My method
You can't act on Clients you don't see. So we start by seeing them clearly, then hand that reading to your team, then put it to work. Three steps in order, each one standing on the last.
01
Client Portfolio Dynamics
First we read the database you actually have (not the one you wish you had): what's usable, what's missing, what can be answered today. Then we read the Clients themselves, sorted by what they do rather than what they bought:
- how many are new
- how many stayed
- how many came back
- how many slipped away
…and what each group is worth.
One year, up close.
New - 34%
first purchase this year
Retained - 51%
bought this year and last
Reactivated - 15%
back after a gap
Each square is 1% of your active Client base. Illustrative example, inspired by experience but not real Client data.
2023 · 200
2024 · 250
2025 · 290
Active Clients per year
A major KPI: the Retention Rate
How many Clients of last year purchased again this year?
2023 → 2024
61%
2024 → 2025
59%
Illustrative example, inspired by experience but not real Client data.
The engagement matrix
Since Covid, the best luxury houses have shifted their attention from what a Client buys to what a Client feels. Engagement is emotion made visible, and the way to read it is to ask what a signal costs the person sending it.
Money is the counterintuitive part: for a wealthy Client it's the cheapest thing to give.
- Attention costs more.
- Time costs more again.
- Taking an initiative, reaching out unprompted, costs more still.
- And endorsing you to someone else, putting their own name behind yours, is the most expensive signal there is.
Those are the layers, and they don't move together with spend. To visualise it, put value on one axis and engagement on the other: your Client base then separates into four groups, each needing different intentions.
Heavy Buyers
High value, weak bond. Here for the product, not for you.
Intention: create the bond
- Invite them to something they cannot buy
- Hold a smaller room for them
Ambassadors
Value and bond together. They bring others in.
Intention: protect and reward
- Open a door that stays closed to everyone else
- Give them a seat at a decision
Entry
New, or still at a distance. Everything is still to be built.
Intention: prove it was worth joining
- Show what makes you singular
- Make the first contact personal, unprompted
Enthusiasts
Strong bond, smaller spend. They fill the room and carry the mood.
Intention: recognise, then grow
- Offer the rare thing first, even on a smaller budget
- Take time to understand their potential
Value ↑
Engagement →
What counts as engagement isn't the same for two businesses. For a restaurant it might be who books rather than walks in. For an estate, who reads the harvest letter, who comes to the domaine, who brings a friend. Working out which signals matter for you, and which ones your data can already see, is one of the better conversations I get to have.
The data-gap map
The questions come out of our conversations, once I've seen your data. You come away with a map: what your records answer today, what they can't yet, and the short list of fields to start capturing so next year's read goes further than this year's.
Most of that list is small, a field at the till, a date recorded properly, a Client reference that doesn't change.
02
Transmission
Your team learns to produce the read themselves: how the Client groups are built, how to rebuild them next year, and how to read what changed. We work on your own data, not a training set, so what comes out of the session is usable the same week.
The format is decided together:
- one workshop or several
- one person to train or a whole team
- on site, or remote if your team is already comfortable with the data
The point isn't to make analysts out of them. It's that nobody has to wait for me, or for anyone, to know where the Client base stands.
03
Accompagnement
Each Client group gets an intention and a move to match.
- The ones who stayed get depth, something that recognises how long they've been with you.
- The ones growing get the next tier offered before they go looking for it elsewhere.
- The ones who slipped away get a reason to come back that isn't a discount.
Drawing on my years inside Luxury Maisons, I give your team the direction to turn these intentions into the kind of Client gestures those houses run as standard: the ones that deepen a Client's attachment to the Maison.
We set the cadence together, monthly, quarterly, or around your own calendar, and each round starts by looking at what the last one moved. Nothing here needs a new tool. It needs someone to decide who gets an action, and why, and then to check whether it worked.
How I work
Fifteen years in Client intelligence across luxury and retail, five at Chanel and two at Richemont, in Europe and Asia. An MSc in mathematics and applied statistics behind the analysis. The tools stay plain: Excel, R, and AI used the way you'd use any other tool, to go faster and see more, never as a black box. I'm certified in Anthropic's AI Fluency for Small Businesses, which mostly means I know where these tools are reliable and where they aren't.
Pseudonymisation
Your export reaches me pseudonymised. Every Client is a reference number, never a name. The table linking each number to a real person stays entirely on your side and never leaves your systems, so I can see that Client 4471 bought twelve times in three years and drifted last spring without ever knowing who they are. No names, emails, phone numbers or addresses touch my side.
It's pseudonymised rather than anonymised on purpose. Anonymised data can't be traced back to anyone, which also means you couldn't act on what it shows. Pseudonymised keeps the link, held by you, so when the analysis says Client 4471 is worth calling, you're the one who can put a name to it.
If building that extract is what's stopping you, I'll come to your office and we make it together. Nothing starts before an NDA is signed.
This setup keeps you compliant with the GDPR in Europe and the PDPO in Hong Kong: personal data stays with you, as its controller, and never transfers.
You've got the data. One message and we'll see what's in it.
Let’s talkA first conversation, no strings. You'll leave knowing what your data can tell you.