spine applied to targeting.
Data Science & ML →
Activation and integration.
Segments and scores into the live environment: CRM (high-propensity to VIP routing, low to automated nurture), ad platforms (custom and lookalike audiences), direct-mail selections and the dialler. Built via model APIs with MLOps monitoring.
Conversational segment personas (emerging).
Conversational AI personas grounded in your real customer data and research, that marketing, product and CX teams can interrogate before committing spend. A bespoke, emerging capability.
Conversational AI →
Results

What targeting results has Fifty One Degrees delivered?

Indicative ranges across Fifty One Degrees targeting engagements.

~30%
improvement in sales productivity across Fifty One Degrees engagements
up to 40%
reduction in churn across Fifty One Degrees engagements
~15%
reduction in customer acquisition cost
~7 weeks
to a working model, PoC-first (~2 weeks discovery, 3 build, 2 deploy)
Case study · Resi

Resi: scoring registrations by conversion probability

SituationResi, the UK’s largest residential architecture platform, needed to prioritise the registrations most likely to convert.
ApproachFifty One Degrees combined Resi’s own first-party registration data with UK public datasets: the exact Living Segmentation Model thesis, built proof-of-concept-first.
SolutionA predictive model scores new registrations by conversion probability, embedded in the workflow so the team works the best leads first.
OutcomeEvery lead scored into a decile in real time, on a production model Resi owns.
Read the Resi story →
Resi
Fit

Who is this for?

Segmentation pays back wherever spend is targeted at people, and being precise about which people changes the return.

01
Financial services & insuranceAcquisition, affordability and risk-segmented targeting.
02
Retail & e-commerceRange, lifetime value, catchment.
03
Home services & improvementProperty-contextual targeting.
04
Subscription & recurring-revenue B2CChurn and lifetime value.
05
Charities & membershipSupporter segmentation and fundraising response.
06
PropertyTenant mix, catchment, investment.
Why Fifty One Degrees

Why trust Fifty One Degrees with segmentation?

This is not a sideline. Mark Somers, Fifty One Degrees’ co-founder, built 4most into the UK’s largest independent credit-risk and analytics consultancy, 200+ staff, a business built on segmentation and risk modelling at national scale. Nick Harding scaled Fluro’s customer-acquisition engine to process 4 million credit applications a year. The people behind this page built careers on exactly this, which is what separates Fifty One Degrees from both the data vendors and the generalist consultancies.

About Fifty One Degrees →
How it runs

How does a segmentation engagement run?

Proof-of-concept-first to de-risk. You own the model and the IP, and your team is trained to run it.

Weeks 1–2
Discovery & data audit
Audit your first-party data and agree the segmentation goal.
Weeks 3–5
Build, train & validate
Build and validate the model with a human in the loop.
Weeks 6–7
Deploy & integrate
Into your CRM and ad platforms with MLOps monitoring.
FAQ

Questions buyers ask Fifty One Degrees about segmentation

What is the difference between Acorn, Mosaic and CAMEO?

Acorn (CACI), Mosaic (Experian) and CAMEO (TransUnion) are the three main UK geodemographic classification tools. All segment UK postcodes and households into named consumer types from census and commercial data; they differ in data heritage (CAMEO leans on credit and affordability data via TransUnion, Acorn uses current-year data plus the 2021 Census), granularity and coverage. All three are off-the-shelf classifications of the whole population, not models of your specific customers. Fifty One Degrees builds the latter.

Should I license Acorn or Mosaic, or build my own segmentation?

Off-the-shelf is fast and useful as an input; bespoke is specific, predictive, owned and activated. Most businesses are better served by a model built on their own data, enriched with third-party signals where useful. The Fifty One Degrees Living Segmentation Model gives you a model of your actual customers that you own and that flows into your CRM and ad platforms, rather than a national classification you rent.

How is bespoke segmentation different from off-the-shelf geodemographic data?

Off-the-shelf describes who people are nationally and refreshes slowly; the Fifty One Degrees bespoke segmentation describes and predicts what your customers do, refreshes continuously from behaviour, and is built into your systems. One is a rented national average; the other is an owned model of your business.

What data do you need to build a customer segmentation model?

Your first-party customer and transaction data is the foundation. Fifty One Degrees enriches it with public and commercial sources (Census 2021, ONS Output Area Classification, property and EPC data, Land Registry) where they add predictive lift. PII is not always required, and you keep ownership of the data and the model.

What is the difference between segmentation and propensity modelling?

Segmentation groups customers by shared characteristics (descriptive); propensity modelling predicts a specific future action, such as buy, churn or respond (predictive). The Fifty One Degrees Living Segmentation Model combines both, so you understand your customers and can act on what they will do next.

How long does it take to build a customer segmentation model?

Typically around seven weeks end to end on a proof-of-concept-first basis: about two weeks of discovery and data audit, three weeks of model build, and two weeks of deployment and integration into your CRM and ad platforms.

Is the ONS Output Area Classification a free alternative to Acorn?

Yes. The ONS Output Area Classification (OAC) is the open, census-based classification from ONS and UCL. It is a legitimate free input, but it is census-only and not tailored to your business. Fifty One Degrees uses it as enrichment, not as a substitute for a model of your own customers.

Do we own the segmentation model, or does Fifty One Degrees?

You own it. With the Living Segmentation Model the model and the IP transfer to you, unlike a licensed classification where the vendor keeps ownership and you pay an annual fee. Your team is trained to run and refresh it.

Go deeper

Part of a broader practice

Customer segmentation is one application of the Fifty One Degrees data science practice.

SolutionData Science & MLRead on 51d.coCase studyFifty One Degrees partners with Resi on data scienceRead on 51d.co
Next step

See what a segmentation model of your own customers would reveal.

Book a 30-minute discovery call and we’ll show you what a segmentation model of your own customers would reveal.

Book a Discovery Call →