Ask Holdfast AIbeta
Holdfast Intelligence

NewHoldfast Retention Lab · Retail

See which customers are drifting away, what they are worth and how to win them back.

In retail, most customers don’t leave with a goodbye. They come less often, spend less, or move part of their basket to a competitor. Retention Lab combines machine learning, AI and our consumer research to measure that silent churn, predict it and act on it before it shows in your sales.

What you will measure

  • Probability each customer is still active
  • Partial churn: customers buying less
  • Cohort retention
  • Customer lifetime value
  • Revenue at risk
  • Repeat purchase rate
01

Diagnosis

Where you stand today

The six retention indicators by segment, region and store format, built from your own transaction history.

02

Prediction

Who is at risk, and why

A risk and value score for every customer, with the reasons behind it explained in business language.

03

Simulation

What retention is worth

Scenarios in money: what keeping more customers would mean for your revenue and margin, before you commit a budget.

04

Action

What will bring them back

Test win-back offers and messages with Synthetic, and ask at-risk customers directly with Pulse, before you invest.

The engine, in plain words

Proven machine learning models, read by senior researchers.

Every model is checked against your own history before we use it, and every result is reviewed by a senior analyst before it reaches you.

Purchase-pattern models (BG/NBD)
Estimate whether each customer is still active, from how often and how recently they buy.
Value models (Gamma-Gamma)
Project how much each customer is likely to spend in the future.
Gradient boosting with SHAP
Rank the risk of each customer and show which factors are driving it.
Survival analysis
Estimate when a customer is likely to lapse, not only whether.
Time-series forecasting
Project revenue at risk month by month.
Language models on open text
Read surveys, reviews and service notes to explain the why behind the numbers.

What we need from you

  • 24 to 36 months of transactions with an anonymised customer ID.
  • Store, channel and product for each purchase.
  • Optional: loyalty programme, CRM, surveys and reviews.

How we protect it

  • Customer data is anonymised before it reaches us.
  • A data processing agreement under GDPR or your local law.
  • A working environment agreed with your IT team.
  • Your data is never used to train third-party models.

How we start

An 8-week pilot, with a fixed scope and price.

One region or store format, one data extract, and a clear answer at the end.

  1. Weeks 1–2

    Data and agreements

    Scope, data agreement and extract from one region or format.

  2. Weeks 3–5

    Diagnosis and models

    The six indicators, cohorts and models validated on your history.

  3. Weeks 6–8

    Prediction and plan

    Risk scores, simulation in money and prioritised actions.

You receivean executive report, your customer base with risk scores, the simulator and an action plan. After the pilot, a monthly retainer keeps scores, dashboard and simulations up to date, with a monthly session with a senior consultant.

Why Holdfast Intelligence

Prediction tools tell you who. We focus on why, and what to do next.

Research methodology at the core

Our foundations come from more than two decades of customer and market research at Qualtrics, Ipsos, GfK and Nielsen. Holdfast’s AI engine does the heavy lifting; proven methodology keeps the answers reliable.

Explanations, not just scores

A risk score without a reason is hard to act on. Every insight comes with the drivers behind it, in plain language your team and your board can understand.

Works with the tools you have

No rip-and-replace. We build on your existing CRM, product analytics, support and survey data, so your team keeps its current workflows.

European and privacy-first

Based in Paris and designed for European companies, with GDPR and data protection built in from the first day rather than added later.

Built around measurable outcomes

We agree on the retention metrics that matter to you at the start, and measure our work against them. If it doesn't move the numbers, it isn't working.

Specialists, not generalists

We do not sell broad transformation programmes. We focus on one thing: helping companies understand, keep and grow their customers.

The problem

Most companies find out a customer is leaving when it is already too late.

Signals arrive too late

Satisfaction scores and alerts flag risk once behaviour has already changed. By then, the decision has usually been made.

Data lives in silos

Sales data, CRM, service interactions and surveys each tell part of the story. Nobody sees the whole customer.

Alerts without answers

Knowing an account is at risk is not enough. Teams need to know why, and which action is most likely to work.

Our approach

Research-grade customer understanding, delivered at the speed of AI.

We combine two decades of customer research methodology with modern AI, so your team gets explanations it can trust and act on, not another dashboard.

01  ·  Connect

One view of every account

We bring together usage, CRM, support and feedback data into a single picture of each customer relationship.

02  ·  Explain

The reasons behind the risk

AI models trained on your history surface early warning signs and, crucially, the drivers behind them.

03  ·  Act

Clear next steps for your team

Each at-risk customer comes with a recommended action, so your teams know exactly where to focus.

For retailers: discover Holdfast Retention Lab →

Use cases

Built for the decisions that protect revenue.

From financial services and retail to technology and SaaS, each use case answers a question your leadership team already asks every quarter.

Customer teams

Early churn warning

Spot customers drifting away before they leave or reduce their spend, with the specific behaviours and feedback that explain the risk.

“Which customers should we save this month?”

Finance & Commercial

Revenue and renewal forecasting

Replace gut-feel estimates with a forecast grounded in behaviour, sentiment and engagement data across your whole customer base.

“How much revenue will we really keep next quarter?”

Account Management

Expansion signals

Identify loyal customers showing the patterns that typically precede upgrades, cross-selling or higher spend.

“Where is our next growth revenue?”

Product & CX

Voice of the customer, at scale

Holdfast’s AI engine reads every survey answer, support ticket and call note, then turns them into themes ranked by their impact on retention.

“What are customers really telling us?”

Onboarding

Onboarding risk

Detect new customers who are not reaching value fast enough, while there is still time to change the outcome.

“Which new customers are stuck?”

Leadership & Board

Retention narrative for boards and investors

Explain customer retention and loyalty with evidence: what drives it, which segments are at risk and what the plan is to improve it.

“What do we tell the board about retention?”

The business case

Every point of churn is worth more than it looks.

Keeping a customer is almost always cheaper than winning a new one, and retained revenue compounds year after year. Customer retention is also one of the clearest signals boards and investors use to judge the health of a business. Small improvements add up fast.

Estimate your retention value

€20M
10%
2.0 pts
Revenue lost to churn each year today €2M
Revenue protected each year €400K
Share of today's churn recovered 20%

Simple arithmetic for illustration, not a guarantee of results. Excludes expansion revenue and the compounding effect over multiple years.

Three illustrative scenarios

SaaS scale-up

€10M ARR, 12% annual churn: €1.2M lost every year.

€200K

protected per year by reducing churn 2 points, roughly the cost of two senior hires.

B2B distribution company

€25M from repeat customers, 10% lost each year: €2.5M.

€375K

protected per year by reducing customer loss 1.5 points, before any growth in spend.

Insurance or banking portfolio

€50M in annual premiums or fees, 8% lapse rate: €4M lost every year.

€500K

protected per year by reducing cancellations just 1 point, with a direct effect on profitability.

Hypothetical examples based on simple arithmetic. They are not client results.

Early partners

Help shape how companies keep their customers.

We are working with a small group of customer, CX and commercial leaders across industries. If retention is on your agenda, let's talk for 25 minutes about how you handle it today.

Book a 25-minute conversation
Back to top