Case 4 · Customers¶
Data used on this page
Online Retail II - real public data · NPS survey (simulated) - synthetic. Details on the data sources page. With synthetic data the findings illustrate the method; they are not claims about a real organisation.
Context¶
The same retailer as case 3. 5,852 identified customers placed orders between Dec 2009 and Dec 2011; guest checkouts have no customer id and represent 13% of revenue (they are excluded here and measured in case 5). An NPS survey reached the customers active in the last 12 months (1,303 responses, simulated).
Question¶
Are we keeping our customers - and whom should we call first?
KPIs · last 12 months to 09 Dec 2011¶
Retention is the outcome; revenue at risk is the leading indicator and, more usefully, a list of names. NPS adds the customer's voice - read with its sample size.
Finding¶
Show the data behind this chart
| quarter | value |
|---|---|
| 2010Q1 | 63% |
| 2010Q2 | 61% |
| 2010Q3 | 58% |
| 2010Q4 | 63% |
| 2011Q1 | 42% |
| 2011Q2 | 59% |
| 2011Q3 | 61% |
Show the data behind this chart
| segment | customers | revenue | share |
|---|---|---|---|
| Champions | 1096 | 5526916.21 | 66% |
| Loyal | 809 | 1324116.53 | 16% |
| At risk | 652 | 795425.83 | 9% |
| Hibernating | 1052 | 501528.80 | 6% |
| Needs attention | 340 | 154663.49 | 2% |
| New / promising | 312 | 126274.66 | 1% |
Show the data behind this chart
| segment | n | nps | moe |
|---|---|---|---|
| At risk | 193 | -28.5 | 10.7 |
| Hibernating | 348 | -18.1 | 8.0 |
| Needs attention | 92 | 10.9 | 15.8 |
| New / promising | 101 | 21.8 | 15.9 |
| Loyal | 244 | 32.8 | 9.2 |
| Champions | 325 | 48.9 | 7.2 |
The NPS survey is simulated and its scores were drawn per RFM segment, so the link between segment and NPS is an assumption of the generator, not a finding. The chart shows how to read NPS: by segment, with n and a margin of error.
Decision¶
- Win-back campaign on the 'At risk' list, ordered by revenue. The first 104 customers cover half of the revenue at risk - a call list for account managers, not a mass e-mail.
- Close the loop with detractors within 48 hours of their answer; their verbatims feed the service and inventory cases (late or short orders are the usual reasons).
- Measure retention by acquisition cohort before setting a single target: new cohorts behave differently from long-standing wholesale customers.
- Reduce guest checkouts (13% of revenue has no customer id) - every anonymous order is invisible to retention and RFM (see case 5).
How we keep measuring¶
| KPI | Owner | Cadence | Target | Trigger for action |
|---|---|---|---|---|
| Revenue at risk (RFM "At risk") | CRM Lead | Monthly, with the customer list (leading) | ≤ 10.0% | Above 20% - extend the campaign |
| Customer retention (quarter over quarter) | Customer Success / CRM Lead | Quarterly, per acquisition cohort | ≥ 50.0% | Below 40% |
| Repeat purchase rate | CRM Lead | Monthly, rolling 12 months | ≥ 60.0% | Below 50% |
| Net Promoter Score (simulated survey) | Customer Experience Lead | Per survey wave, with n and margin of error | ≥ +30 | A drop larger than the margin of error |