Gross Retention vs Net Retention in SaaS IPOs

The most important retention number in a SaaS IPO is not the one the founder puts on the slide. It is the one the filing leaves out.

Gross retention vs net retention in SaaS IPOs is the single comparison that separates a durable business from a flattering one. Net revenue retention is the headline founders love, because expansion revenue hides churn inside a number above 100%. Gross retention is the number that cannot lie, because it strips out every upsell and shows how much revenue actually stays.

Read together, they tell two different stories about the same company. NRR is the growth story. GRR is the foundation under it. When a prospectus shows a glowing NRR and goes quiet on GRR, the silence is the signal.

This piece reads that gap through public filings, not pitch decks. Every company figure below ties to a specific 10-K, 10-Q, or earnings release and fiscal period. The framing is analytical: how to read the two metrics against each other, not what to do about any stock.

Key takeaways

• NRR includes expansion; GRR does not. Net revenue retention counts upsells and upgrades; gross revenue retention strips them out and shows only what survives churn and downgrades. The gap between them is the expansion engine.
• Disclosure is lopsided. Roughly 5% of public SaaS companies report a gross retention metric while about 67% report net retention (Ordway Labs SaaS retention analysis). GRR is the number more often withheld.
• Datadog shows both. Trailing-12-month net retention was about 120% with gross dollar retention in the mid-to-high 90s as of Q2 2025 (Datadog Form 10-Q, quarter ended June 30, 2025). That pairing is a strong foundation plus a real expansion engine.
• NRR decays. Snowflake’s net revenue retention was 126% in FY2025, down from a 177% peak in FY2022 (Snowflake Form 10-K, FY2025). A high NRR is a snapshot, not a constant.
• Identical NRR can hide very different GRR. Two companies at 120% NRR can sit at 95% versus 80% GRR. Same headline, opposite durability.
• The gap is the diligence question. NRR minus GRR is the expansion lift. A wide gap on an undisclosed GRR is the tell an underwriter looks for.

What is the difference between gross retention and net retention in SaaS?

Gross retention (GRR) measures how much recurring revenue you keep from existing customers after churn and downgrades, with no credit for upsells. Net retention (NRR) counts that same expansion in the calculation. A company can post high NRR but lower GRR, meaning the growth leans on selling more to the base rather than on retention itself.

Net revenue retention (NRR) measures recurring revenue from a fixed cohort of existing customers one year later, including expansion from upsells and upgrades, net of downgrades and churn. It answers a single question: did this group of customers, as a group, spend more or less than a year ago?

Above 100% means the cohort grew without a single new logo. That is the property founders love, and rightly so. A business that expands inside its installed base can grow even when new-customer acquisition slows. It is also the cleanest evidence of product value: customers who use more and pay more are voting with their budgets.

The mechanics are straightforward. Take the annual recurring revenue of a customer cohort at the start of the period. Add expansion, subtract downgrades and churned revenue, and divide by the starting figure. The same expansion-driven logic governs whether a company should price by seat or by consumption, the tradeoff dissected in usage-based pricing vs seat-based pricing.

The reason founders lead with NRR is not deception. It is selection. NRR is the metric that most flatters a healthy expansion motion, and a prospectus is a sales document. The job of the reader is to notice what the chosen metric leaves out.

Gross retention: the revenue that never left

Gross revenue retention (GRR) measures the same starting cohort one year later, but counts only what stayed, with no credit for expansion. It subtracts churn and downgrades from the base and stops there. GRR can never exceed 100%.

That ceiling is the whole point. NRR can paper over a leaky base because one whale’s expansion offsets ten small accounts walking out the door. GRR cannot. It exposes the raw rate at which a company holds the revenue it already had, before any sales rep upsells a single seat.

Think of two vessels. NRR is the water level after you both fill from the top and let some leak from the bottom. GRR is the leak rate alone. A high water level tells you nothing about how fast the bottom is draining if the tap is running hard enough to mask it.

Healthy GRR sits in a recognizable band. Industry benchmarks place 85% to 95% in the healthy range, with the strongest businesses above 95% (ChartMogul and MetricHQ retention benchmark resources). Below 85%, a company is losing more than a seventh of its gross revenue every year to churn and downgrades, a hole that expansion has to refill before the business grows at all.

Why do most SaaS companies disclose net retention but not gross retention?

Net retention tells a better growth story because it includes expansion revenue. Gross retention strips out upsells and shows how much revenue naturally stays, and in a slowing market that number can be weak. Only about 5% of public SaaS companies disclose a gross retention metric, which points to intentional opacity around baseline churn.

Because GRR has a ceiling and a worse story, most companies simply do not report it. Across 135 analyzed public SaaS companies, about 5% report a gross retention metric while roughly 67% report a net retention metric (Ordway Labs public-company SaaS disclosure analysis). Among IPO-era filers, the imbalance shows up in the S-1: of 41 SaaS companies that disclosed net dollar retention, only 5 or 6 also disclosed gross dollar retention (Blossom Street Ventures and Ordway Labs research on public disclosures).

This is the opacity problem, and it is not an accident of accounting. Neither metric is mandated by GAAP, so disclosure of both is a choice. A company that holds 95% of its revenue and expands the rest to 120% has every reason to show GRR, because the foundation is strong. A company that holds 80% and leans on a few large expansions to reach the same 120% has every reason to stay quiet.

The pattern of disclosing the flattering metric and withholding the unflattering one is familiar to anyone who reads prospectuses closely, a discipline laid out in how to read a tech S-1 like an operator. The absence of a metric in a filing is itself information. When a company reports NRR to two decimal places and never once names GRR, the reader should treat the omission as the most interesting line on the page.

How do you read gross retention vs net retention in SaaS IPOs together?

Compute the expansion ratio: NRR minus GRR equals the points of growth that come from selling more to existing customers rather than keeping them. A 120% NRR on a 95% GRR is a strong base plus a bonus; the same 120% on an 80% GRR is a leaky base masked by expansion. If a filing hides GRR, treat the omission as the signal.

The expansion ratio is the analytical key: NRR minus GRR equals the points of growth that come from selling more to existing customers rather than from keeping them. It isolates the expansion engine from the retention base.

A company at 120% NRR and 95% GRR has a 25-point expansion lift on a near-watertight base. A company at 120% NRR and 80% GRR has a 40-point expansion lift bolted onto a base losing a fifth of its revenue annually. Identical headline. Opposite resilience.

The gap matters most when growth conditions change. Expansion is the most cyclical component of retention: it is the first thing to compress when customers freeze budgets, consolidate vendors, or cut seats in a downturn. A business that depends on a wide expansion ratio to clear 100% NRR is borrowing its retention story from a sales motion that softens exactly when it is needed most.

the Retention Decoder

This is the framework this piece contributes: the Retention Decoder, a matrix that defines NRR and GRR side by side, states what each includes and excludes, gives the healthy band, and reads what the gap between them reveals. It is an original analytical asset; the benchmark bands draw on the public retention research cited above. Name it so it is referenceable: you can run any SaaS filing through these five rows and know within minutes what the company is and is not telling you.

Dimension	Net Revenue Retention (NRR)	Gross Revenue Retention (GRR)
Includes	Upsells, upgrades, cross-sells, price increases	Nothing additive; base revenue only
Excludes	Nothing; it is the all-in figure	All expansion revenue
Subtracts	Churn and downgrades	Churn and downgrades
Can exceed 100%?	Yes, expansion lifts it above	No, capped at 100% by definition
Healthy band	110%+ enterprise, 100%+ broadly	85% to 95%, top tier 95%+
What it reveals	The growth story and expansion engine	The foundation and true churn rate
The gap (NRR − GRR)	Expansion lift in points	The cushion that disappears in a downturn

Run a filing through the Decoder and one column usually goes missing. The discipline is to fill it in from context, label it as an estimate, and never let a strong NRR stand in for a GRR the company declined to show. The same read-both-numbers logic underpins why gross margin is destiny in SaaS: a single headline figure rarely survives contact with its underlying components.

Case study: Datadog discloses both, and the pairing tells the story

Datadog is the rare large-cap that publishes both numbers, which makes it the cleanest teaching example. As of Q2 2025, trailing-12-month dollar-based net retention was approximately 120%, and trailing-12-month gross dollar retention remained stable in the mid-to-high 90s (Datadog Form 10-Q, quarter ended June 30, 2025; Datadog Supplemental Financial Information, Q2 2025).

Datadog (Q2 2025, TTM)	Value	Source
Net dollar retention	~120%	Datadog Form 10-Q, quarter ended June 30, 2025
Gross dollar retention	Mid-to-high 90s	Datadog Supplemental Financial Information, Q2 2025
Implied expansion ratio (NRR − GRR)	~20 to 25 points	Derived from the two disclosed figures

Read through the Decoder, this is a strong business. The mid-to-high-90s GRR means the base is nearly watertight; Datadog keeps almost every dollar it already had. The ~120% NRR then adds a 20-to-25-point expansion lift on top of that foundation, not in place of it. The expansion is a bonus, not a crutch.

The reason Datadog can show GRR is that GRR is a strength. A usage-based observability platform that customers expand as their infrastructure grows tends to hold its base tightly, because ripping out monitoring is operationally painful. The disclosure choice and the business model point the same direction.

Case study: Snowflake’s NRR decay and the absent GRR

Snowflake illustrates the opposite disclosure posture and a second lesson: NRR decays. Net revenue retention was 126% in FY2025, down from a peak of 177% in FY2022 (Snowflake Form 10-K, FY2025). The company does not disclose a gross dollar retention figure.

Snowflake net revenue retention	Value	Source
FY2022 (peak)	177%	Snowflake Form 10-K, FY2025
FY2025	126%	Snowflake Form 10-K, FY2025
Gross dollar retention	Not disclosed	Snowflake Form 10-K, FY2025

Two things stand out. First, a 177% NRR was never a steady state; it reflected a period of explosive consumption growth that has normalized as the customer base matured. A net retention figure is a snapshot of a moment in the growth curve, not a fixed property of the company. Reading any single year’s NRR as permanent is the error the decay rate corrects.

Second, the missing GRR leaves a hole in the Decoder. At 126% NRR with an undisclosed gross figure, the reader cannot tell whether the foundation is a watertight mid-90s or a leakier low-80s offset by heavy expansion from large accounts. For a consumption-priced platform, that distinction matters, because consumption can fall as fast as it rises. The dependence on a concentrated set of large, expanding accounts is exactly the exposure mapped in customer concentration risk in SaaS filings.

None of this says the business is weak. It says the filing shows the flattering metric and withholds the one that would let an outsider judge the base. That is the opacity problem in a single 10-K.

Case study: HubSpot’s high-80s GRR paired with low-100s NRR

HubSpot anchors the third pattern: a company serving a high-volume, SMB-weighted base, where gross retention runs lower by structure. HubSpot disclosed customer dollar retention (its gross measure) in the high-80s in Q3 2025, and reported net revenue retention of 103% in Q1 2026 (HubSpot Q3 2025 earnings; HubSpot Q1 2026 earnings results).

HubSpot retention	Value	Source
Customer dollar retention (gross)	High-80s	HubSpot Q3 2025 earnings, September 2025
Net revenue retention	103%	HubSpot Q1 2026 earnings results
Implied expansion ratio (NRR − GRR)	~15 points	Derived from the two disclosed figures

Through the Decoder, this is a different shape of healthy. A high-80s GRR is solidly in the benchmark band but below an enterprise platform’s mid-90s, which is what a large base of smaller customers tends to produce: SMBs churn more often because they go out of business, change tools, or cut costs faster than enterprises. The 103% NRR then sits only modestly above 100%, meaning expansion covers the higher churn and leaves a thin margin of net growth from the existing base.

That is not a flaw. It is the math of HubSpot’s market. The instructive part is that HubSpot discloses both, so the read is honest: you can see that net growth from the base is real but slim, and that new-logo acquisition has to do more of the work than it does at a mid-90s-GRR enterprise vendor.

A clearly labeled illustrative example: identical NRR, opposite GRR

To make the gap concrete, here is an illustrative, hypothetical comparison. These numbers are invented to show the mechanism, not drawn from any filing.

Two companies each report 120% net revenue retention. On the headline metric, they look identical. The Decoder pulls them apart.

Illustrative (hypothetical)	Company A	Company B
Net revenue retention	120%	120%
Gross revenue retention	95%	80%
Expansion ratio (NRR − GRR)	25 points	40 points
Annual gross revenue lost to churn	5%	20%
Foundation	Near-watertight	Leaky, masked by expansion

Company A keeps 95% of its base and adds 25 points of expansion. Company B keeps only 80% and has to manufacture 40 points of expansion just to reach the same 120%. In a good year, both print 120% and look the same to a casual reader.

Now imagine expansion compresses by half in a downturn, a realistic move when customers freeze budgets. Company A falls to roughly 95% + 12.5% = 107.5% NRR, still growing its base. Company B falls to roughly 80% + 20% = 100% NRR, now treading water, with a 20% gross leak that expansion no longer covers. Same starting headline. The GRR decided which one survives the cycle.

What operators should take from this

The transferable lesson is not “GRR good, NRR bad.” It is that you read them together or you read neither. Here is the playbook, concrete moves for a founder, operator, or analyst reading filings.

1. Compute the expansion ratio first. Subtract GRR from NRR before anything else. That single number tells you how much of the growth story is retention and how much is upsell. A 10-to-15-point gap is a balanced business; a 35-plus-point gap is an expansion bet, and you should price the cyclicality of that expansion into your read.
2. Treat a missing GRR as a red flag, not a neutral omission. If a company discloses NRR and withholds GRR, ask why. In private diligence, request it directly. In public filings, estimate it from cohort churn disclosures and label the estimate. Never let a strong NRR stand in for the GRR a company declined to show.
3. Read NRR as a curve, not a point. Snowflake’s slide from 177% to 126% (Snowflake Form 10-K, FY2025) shows net retention decays toward a steady state as a base matures. Pull three to four years of NRR and look at the trend; a single high year is the least informative version of the metric.
4. Match GRR to the customer base. A high-80s GRR is healthy for an SMB-weighted business like HubSpot and worrying for an enterprise platform that should sit in the mid-90s. Benchmark GRR against the company’s segment, not against an absolute number.
5. Cross-check the expansion against unit economics. A wide expansion ratio is only durable if the expansion carries margin and the base is cheap to retain. Read it alongside gross margin and CAC payback; the same composite discipline drives the Rule of 40 and what it actually tells you. High expansion on thin margin is not the same quality as high expansion on fat margin.
6. Watch the leading indicator. For any expansion-heavy business, the number that moves first in a downturn is the expansion rate, which shows up as NRR compressing toward GRR. Put the gap on a dashboard and track whether it is widening or narrowing quarter over quarter.

The discipline scales down to a one-person SaaS. Running retention on PDF9to5, the difference between a base that holds at 90% and one that holds at 75% but looks fine on NRR because a few power users keep upgrading is the difference between a durable product and a treadmill. The headline can hide the leak. The expansion ratio cannot.

The bear case: where the retention read breaks down

A framework this clean deserves its own counterexamples. The Decoder sharpens the read; it does not settle every case.

GRR definitions are not standardized. Because neither metric is GAAP-mandated, companies define the cohort, the period, and what counts as a downgrade differently. One company’s “gross dollar retention” and another’s “customer dollar retention” may not be computed the same way. HubSpot’s customer dollar retention and Datadog’s gross dollar retention are both gross measures, but comparing them point-for-point overstates the precision. Compare the definitions before comparing the numbers.

A low GRR can be a deliberate market choice. A company serving small businesses or running a freemium-to-paid motion will structurally churn more than an enterprise vendor, and a high-80s GRR may be the best achievable in that segment. Read against the wrong benchmark, that looks like a broken business when it is simply a different one. The same disclosure honesty that makes HubSpot easy to read also makes it easy to misjudge against an enterprise yardstick.

Expansion can be high-quality. A wide NRR-GRR gap is not automatically fragile. If expansion comes from customers adopting more of a platform they are operationally locked into, it can be as durable as the base itself, the kind of stickiness examined in Microsoft Copilot and enterprise lock-in. The gap flags a question to investigate, not a verdict to pronounce.

A snapshot misses the trajectory. A single year’s GRR, like a single year’s NRR, can mislead. A business improving its base from 82% to 88% over three years is a very different read from one sliding the other way, even if both show 85% today. Read the direction, not just the level.

None of this overturns the thesis. It bounds it. Reading GRR and NRR together is the strongest single way to judge retention quality, and it is not the only thing that decides durability.

Methodology: how to score the retention gap

When you compute an expansion ratio and draw a conclusion about durability, here is the frame to keep it honest.

• Inputs: disclosed NRR and GRR (or customer dollar retention) from the most recent filing, by company and fiscal period. Datadog ~120% NRR and mid-to-high-90s GRR (Form 10-Q, quarter ended June 30, 2025); Snowflake 126% NRR, GRR undisclosed (Form 10-K, FY2025); HubSpot high-80s gross and 103% net (Q3 2025 and Q1 2026 earnings).
• Calculation: expansion ratio = NRR − GRR, expressed in percentage points. Where GRR is undisclosed, mark the ratio as not computable rather than estimating without a basis.
• Assumptions: that each company’s cohort definition and measurement window are consistent year over year, and that the disclosed figures use comparable bases. Cross-company comparison assumes definitions align, which they often do not.
• Sensitivity: halve the expansion rate to simulate a downturn (the example above), and re-read NRR. A business that drops below 100% under that stress depends on expansion for its growth story; one that stays above relies on a durable base.
• What this misses: public filings rarely break retention by segment or cohort vintage, so a blended figure can mask a strong enterprise base subsidizing a leaky SMB one, or the reverse. Retention also says nothing about gross margin or acquisition cost, both of which can rescue or sink a business independent of its retention.

This is a framework for understanding retention quality, not a model that outputs a target price.

How the pieces fit together

Gross retention vs net retention in SaaS IPOs comes down to one habit: read both, and notice which one is missing.

NRR is the growth story, and it is a real one when the foundation is sound. GRR is that foundation, the rate at which a business holds what it already earned before a single upsell. The gap between them is the expansion engine, and the durability of that engine is the question a careful reader actually wants answered.

When a filing shows a 120% NRR and never names GRR, the analytical move is not to admire the headline. It is to fill in the missing column, label the estimate, and weigh how much of the story depends on expansion that a downturn can erase. The metric a company chooses to show is a sales decision. The metric it chooses to hide is the one worth reading.

That is the whole comparison. NRR shows you the height of the water. GRR tells you how fast it drains.

---

Analysis, not investment advice. Figures are drawn from public SEC filings and earnings releases cited inline by company and fiscal period (Datadog Form 10-Q Q2 2025; Snowflake Form 10-K FY2025; Okta Form 10-K FY2025; HubSpot Q3 2025 and Q1 2026 earnings; MongoDB Form 10-K FY2025) and from public retention research by Ordway Labs and Blossom Street Ventures. Frameworks here are for understanding SaaS business models and tradeoffs, not for making buy or sell decisions.

Want the full toolkit for reading filings like this, the retention-decoder matrix, the expansion-ratio worksheet, and the cohort-durability checklist used above? It’s in the Tech Business Analysis Playbook.