NRR as the north star. The honest SMB version.

The components before the number

Here's the useful part of having backed into NRR rather than leading with it: we were tracking everything that composes it long before we rolled it up. MRR, expansions, downgrades, and cancellations were on the board weekly and often daily, and each had an owner — sales and success carried their pieces. The number nobody was watching was the one that ties them together.

If you're a founder today, that's the easier path in: you almost certainly already track the components. NRR isn't a new measurement project — it's a roll-up of numbers your sales and success people are already responsible for. Which means adopting it as a north star is mostly a decision to look at the combined figure on purpose, with one person accountable for it, instead of letting it stay implied across four separate line items.

The honest SMB number nobody quotes you

The playbooks preach 120%-plus or you're broken. Our real ballpark was 90–95%. It wasn't great — and I want to be honest about why, because the reason matters more than the number. We didn't have many expansion levers from the start, so we were constrained no matter how well customer success executed.

This is the same lesson as the expansion piece, viewed through the metric: you cannot customer-success your way past an empty shelf. If your only levers are more seats and a plan upgrade, a 90–95% NRR isn't a sign your retention team is failing — it's the mathematical ceiling of your product surface. So before you set an NRR target, look at what you actually have to sell an existing customer. The benchmark you can hit is a product-strategy fact, not a motivation problem. Chasing 120% with two items on the shelf just demoralizes a team doing fine work against a ceiling someone else built.

Why it's still the cleanest number you have

None of that makes NRR overrated. It's an all-encompassing number that signals the strength of your customer base and the value you provide, in a single figure.

That thought experiment is why I'd make it the north star now. Keep 100% of your revenue from the customers you already have and you're fine. Become more valuable to them — seats expanding, plans upgrading, add-ons selling — and you're doing amazing, at 105% or higher. It's a very clean metric: it folds churn, contraction, and expansion into one honest verdict on whether your existing base is getting more or less valuable to you over time. Most metrics measure an activity. NRR measures the result of all of them.

Reading the components when it moves

Because we tracked all four pieces, it was easy to spot what was actually going on when the number moved — whether churn, contraction, or expansion was driving it. The one that needs judgment is contraction: accounts shrinking their spend without leaving. Sometimes that's pre-churn, an early tremor before a cancellation. Other times it's nothing of the sort — the customer is just downsizing their own business, trimming costs in a way that has nothing to do with you.

Telling those two apart is a CSM's job, not a dashboard's, and it changes the response entirely: pre-churn contraction triggers a save play; a genuine downsizing usually just needs to be understood and accepted. Don't burn save energy fighting a customer's own cost-cutting.

Who owns it, and where the point comes from

If a founder adopts NRR as their north star this week, here's the concrete version. You can manage it weekly or monthly; a good weekly proxy is Net New MRR, which moves with the same forces. The ownership map that worked for us, lightly idealized: the success team owns expansion, contraction, and churn; sales owns the MRR dollars coming in; and the overall number probably belongs to a co-founder, because no single function controls all of it.

On goals, you can run it either way. Early on ours were focused on churn and mitigating it; over time we built both expansion and churn into the goals. A combined retention number on one scorecard works; separate expansion and churn goals work too. What matters is that the components have owners, which — if you've tracked them all along like we did — they probably already do.

Then the real question: where does the next point of NRR actually come from? You find the weakest component dragging the number down and ask whether you can realistically move it. Usually it's one of two paths — experiment with ways to improve churn, or experiment with ways to improve expansion. Contraction is usually the lowest-leverage place to push: some of it you're already addressing if you're working on churn, and the rest is a customer cutting costs outside your control. Put your experiments where churn or expansion live, because those are the two levers you can actually pull.