Research Brief | Retention · Growth · Performance | March 3, 2026

CORE THESIS Retention becomes a growth constraint when the existing customer base loses revenue faster than new business can economically replace it. NRR shows whether the installed base compounds after expansion; GRR shows how much revenue survives before expansion. Read together, they reveal whether growth is being amplified by customer value or financed by a recurring replacement burden.

 

Executive Summary

Retention is often summarized into one number, usually NRR. That is useful, but incomplete. NRR can be above 100% while a meaningful share of customers or contracted revenue is still being lost, because expansion from surviving accounts offsets churn and contraction. GRR removes expansion and therefore exposes the durability of the opening revenue base.

For the founder-led B2B SaaS companies closest to Woldmark’s initial focus, SaaS Capital’s 2026 data provides a useful reference point: bootstrapped companies with $3M–$20M ARR reported median NRR of 103% and median GRR of 91%, while the 90th percentile reached 117.9% NRR and 100% GRR. [1]

The broader market is not uniformly improving. Benchmarkit’s 2026 benchmark release reported market-wide GRR falling from 88% to 84%, including a decline at the 75th percentile from 95% to 91%. [2]

At the same time, stronger retention remains associated with stronger growth. SaaS Capital’s 2026 survey of more than 1,000 private B2B SaaS companies found that moving from the 90%–100% NRR range to 100%–110% was associated with a five-percentage-point increase in growth, and the highest-NRR cohort reported median growth 173% above the population median. [3]

The operating conclusion is not that a single retention threshold determines company quality. It is that retention sets the amount of new revenue a company must create merely to stand still — and NRR alone can conceal the source of that burden.

1. NRR and GRR answer different questions

ChartMogul defines NRR as the percentage of starting recurring revenue retained after expansion, contraction, and churn, while GRR excludes expansion. New customers acquired during the measurement period are excluded from both calculations. [4]

That distinction matters because the two metrics answer different management questions:

• NRR: Does the opening customer base generate more or less recurring revenue after churn, contraction, and expansion?
• GRR: How much of the opening recurring revenue survives before expansion is allowed to compensate for losses?

2. Why NRR can hide customer loss

Consider a simple annual cohort. A company begins with $10 million of recurring revenue. During the year it loses $2 million to churn and contraction but adds $2.5 million of expansion from surviving customers. GRR is 80%; NRR is 105%.

Both numbers are correct. But they describe different realities. The company has an installed base that expands in aggregate, yet one fifth of the opening revenue disappeared before expansion. If that loss is concentrated in a weak segment, a product gap, a cohort, or a customer-size band, the aggregate NRR can make the problem look smaller than it is.

This is why a high NRR should not be interpreted as proof of uniformly strong retention. It can coexist with meaningful gross loss, particularly when expansion is concentrated among a subset of large or successful accounts.

Opening ARR	Churn + contraction	Expansion	Result
$10.0M	-$2.0M	+$2.5M	GRR 80% · NRR 105%

3. The retention ceiling is a replacement burden

There is a simple way to see when retention becomes a binding constraint on growth. If opening ARR is normalized to 100, annual target growth is g, and NRR is r, then required new-logo ARR is approximately:

Required new ARR = 100 × (1 + target growth − NRR)

This is arithmetic, not a benchmark. It shows the burden placed on acquisition by retention.

NRR	Target growth	New ARR required vs. opening ARR
90%	25%	35%
100%	25%	25%
105%	25%	20%
115%	25%	10%

At 90% NRR, a company targeting 25% annual growth must generate new ARR equal to 35% of its opening ARR just to reach the target. At 115% NRR, the required new ARR falls to 10%. The difference is not cosmetic: it changes sales capacity requirements, CAC exposure, hiring pressure, and the amount of growth that must be purchased from outside the installed base.

4. Retention becomes more important as new business gets harder

ChartMogul’s retention research, based on more than 2,500 SaaS businesses, found that in 2024 companies with at least 100% NRR grew at a median 48% year over year — more than twice as fast as companies below 100% NRR. It also found that companies with high NRR derived more than half of their growth from expansion, while low-NRR companies depended far more heavily on new business. [5]

The same report observed that expansion represented 40% of growth for companies with roughly $15M–$30M+ ARR in 2024, up from about 30% in early 2021. [5]

For smaller founder-led companies, the exact mix will differ. The important point is structural: as acquisition becomes slower, more expensive, or less predictable, weak retention raises the amount of external growth the company must continually recreate.

5. A good NRR can still be fragile

Three common patterns can make aggregate NRR look healthier than the underlying customer base:

• Expansion concentration. A small group of large accounts expands enough to offset broad contraction or churn elsewhere.
• Segment mixing. Enterprise customers may retain differently from SMB customers; a single company-wide NRR can hide a deteriorating segment.
• Pricing mechanics. Usage, seat, hybrid, and other models produce different expansion and contraction behavior, so identical NRR values can arise from different economics.

Benchmarkit’s 2026 data is a useful reminder on the pricing point: the report shows median NRR of 108% for usage-based models versus 98% for seat-based models. [2]

That does not mean usage pricing is universally superior. It means pricing architecture changes the mechanics of retention. A founder should therefore avoid comparing NRR across materially different models as if the metric were context-free.

6. GRR identifies the part expansion cannot repair

GRR is deliberately unforgiving: expansion does not count. That makes it useful for identifying whether the existing revenue base is intrinsically durable.

In SaaS Capital’s 2026 cohort of bootstrapped companies with $3M–$20M ARR, median GRR was 91%, and the 90th percentile was 100%. [1]

High Alpha’s 2025 benchmark report also found materially stronger median growth among companies with high GRR and high NRR than among weaker-retention cohorts, while noting that the relationship is correlational rather than deterministic. [6]

The practical value of GRR is not that it should replace NRR. The two should be read as a pair: NRR captures the compounding behavior of the installed base; GRR shows the loss that expansion is covering.

7. What founders should inspect below the headline retention rate

• Retention by cohort. Are newer cohorts retaining better or worse than older ones?
• Retention by customer size. Is the aggregate supported by a few larger accounts while smaller customers decay?
• GRR versus NRR spread. Is expansion amplifying a sticky base or compensating for large gross losses?
• Expansion concentration. How much expansion comes from the top accounts, and is that concentration increasing?
• Churn versus contraction. Are customers leaving entirely, or remaining while reducing spend?
• Renewal versus usage behavior. Does reported retention lag a visible decline in product usage, seat count, transactions, or engagement?
• Retention by product and pricing model. Are changes caused by product value, customer mix, packaging, or the revenue model itself?

The useful question is not “Is our NRR above 100%?” It is “How much of our opening revenue survives without expansion, where are the losses occurring, and how much acquisition effort is required to compensate for them?”

Evidence Notes

This brief combines survey benchmarks and aggregated revenue data from several SaaS research providers. Samples, company sizes, pricing models, time periods, and metric definitions differ. Source-specific figures are therefore not blended into a single universal benchmark. Reported relationships between retention and growth are treated as associations, not proof of causation. The “retention ceiling” calculation is a Woldmark analytical illustration derived directly from the arithmetic of NRR and target growth.

Sources & References

1. SaaS Capital — 2026 Benchmarking Metrics for Bootstrapped SaaS Companies — Source
2. Benchmarkit — 2026 B2B SaaS & AI-Native Metrics — Source
3. SaaS Capital — 2026 Private B2B SaaS Company Growth Rate Benchmarks — Source
4. ChartMogul — Benchmarks: NRR and GRR Definitions — Source
5. ChartMogul — The New Normal for SaaS: Retention Report — Source
6. High Alpha — 2025 SaaS Benchmarks Report — Source

About Woldmark

Woldmark is an independent intelligence firm for founder-led companies. We publish research and analysis and provide recurring independent business performance reviews focused on material change, key assumptions, performance interpretation, and emerging risk. woldmark.com · vitaly@woldmark.com