Back to Salon Industry Statistics
Data & tools

Retention Rate Benchmarks Checklist

An audit checklist to run before you trust a retention number: duplicate client records, missing visits, unwritten definitions, and unfair comparisons.

Cited Canadian evidence asset (checklist).

A retention figure is only as good as the records underneath it. Most of the work in producing a trustworthy number is not arithmetic; it is auditing the inputs before the division happens. Run this list in order, once, and then again whenever the figure moves in a way nobody can explain.

Audit the client list first

  • One record per person. Duplicates arrive when a client books by phone with a mobile number and online with a home number, and they double-count that client in the group who were due.
  • Choose an identity key and hold to it. A normalised mobile number is the usual choice, with email as a fallback for clients who book without one.
  • Remove the placeholders. Walk-in records, cash-sale records, and test entries made during setup are not clients, and they cannot lapse.
  • Mark inactive records. A client who asked to be removed, moved away, or died should be excluded rather than counted as a failure to return.
  • Note shared numbers. Two members of one household booking on a single phone need separate records with the number repeated, or they will look like one client who visits twice as often.
  • Check the display name field is usable. A list you cannot read by eye will not be checked by anyone, and an unchecked list drifts within a few months.

Audit the calendar

  • Every appointment lives in one system. A paper book beside an online calendar guarantees visits that never reach the count.
  • Cancellations are logged as cancellations. Deleting them is the single most common reason a rate looks better than the business feels.
  • Reschedules are logged as reschedules. The original slot was released and a new one taken; treating that as a cancellation and then counting the new visit describes the same client twice in two directions.
  • No-shows are marked as no-shows. They are a different failure from a cancellation and they belong on their own line.
  • Past appointments are not edited after the fact. Corrections belong in a note, not in a silent change to a date, because the history is the evidence.

Audit the definition in writing

  • Who counts as due, in one sentence
  • How long the window is, and why that length was chosen
  • Whether a cancelled-then-rebooked visit counts as a return
  • Which date keys the cohort, when a cohort is being used
  • Where the definition is stored, so the next person to run the numbers uses the same one

A definition that lives only in someone's head leaves with that person.

Audit the comparison

  • Same definition, same window, same segments on both sides of any comparison.
  • Season against season. This month against the same month a year earlier survives a slow stretch; this month against last month mostly measures the calendar.
  • Small groups move. Treat a change in a group of a few clients as a question, not a finding.
  • Note what else changed in the same month: a price change, a new staff member, a booking channel switched on or off, a holiday week.

Audit the follow-up

The checklist is not finished when the number is correct. It is finished when someone has read it.

  • One named person owns the figure and its explanation.
  • The lapsed list is worked, not just counted, with a call or a message to each client.
  • The rebooking-at-the-desk figure sits beside the rate, because that is the lever the front desk controls.
  • Last month's notes are read before this month's number is discussed.

Elevoi supports the last item on that list: it answers calls, offers openings to the clients who ring back, and records what happened against the client. The retention rate benchmarks calculator covers the arithmetic once the inputs pass this audit.

Frequently asked questions

What is the first thing to check before calculating retention?

Whether one person appears once in the client list. Duplicate records inflate the group who were due and pull the rate down, and no amount of careful arithmetic fixes a list that counts the same client twice.

Do deleted appointments matter?

Considerably. A cancellation that was deleted instead of logged leaves a client looking as though they were never due, which quietly flatters the rate. Log the event and keep the record.

How many months of history are needed before the checklist is worth running?

Enough that each client in the group has had a fair chance to return within the window. Below that, the checklist still catches data faults, but the rate it produces describes too short a stretch to act on.