Early-Stage SaaS Conversion
Review retention, growth, and first-revenue evidence from prelaunch to $1 million ARR, and identify what evidence can and cannot justify discounting mature-company benchmarks.
Written and reviewed by Eric Han · Co-founder & CTO, reopt
Key takeaways
- We found no public free-to-paid or trial-to-paid rates by ARR band. Multiplying mature-company benchmarks by a discount such as 0.5 has no evidential basis.
- Retention does have band-level data: median annual customer retention is 53% below $300K ARR, 65% at $300K–1M, and 68% at $1–3M (2023, only companies surviving 12 months).
- Surveys and billing data diverge sharply. Median GRR below $1M is 90% in a survey versus 53–62% in billing data. Do not combine them.
- Dispersion matters more than the early-stage median. Below $300K, annual growth through March 2023 was 663% at the top-10% threshold, 37% at the median, and −5% at the bottom quartile.
- Soon after launch, the evidence supports examining time to first revenue separately, rather than applying a discount to conversion rates. Of Stripe Atlas startups, 80% had no payment within 30 days of incorporation (calculated).
What the Evidence Can Establish
| Question | Answer | Evidence |
|---|---|---|
| How much lower is early-product free-to-paid conversion? | Unknown | No public stage-level values |
| How much lower is early-product retention? | Versus $1–3M, below $300K is about 22–27% lower; $300K–1M about 4–9% lower | Cross-sectional comparison within one source, calculated |
| Is median growth lower? | It is higher, with much wider dispersion | Billing data |
| How long to first revenue? | Only part of the distribution is public | Stripe Atlas, not SaaS-only |
Follow the Self-Serve SaaS principle of planning around outcomes near the lower end of the interval without attaching an unsupported numerical discount.
Retention and Churn by ARR
ChartMogul SaaS Benchmarks 2023 uses billing data from more than 2,100 companies over the 12 months through March 2023, including only those active throughout the period (vendor customers, grade B). Excluding closures biases results upward. Conversely, the report describes the period as especially difficult for retention, pushing values downward. The magnitudes of these biases are unknown.
- ARR below $300K53%
- ARR $300K–1M65%
- ARR $1–3M68%
- ARR $15–30M66%
| Metric, median | Below $300K | $300K–1M | $1–3M |
|---|---|---|---|
| Annual gross revenue retention, GRR | 53% | 62% | 68% |
| Annual net revenue retention, NRR | 56% | 70% | 77% |
| Monthly customer churn | 6.5% | 4.1% | 3.7% |
| Monthly net MRR churn | 6.2% | 3.1% | 2.3% |
Ratios relative to $1–3M are below (calculated). They do not isolate customer-mix differences and are not causal.
| Metric | Below $300K | $300K–1M |
|---|---|---|
| Annual customer retention | About 0.78× | About 0.96× |
| GRR | About 0.78× | About 0.91× |
| NRR | About 0.73× | About 0.91× |
| Monthly customer churn | About 1.8× | About 1.1× |
| Monthly net MRR churn | About 2.7× | About 1.3× |
Do not mix surveys with billing data
Below $1M, median GRR is 90% in the Benchmarkit 2025 survey versus 53–62% in ChartMogul. Benchmarkit itself notes that immature measurement below $5M may overstate retention. Timing and B2C inclusion also differ.
Growth Varies Widely at the Earliest Stages
| ARR band | Top 10% threshold | Top 25% threshold | Median | Bottom 25% threshold |
|---|---|---|---|---|
| Below $300K | 663% | 153% | 37% | −5% |
| $300K–1M | 224% | 97% | 32% | 4% |
| $1–3M | 183% | 70% | 27% | 7% |
These are annual ARR growth rates from the same ChartMogul dataset. The gap between the top-decile threshold and bottom quartile is 668 percentage points below $300K, versus 176 points at $1–3M (calculated). Use scenarios rather than a point estimate.
- ChartMogul Growth 2023: median monthly growth stays around 2–2.5% across the lifecycle; the top 10% grow 10–17% monthly early on. Median time from first paying customer to $1M ARR is about two years and nine months, conditional on reaching it. After ten years, 13% reach $10M among ChartMogul customers; the report acknowledges survivorship bias.
- High Alpha / Growth Unhinged 2025: self-reported median growth below $1M is 75% for B2B and 100% for AI-native companies. The 2025 survey of over 800 companies is higher than 2023 billing data; survey bias cannot be separated from period effects (C).
- Paul Graham's 5–7% weekly growth is an operating target during YC batches, not a measured benchmark.
Time to First Revenue
| Metric | Value | Conditions | Evidence |
|---|---|---|---|
| First payment within 30 days of incorporation | 20%, versus 8% in 2020 | About 23,000 Atlas startups, not SaaS-only | Stripe Atlas 2025, B |
| Median time to first payment | 34 days | Only companies paid within three months | Same source, B |
| Products with zero revenue | More than 54% | 937 Stripe-verified Indie Hackers products, 2022 | DEV analysis, D |
| Reaching $100K annually | About 5% | Same sample | Same source, D |
| Median monthly revenue | $169 | 5,079 voluntarily listed products, March 2026 | Indie Hackers post, D |
Voluntary disclosure does not have a single predictable bias direction: success stories are easier to publish, but zero-revenue projects also remain listed. The supported conclusion is that having no paying customers yet is common, rather than a specific low conversion rate.
Prelaunch Funnels
Apart from the two Unbounce rows (C), these are anecdotes or secondary citations. Check the numerator; do not use them as baselines.
| Metric | Value | Denominator / context | Grade |
|---|---|---|---|
| SaaS landing-page conversion | Median 3.8%, versus 6.6% across industries | SaaS subset, size undisclosed, of more than 41,000 pages; July 2023–July 2024. Demos, trials, and downloads mixed. Pages with fewer than 50 visits or zero conversions excluded, optimistic for new products | C |
| By landing-page traffic source | Email 16.9%, paid search 4.1%, paid social 2.9%, display 0.3% | Same source | C |
| Waitlist → paid | 5–25% | Secondary citation via a16z; original sample unclear | D |
| All launch-day visits → free signups on Product Hunt | 10.73% | Ranked fourth; 1,240 visits including other sources; free community | D |
| First Hacker News visit → signup | About 0.39%, calculated | 37signals, mature company, 2012 | D |
| Prospective design partners → paid | 6 of 40, or 15%, calculated | Sierra, exceptional founder-network advantages | D |
Product Hunt's 10.73% and Hacker News's 0.39% differ in product and denominator and must not be combined into a range. We found no sample following these users through paid conversion, no design-partner-to-paid baseline, and no seed-stage cold-outreach win rate. Sales-Led Conversion offers the best available outbound inputs; no evidence supports a premium for founder-sent outreach.
The 40% PMF Rule and Failure-Rate Myths
Sean Ellis's 40% threshold is relayed through Rahul Vohra's account of “nearly 100” companies, without public underlying data. At 40 respondents and a 40% response share, a 95% Wilson interval is about 26–55% (calculated). Superhuman increased the score from 22% to 32–33% by narrowing the segment, so compare scores carefully.
Do not use “90% of startups fail.” It is a claim in the Startup Genome introduction, not a cohort measurement. Official business-survival statistics are not early SaaS benchmarks. Five-year survival was 50.6% for US establishments born in 2013 (B), 36.5% for Korean new enterprises (B via press), and 59.3% for selected Korean pre-startup support recipients (B via press). Definitions differ and prevent direct comparison. Carta's seed-to-Series-A rate was about 25–30% for 2018 and 17% for 2022: a funding-progression measure (B).
We found no public Korean early-SaaS conversion or churn data. Nearby figures appear in Korean Cases.
Simulation Inputs
| Stage | Input | Evidence and conditions |
|---|---|---|
| Visit → signup | Mean 4.5%, | ChartMogul 2026 trial products, self-reported survey C; 34% of sample below $1M ARR |
| Signup → paid | Mean 8%, | Overall median in the same source; no evidence for an early-stage discount |
| 12-month retention | Mean 53%, | ChartMogul 2023 median below $300K ARR, B |
| First revenue | Consider separately from this calculation | Stripe Atlas: 20% paid within 30 days of incorporation; not SaaS-only |
Do not use Benchmarkit's 90% GRR, waitlist-to-paid 5–25%, or Product Hunt's 10.73% as inputs.
Sample size itself is the first limitation: eight conversions from 100 trials give a 95% Wilson interval of about 4.1–15.0% (calculated). This motivates low concentration .
Early SaaS paying-customer intervals
An example using self-reported survey rates with low concentration. Means remain unchanged, but skew makes the displayed median about half the 7.2-customer point estimate.
Each input alone moves to its 5th and 95th percentile with others at the mean. Measure the top one first.
- Visit → signup0.5customers ~ 20.6customers
- Signup → paid0.7customers ~ 19.3customers
8,000 runs. Stages are independent. If they move together the real interval is wider; if they move in opposite directions it is narrower. The 90% interval holds only rate uncertainty around the expected value; the actual count adds chance variation in counts. Vertical lines mark the 5th percentile, median (darker), and 95th. The last bin collects values above the 99th percentile. Concentration k is an author judgment.
Self-Serve SaaS Conversion
Compare freemium, trials, card requirements, and activation using vendor surveys, transaction logs, and randomized experiments to establish SaaS forecasting baselines.
Sales-Led Conversion
Review lead-to-opportunity-to-win rates, lead-source differences, and email, phone, and LinkedIn outreach metrics while excluding figures with unclear origins.