Responsible AI

For upcoming appointments and active clients, we calculate a risk score (for example, cancellation, no-show, or churn) from your booking behaviour with the salon — such as prior cancellations or no-shows, whether a deposit is held, how far ahead the booking was made, and how the timing compares to your own usual visit rhythm. Scores let the salon send a timely reminder, offer a reschedule, or open a standby slot.

Scores are built only from behavioural and transactional signals. We deliberately exclude — and block at the data layer — protected attributes and their known proxies, including race, ethnicity, gender, age, religion, nationality, disability, and location-based proxies such as postal code. This exclusion is reviewed in periodic fairness reviews.

No solely-automated decision is made where it would have a legal or similarly significant effect on you. Actions that could disadvantage a client — such as requiring a deposit or prepayment — are never applied automatically: they are queued for a salon team member to review and approve. Automated touches are limited to reminders and optional offers.

Each score stores human-readable reason codes describing the top factors behind it (for example, “2 prior no-shows” or “overdue versus usual 4-week cadence”). Salon staff see these alongside every prediction, and you can request the explanation behind a decision that affects you.

• Opt out of profiling — stop risk scoring of your account entirely.
• Opt out of AI-driven messaging — stop predictive reminders, confirmation prompts, and win-back offers (you still get the booking confirmations and reminders you choose).
• Request human review — require a person to review any automated decision that affects you.

Manage these any time in your account settingsunder “Automated decisions & privacy”. Changes take effect immediately. See also the Privacy Policy for your GDPR Article 22 and CCPA rights.

AI-driven messages respect consent (opt-in where required by CAN-SPAM / TCPA), shared frequency caps, and quiet hours. Every message offers a one-tap opt-out, and replying STOP to any SMS stops texts immediately at the carrier level.

A portion of clients is held out of automated actions as a control group so we can measure whether interventions genuinely help. We monitor accuracy and calibration against real outcomes, and review the model and its inputs for fairness on a periodic cadence.