Logo churn, revenue churn, NRR: what matters under $100k#

Under $100k MRR, logo churn - the plain count of subscribers who cancelled, divided by subscribers you started with - is the metric to run monthly and act on. Revenue churn earns a quarterly look, mostly to catch downgrade leaks and whale risk. NRR, the enterprise SaaS favorite, is close to noise at small scale: without meaningful expansion revenue it collapses into 100 minus gross churn, while single customers swing it by whole points.

Subscription analytics content is written for VC-backed SaaS, so small WooCommerce stores inherit metrics designed for companies with sales teams, seat expansion, and ten thousand accounts. Most of that inheritance is dead weight. Here is which of the three headline churn metrics deserves your attention below $100k MRR, which is a quarterly hygiene check, and which you can ignore until your business changes shape.

The three metrics, defined#

Quick definitions so the rest of the post is unambiguous; the glossary has the full set.

Baremetrics, whose revenue churn guide these formulas match, draws the key distinction plainly: customer churn counts people cancelling, revenue churn counts dollars lost including downgrades and failed charges. NRR then nets expansion against those losses. Three lenses, one underlying event stream.

Logo churn: the one to run monthly#

For a store under $100k MRR, logo churn is the primary metric because its assumptions actually hold at your scale. Three reasons.

First, your customers are roughly equal-weighted. A WooCommerce subscription store typically spans a narrow price band - maybe $25 to $60 a month - not the 100x spread between an enterprise SaaS starter seat and a six-figure contract. When every customer is worth about the same, counting customers and counting dollars tell the same story, and counting customers is simpler, faster, and harder to fool.

Second, logo churn is the direct output of the thing you can influence this month. Every unit of logo churn is a person who clicked cancel or a card that failed. Cancel-flow changes, cadence options, dunning - each shows up in the count within one or two billing cycles. It is worth splitting the count into voluntary and involuntary while you are at it, using Recurly's definitions: voluntary is an active cancellation, involuntary is a payment failure. The two move with completely different levers, and their network data shows involuntary churn is roughly a third of the total, which is too large a slice to leave unlabeled.

Third, it is the most noise-resistant of the three at small n - and it is still plenty noisy, as the sample-size section below shows.

Measure it as cancels during the month divided by active subscribers at the start of the month, and count pending-cancellation subscribers as churned the day they cancel, not when the prepaid term runs out. The churn rate calculator handles the arithmetic and the annualization if you want to sanity-check your spreadsheet.

Revenue churn: the quarterly hygiene check#

Gross revenue churn matters at small scale only where it diverges from logo churn, and it diverges in exactly three situations: you sell at meaningfully different price points, customers can downgrade between tiers, or a handful of large accounts dominate MRR. If none of those describe your store, revenue churn is logo churn wearing a dollar sign.

When they do diverge, the divergence is the signal. Logo churn steady at 5% while revenue churn runs 8% means your expensive customers are the ones leaving, or downgrades are leaking dollars invisibly - Baremetrics' guide makes precisely this point, noting that customers not cancelling does not mean you are not losing money. Their observed range across SaaS and subscription companies is roughly 4 to 8% monthly revenue churn, wide enough that they recommend benchmarking against similar-ARPU businesses rather than the blended average.

Price point turns out to be one of the strongest churn predictors in public data. Recurly's benchmark research (median annual churn, July 2026 network data) shows total churn falling as average revenue per customer rises - 4.29% for $10-25 products against 2.87% for $100-250 - with involuntary churn dropping from 1.30 points to 0.46 across the same span. Cheap subscriptions ride on weaker cards and weaker commitment. If your store spans tiers, expect the cheap tier to churn structurally worse, and read revenue churn per tier before concluding anything.

Quarterly is the right cadence: downgrade leaks develop over months, and at small scale a single month of revenue churn is one whale's decision away from meaningless.

The annual-plan wrinkle#

If you sell annual or prepaid multi-month plans alongside monthly, every metric above needs one adjustment: segment by billing interval before computing anything. An annual subscriber can only churn in their renewal month. Blending them into monthly logo churn deflates the rate all year - they sit in the denominator for twelve months while contributing at most one cancellation event - and then a cluster of January renewals makes January look like a crisis.

The clean pattern for a small store is two numbers instead of one blend: monthly-plan logo churn, measured monthly as above, and an annual renewal rate, measured as renewals divided by subscriptions up for renewal that month. The second is a tiny sample almost by definition, so report it quarterly or half-yearly as a running total. Same events, same spreadsheet tab, but the two populations stop contaminating each other's rates.

NRR: an enterprise metric in a small-store costume#

Net revenue retention is the headline metric of subscription investing, and the benchmarks around it are all drawn from a world that is not yours. Paddle's NRR guide pegs 109% as the target worth aiming for, cites public SaaS averaging around 114%, and lists IPO-era standouts like Snowflake at 158%. Those numbers are achievable only because enterprise SaaS has enormous expansion surface: more seats, higher tiers, usage growth.

A WooCommerce subscription store mostly does not. Your expansion events are cadence upgrades, box-size bumps, and the occasional cross-sell - real, but thin. With expansion near zero, the NRR formula degenerates: NRR is approximately 100 minus gross revenue churn. At that point the metric adds no information you did not already have, and it subtracts some, because netting hides composition. An NRR of 97% could be 3% churn with no expansion or 8% churn papered over by 5% expansion - radically different businesses, same number.

The noise problem is worse for NRR than for the other two because both tails move it. One $200/month account upgrading or cancelling in a 150-subscriber store swings NRR by multiple points in either direction. A monthly NRR series at small scale is a random walk with a story attached.

When does NRR start mattering? When expansion revenue becomes a real motor: a genuine multi-tier ladder, usage-based pricing, or a deliberate upsell program - typically alongside crossing into six-figure MRR. Until then, track gross numbers and skip the netting.

The small-sample problem nobody adjusts for#

All three metrics share a failure mode below $100k MRR: the denominator is small, so ordinary randomness produces movements that look like trends. This is arithmetic, not opinion - here is an illustrative store.

At 150 subscribers, one cancellation is 0.67 points of monthly logo churn. Suppose the true underlying rate is 5%, meaning about 7 or 8 cancels in an average month. Perfectly ordinary monthly draws around that rate will land anywhere from 4 to 11 cancels - which your dashboard renders as churn "swinging" between 2.7% and 7.3% while nothing whatsoever changed. A merchant who reacts to that swing ships a fix for a problem that does not exist, then credits the fix when the number regresses to the mean.

Baremetrics' LTV guide gives the same warning from the statistics side: with under 100 customers, they suggest a metric needs to draw on at least half your user base before it is worth trusting. The practical translation for churn metrics:

• Judge on a 3-month rolling average, never a single month.
• Think in counts first, rates second. "9 cancels against 7 expected" invites the right question - who were they? - where "6.2% churn" invites panic.
• Before reacting to any move, ask whether it is at least 3 or 4 customers beyond normal. If not, wait a month.

This is also why we keep repeating that the early ChurnStop install cohort is too small to publish benchmarks from - the same arithmetic that makes your monthly churn noisy makes our fleet-wide averages noisy, and pretending otherwise would be exactly the dashboard theater this post is arguing against.

Cadence, by store size#

A sensible measurement schedule, assuming a single-product store in the common WooCommerce price band:

Weekly churn tracking is absent from the table on purpose. At every size in it, a week of churn data is a handful of events, and a handful of events is an anecdote.

The one-sheet setup#

One spreadsheet tab, updated on the first of each month, beats any dashboard at this scale:

1. Subscribers at month start, new, cancelled (voluntary), cancelled (failed payment), month-end count.
2. Logo churn for the month, plus the 3-month rolling average next to it - the rolling column is the one you act on.
3. A short notes column: promos running, price changes, anything that explains a cohort. Numbers without annotations rot.
4. Quarterly, add MRR lost to cancels and downgrades for the revenue churn check.
5. Once your counts are trustworthy, feed them forward: churn drives lifetime math, and the LTV spreadsheet post shows how to turn these same columns into a defensible CAC ceiling.

The discipline is the point. Small stores do not fail at churn measurement because they lack metrics; they fail because they track too many, weekly, and react to noise. One honest count, split by cause, smoothed over a quarter, will out-inform an NRR chart every month of the year.