Analytics · 9 min read · 2026-10-11
Logo Retention vs Revenue Retention: When They Diverge
A SaaS cohort can lose 40% of its customers and still grow its revenue. That is not a paradox or an accounting trick — it is what happens when the customers who stay expand faster than the ones who leave shrink the base. As one retention guide puts it bluntly, "you can have 95% logo retention and still be shrinking if those retained customers are downgrading," and conversely "85% logo retention with 115% net revenue retention means you're growing despite the churn" (UserJot, SaaS retention metrics).
When logo retention and net revenue retention point in opposite directions, a single headline number lies to you. This post is a diagnostic: why the two metrics diverge, which direction of divergence is dangerous, and how to instrument both so you act on the real signal instead of the comforting one. It builds on the definitions in churn rate measurement: logo vs revenue, gross vs net, but the focus here is the divergence itself — the moment the two curves split and one of them is quietly wrong.
Two metrics, two questions
Logo retention and revenue retention answer different questions, which is why they can disagree. Logo retention counts accounts: of the customers you started the period with, how many are still customers at the end, regardless of what they pay. Net revenue retention counts dollars: of the recurring revenue you started with from that group, how much remains after churn, downgrades, and expansion. A $5,000-a-month account and a $50-a-month account are one logo each in the first metric and a hundredfold apart in the second. The instant your customer base is not uniform in size — which is every real SaaS — the two metrics stop tracking each other.
The divergence is structural, not a measurement error. Because NRR includes expansion, it can exceed 100% while logos fall, since a handful of growing accounts can more than replace the revenue of many departed ones. Gross revenue retention, which excludes expansion and caps at 100%, sits between the two and is the honest floor: it tells you how much revenue you kept before any upsell papered over the losses. Reading GRR, NRR, and logo retention together is the only way to see all three stories at once, which is the same layered view the expansion revenue and NRR analysis argues for.
The worked example: 60% logos, 110% revenue
Make it concrete. Start a cohort of 100 customers each paying $500 a month, for $50,000 in monthly recurring revenue. A year later, 40 have churned and 60 remain — logo retention of 60%, which on its own reads as a disaster. But those 60 survivors expanded: as one cohort guide describes the same case, they grew "from $500 to roughly $917 average MRR," so the cohort now bills about $55,000 a month — net revenue retention of 110% (CustomerScore, cohort analysis). The cohort lost nearly half its customers and grew its revenue by a tenth, simultaneously.
Which number is right? Both, and that is the point. If you tracked only logo retention you would launch a churn fire drill and miss that the cohort is net-accretive. If you tracked only NRR you would celebrate and miss that 40% of customers found no lasting value — a signal that your activation or fit for the smaller segment is broken. The two metrics are not competing for the title of the real number; they are describing two real things that happen to be true of the same cohort. The error is choosing one and discarding the other.
Which divergence should scare you
The dangerous direction is high NRR masking falling logo retention. Expansion from a few large accounts hides a leaky base of small ones, and because an NRR above 100% feels like health, it suppresses the urgency to fix the churn driving logos down. The risk compounds: as the small-account base thins, the cohort's revenue leans harder on a shrinking number of whales, until a single large-account loss collapses a number that looked robust for two years. A company "moving upmarket" often shows exactly this profile — logos down, NRR up — and mistakes a concentration risk for a growth story.
The opposite divergence — logo retention high but revenue retention below 100% — is less hidden and more fixable. You are keeping customers while they downgrade or stay under-monetized, which a pricing change, an expansion motion, or a packaging fix can address directly, and the stable logo count tells you the product itself is sticky. The asymmetry is the lesson: a strong NRR is the one that deserves suspicion, because it is the number most able to hide a problem behind an average. That averaging-away is a specific case of Simpson's paradox in aggregate metrics, where a blended figure reverses the story each segment is actually telling.
Segment the divergence before you act
The divergence is almost always a segment-mix effect, so the fix is almost always segmentation. Split logo retention and NRR by plan tier, company size, and acquisition source, and the single confusing cohort usually resolves into two clear ones: an enterprise tier with high NRR and solid logos, and an SMB tier with collapsing logos that the enterprise expansion is averaging over. Once you see that, the blended NRR stops being a health metric and starts being a warning that one segment is quietly leaving while another carries the number.
Segmentation also tells you what to do, which the blended number never does. High SMB logo churn with low SMB revenue is an activation and fit problem for that tier — possibly a signal to stop acquiring it — while strong enterprise expansion is a motion to double down on. Those are opposite actions, and you can only see them once the cohort is cut. The discipline of reading each segment on its own curve is the heart of acquisition versus behavioral cohort analysis, and it is what keeps a comfortable aggregate from hiding a departing segment.
Instrumenting both on one timeline
To see the divergence at all, you have to compute both metrics on the same cohort and the same clock. Assign every account to the cohort of its start date, and measure logo retention, GRR, and NRR at identical offsets — 3, 6, and 12 months after start — so the curves are directly comparable. Logo retention is surviving accounts over starting accounts; GRR is retained recurring revenue excluding expansion over starting recurring revenue; NRR is the same including expansion. Measured at mismatched times or on differently defined cohorts, the two curves are not comparable and the divergence is invisible.
The hard part is the data join, not the formulas. Logo churn lives in your product analytics — who stopped using the product — while revenue change lives in billing, and reconciling them on one cohort timeline is the instrumentation work that makes the divergence legible. If product and billing disagree on who is a customer and when they left, both metrics are wrong before you compare them. Getting that join right is the same reconciliation challenge covered in joining product analytics to CRM and revenue data, and it is the prerequisite for trusting either retention number.
A diagnostic you can run this week
Pull your last several start-cohorts and plot three lines for each: logo retention, GRR, and NRR at the same monthly offsets. Where NRR sits above logo retention, you have the divergence — now split those cohorts by segment and find which tier is carrying the revenue and which is bleeding logos. The pattern you are hunting for is a healthy-looking NRR propped up by expansion in one segment while another segment quietly empties out.
Then act on the segment, not the blend. If SMB logos are collapsing under a strong aggregate NRR, that is an activation or fit problem to fix or a segment to stop acquiring — not a number to keep celebrating. If logos are stable but revenue is eroding, it is a monetization problem for a sticky product. The whole exercise takes an afternoon once the cohort join is in place, and it converts a single reassuring number into a map of which part of your base is actually healthy — which is the only version of retention worth reporting.
Frequently asked questions
The questions below define logo versus revenue retention, show how revenue can grow while logos fall, explain which direction of divergence is dangerous, and cover how to instrument both metrics on one cohort timeline.
FAQ
What is the difference between logo retention and revenue retention?
Logo retention counts customers; revenue retention counts dollars. Logo retention is the share of accounts that are still customers at the end of a period regardless of what they pay, so losing a $50/month account and a $5,000/month account each count as one lost logo. Net revenue retention (NRR) measures the recurring revenue retained from a starting cohort including expansion, upgrades, and downgrades, so one expanding whale can offset many churned small accounts. Gross revenue retention (GRR) is the same without expansion credit, capped at 100%. The two families answer different questions — are customers staying, versus is the revenue base growing — and they routinely move in opposite directions.
Can revenue grow while logo retention falls?
Yes, and it is one of the most common patterns in SaaS. A cohort can lose a large share of its accounts while net revenue retention stays above 100% because the surviving customers expand enough to more than replace the churned revenue. One worked example: a cohort with 60% logo retention and 110% net revenue retention means the 60 surviving customers grew from roughly $500 to about $917 in average monthly recurring revenue each. If you tracked only logo retention you would flag the cohort as failing; if you tracked only NRR you would miss that 40% of customers found no lasting value. Both are true at once.
Which direction of divergence is dangerous?
High revenue retention masking falling logo retention is the dangerous direction, because expansion from a few large accounts hides a leaky base of small ones. NRR above 100% feels like health, so it suppresses the urgency to fix the churn driving logos down — until the base is so thin that a single large-account loss collapses the whole number. The reverse case (logo retention high, revenue retention below 100%) is less hidden: you are keeping customers but they are downgrading or being under-monetized, which a pricing or expansion motion can address. The trap is a comfortable NRR sitting on top of a logo-retention problem nobody is staffing.
Should SaaS companies track logo retention or revenue retention?
Both, always, and segmented. Tracking only one hides the exact failure mode the other would catch: NRR alone hides logo churn behind expansion, and logo retention alone hides revenue erosion behind a stable customer count. The strongest setup reports GRR, NRR, and logo retention side by side for each cohort, then splits them by segment — because the divergence is almost always a segment-mix effect, where an expanding enterprise tier is averaging over a churning SMB tier. Reading the blended numbers alone is how teams celebrate a healthy NRR while a whole customer segment quietly leaves.
How do I instrument logo and revenue retention correctly?
Fix the cohort and the clock first: assign every account to the cohort of its start date and measure both metrics at the same offsets (for example 3, 6, and 12 months after start) so the two curves are comparable. Compute logo retention as surviving accounts over starting accounts, GRR as retained recurring revenue excluding expansion over starting recurring revenue, and NRR as the same including expansion. Then break every number out by segment, plan, and acquisition source. The instrumentation that matters is the join between your product analytics and billing data, because logo churn lives in the product and revenue change lives in billing — reconciling them on one cohort timeline is what makes the divergence visible.