03 — How it works
Predictive analytics for every client, every day.
Rydra doesn’t respond to churn. It predicts it — and acts before the client is already gone.
The model
At the centre of Rydra is a stochastic probabilistic model.
Every client in your database is continuously classified into one of five behavioural states based on their visit history, visit frequency, average gap between appointments, and deviation from their personal baseline.
single visit, habit not yet formed
regular pattern, low churn probability
frequency declining, soft signal
significant gap deviation, intervention window open
beyond recovery threshold
What makes this different
Traditional analytics compares clients to each other. Rydra compares each client to their own historical pattern.
A client who visits every 6 weeks and hasn’t been in for 9 weeks is more at risk than a client who visits every 6 months and last came in 7 months ago. The gap ratio — how far beyond personal baseline a client has drifted — is the core signal.
No other tool in this category models it this way.
The five steps
Your client data comes in
Rydra connects to your existing booking, POS, or reservation system. Every visit, every transaction, every interaction feeds the model. You own the data throughout.
Every client is scored continuously
The Markov model classifies every client in real time. States update as behaviour changes. When a loyal client’s visit frequency begins to decline, the model catches it — often weeks before the client would appear on any traditional report.
The intervention window is identified
There is a precise period — typically 45 to 90 days after a client’s last visit — where a single well-timed message recovers them. After that window, recovery rates decline sharply. Rydra identifies the window for every client, every time, automatically.
A unique message is written and sent
Not a template. Not a discount offer sent to a segment. AI writes a specific message for each individual client — using their name, their history, their barber or stylist. It is sent via SMS. No human writes it. No human approves it. It fires within the intervention window.
Revenue is attributed to the action
When a client returns within 60 days of receiving a message, that revenue is attributed to the outreach. You see exactly: how many clients were contacted, how many returned, what it cost, what it returned. The ROI is a specific number — not an estimate, not an impression count.
What it covers
Hospitality
- Individual guest churn prediction
- Automated at-risk client recovery
- Repeat visit and frequency modelling
- 6-week demand forecasting by day
- Cancellation slot-fill — ranked client list, message written, sent in seconds
- First-party guest data ownership
Beauty & personal care
- At-risk client recovery before they churn
- Rebooking rate modelling and improvement
- No-show prediction and pre-visit confirmation
- Per-stylist and per-barber performance intelligence
- Client lifetime value modelling
- Direct booking migration away from platform fees
How Rydra is categorically different
What the current tools offer
- Scheduling — book and confirm appointments
- Transaction processing — order, pay, split
- Blanket promotions — send a discount to a list
- Retrospective reporting — what happened last month
- Segment-level marketing — broad strokes, no individual signal
What Rydra offers
- Individual churn prediction — who specifically is at risk, today
- Automated recovery — message written and sent in the intervention window
- No discounting — personalised outreach, not margin-eroding offers
- Forward-looking intelligence — what’s going to happen next week
- Client-level attribution — every dollar recovered, traced to the action