The Customer Health & RFM panel on the Insights tab gives every customer a score out of 100 and a named segment, worked out from how recently they ordered, how often they order, and how much they spend relative to your other customers. It is a Pro feature.

The three scores

Each part is scored out of five.

Recency

• 5 – ordered within the last 30 days
• 4 – within 60 days
• 3 – within 120 days
• 2 – within 240 days
• 1 – longer ago than that

Frequency

• 5 – 10 or more orders
• 4 – 6 to 9 orders
• 3 – 3 to 5 orders
• 2 – 2 orders
• 1 – a single order

Monetary

Spend is compared against the average lifetime value of a customer on your store, so this is relative rather than absolute. Three times the store average scores 5, twice scores 4, at or above average scores 3, at least half scores 2, and below that scores 1.

Where the store average cannot be calculated, the score falls back to a flat scale based on total spend alone.

The health score

The three scores are combined into a percentage, weighted 40% recency, 30% frequency and 30% monetary. Recency carries the most weight because a customer who has stopped buying is the most urgent thing the panel can tell you.

The segments

The segment is chosen by working down this list and taking the first match, so a customer with a refund problem is flagged as that regardless of how well they score elsewhere:

• Needs Review – 30% or more of their orders were refunded.
• Dormant – no order for over a year.
• At Risk – past their usual reorder rhythm, meaning the gap since their last order is more than 1.75 times their average gap between orders.
• VIP – a health score of 85 or more with at least three orders.
• Loyal – a health score of 70 or more.
• New Customer – one order or fewer, placed within the last 45 days.
• Discount Driven – 60% or more of their orders used a coupon.
• Active – useful order history with no urgent risk signal.
• No Orders – nothing matched the report’s filters.

Reorder prediction

Alongside the score, the panel shows the expected next order date and a reorder status. The expected date is the customer’s last order plus their average gap between orders, so a customer needs at least two orders before a prediction can be made. When that date has passed, the reorder is reported as overdue.

Because the score depends on the store average, it is calculated over the same date range and status filter as the rest of the report.

Related

• Store comparison benchmarks (Pro)
• Recommended actions (Pro)
• The Insights tab