Clarivoxx Insights · Edition No. 16
The Retention Dashboard I Use: NRR, GRR, Churn and Expansion
Abdessamad Ghanem · September 23, 2026 · Updated
A retention dashboard shows how the recurring revenue of a fixed group of existing customers changed over one period, split into churn, contraction and expansion, with logo churn beside it. NRR, GRR, revenue churn and expansion rate are four views of that one revenue bridge, so NRR = GRR + expansion rate whenever all four use the same customers, the same period and the same revenue basis.
In the last edition I covered how to tell whether a retention playbook really worked. This edition is about reporting the result as one story instead of four separate numbers.
Why do retention numbers disagree?
They usually come from different lists.
I have sat in review meetings where Finance and CS each brought a retention figure, and half the hour went to arguing about whose was right. Both were correct. The math was fine on both sides; the customer list, the period or the revenue basis was not the same. A common split: one team uses annual ARR for every customer billed that year, the other uses monthly MRR and counts customers signed during the quarter.
Which formulas tie the dashboard together?
GRR = (Start − Churn − Contraction) ÷ Start × 100
Expansion rate = Expansion ÷ Start × 100
NRR = GRR + Expansion rate
Gross revenue churn = 100% − GRR
Customer churn = Customers lost ÷ Customers at start × 100
Start = MRR of customers active on day one (new customers excluded)
Churn = MRR of customers who cancelled
Contraction = MRR lost to downgrades from customers who stayed
Expansion = MRR added by customers who stayed
The GRR formula article and customer churn vs revenue churn go deeper on each definition.
What does one quarter look like?
An illustrative quarter: 400 customers and $200,000 in MRR at the start. During the quarter, 12 customers cancel ($9,000), customers who stay contract by $5,000, and they expand by $21,000 ($12,000 seats, $6,000 cross-sell, $3,000 price increase).
| Metric | Calculation | Result |
|---|---|---|
| Customer churn | 12 ÷ 400 | 3% |
| Gross revenue churn | $14,000 ÷ $200,000 | 7% |
| GRR | $186,000 ÷ $200,000 | 93% |
| Expansion rate | $21,000 ÷ $200,000 | 10.5% |
| NRR | $207,000 ÷ $200,000 | 103.5% |
The check: 93% + 10.5% = 103.5%, so the bridge reconciles.
One cohort, four retention metrics
NRR = GRR + expansion rate. Here 93% + 10.5% = 103.5%. Gross revenue churn = 100% − GRR = 7%. The identities hold only when every metric uses the same opening cohort, period and revenue basis.
Read the numbers together, too. Logo churn of 3% next to revenue churn of 7% means the average lost account paid $750 a month, against a cohort average of $500. You are losing bigger customers.
Then stress-test expansion. Without the price increase, NRR falls to 102%. If two accounts produced $10,000 of the expansion, remove them too and NRR drops to 97%. A healthy-looking headline can rest on very few accounts. Try each scenario in the NRR and GRR calculator, and check the monthly revenue churn with the MRR churn calculator.
The quarterly revenue bridge
Without the $3,000 price increase and two accounts adding $10,000, illustrative NRR falls to 97%.
How do I lay out the dashboard?
Four panels, in this order: the headline row (GRR, NRR, customer churn, expansion rate), the revenue bridge, logo churn against revenue churn, and expansion by source with the share from the top accounts. I always put GRR next to NRR, label the period on every tile, and split the bridge by plan or customer size so a strong segment can't hide a weak one.
Try this this week
- Freeze the list of customers active on day one of last quarter, with their opening MRR.
- Classify each account's change as churn, contraction or expansion. An account that downgrades and then cancels counts once, as churn.
- Confirm that NRR = GRR + expansion rate. If it doesn't, find the account that breaks it.
- Write the definitions down and agree on them with Finance.
Once the numbers add up, the argument about which figure is right usually stops.
This edition closes the retention metrics run. Next, we move to support operations.
Read next