Retention Curve Benchmarks 2026: SaaS, Consumer, Marketplace, Fintech
What are 2026 retention curve benchmarks by vertical — SaaS, consumer, marketplace, fintech?
Summary
Retention curves are the single most predictive signal of long-term business value, but "good retention" is meaningless without a vertical. This card consolidates 2026 cohort retention benchmarks across B2B SaaS (SMB, mid-market, enterprise), consumer mobile (gaming, fintech, ecommerce, social, health), consumer subscription (streaming, media), marketplaces (buyer and seller side), and fintech (neobanks). Each vertical uses different time horizons and units — SaaS measures dollar retention over 12 months; mobile measures user activity at D1/D7/D30; marketplaces measure GMV retention by cohort month. [src1, src2, src4]
Data vintage: H2 2025 through Q1 2026. Aggregated from 8 primary benchmark reports covering ~100K+ apps, ~2,500 subscription businesses, and ~940 B2B SaaS companies.
Key shift: SMB SaaS NRR crossed below 100% at median for the first time since 2019 as expansion stalled [src1]. Consumer subscription monthly churn rose to 5.5% (from 2% in 2019) driven by streaming fatigue [src6]. Mobile D30 retention compressed 10–15% across ecommerce, health, and education categories as CAC-driven installs saturated [src2].
Constraints
- Vertical specificity: A 5% D30 retention is excellent for ecommerce and disastrous for fintech. Never transplant a curve shape or threshold across verticals.
- Methodology — three incompatible units: Consumer apps use n-day retention. SaaS uses dollar retention (NRR/GRR) for 12-month cohorts. Marketplaces use GMV retention by cohort month. Do not compare raw numbers across these units.
- Median vs mean: All figures median unless labeled. Means are dominated by outliers. For target-setting, use 75th percentile; for sanity checks, use median.
- Cohort vs snapshot: These are cohort-based (fixed cohort tracked forward). Snapshot retention inflates numbers by 2–4x. If a source doesn't state cohort methodology, treat with skepticism.
- Geographic skew: SaaS data US-centric. Mobile data weighted toward Android. Streaming data primarily US/EU. Emerging markets typically show steeper D30 drop-off but flatter D365 tails.
Metrics
B2B SaaS Revenue Retention
Net Revenue Retention (NRR) — 12-Month Cohort
Definition: (Starting ARR + expansion − contraction − churn) / starting ARR, measured over 12 months for customers who existed at period start. Excludes new logos.
| Segment | Median | 25th Pct | 75th Pct | Top Decile |
|---|---|---|---|---|
| SMB SaaS (ACV <$15K) | 97% | 88% | 103% | 108% |
| Mid-Market SaaS ($15K–$100K) | 104% | 96% | 112% | 120% |
| Enterprise SaaS (>$100K) | 112% | 103% | 122% | 135% |
| Infra / DevTools SaaS | 118% | 108% | 128% | 140%+ |
| PLG / Self-Serve SaaS | 106% | 95% | 118% | 128% |
Trend: SMB NRR compressed 4–6 points YoY in 2025 as expansion stalled and seat-based contraction accelerated. Enterprise and DevTools held steady. Top-decile performers (Snowflake, Datadog tier) still clearing 130%+.
Red flag threshold: SMB NRR <90% = revenue base contracting. Enterprise NRR <100% = churn exceeds expansion, investigate urgently.
Gross Revenue Retention (GRR) — 12-Month Cohort
Definition: (Starting ARR − contraction − churn) / starting ARR. Excludes expansion. Pure survival measurement.
| Segment | Median | Healthy Range | Alarm Threshold |
|---|---|---|---|
| SMB SaaS | 82% | 80–90% | <75% |
| Mid-Market SaaS | 90% | 88–94% | <85% |
| Enterprise SaaS | 94% | 92–97% | <90% |
| Infra / DevTools SaaS | 93% | 90–96% | <88% |
Red flag threshold: GRR gap from NRR >25 percentage points = company is masking real churn with expansion; dangerous at scale.
Consumer Mobile App Retention (n-day)
D1 / D7 / D30 Retention by Category
Definition: Percentage of users from install cohort active on exactly day N. Median across 2025 cohorts.
| Vertical | D1 (Median) | D7 (Median) | D30 (Median) | D30 Top Quartile |
|---|---|---|---|---|
| Mobile Gaming | 31% | 14% | 7% | 14% |
| Fintech (Neobank / Payments) | 27% | 17% | 11% | 22% |
| Social / Messaging | 27% | 10% | 5% | 12% |
| E-commerce / Retail | 22% | 11% | 5% | 10% |
| Health & Fitness | 24% | 7% | 3% | 8% |
| Education / Learning | 21% | 8% | 3% | 9% |
| Productivity / Utility | 28% | 13% | 8% | 16% |
| Media & Entertainment | 25% | 11% | 6% | 13% |
Platform delta: iOS runs ~3pp higher than Android across categories (e.g., iOS D1 27% vs Android D1 24%; iOS D30 8% vs Android D30 6%).
Red flag threshold: D30 below 3% for any category except pure-media apps indicates severe onboarding or product-market fit problems.
D90 and D365 Retention — The Long Tail
Definition: Same cohort still active at 90 days (3 months) and 365 days (1 year) after install.
| Vertical | D90 (Median) | D365 (Median) | D365 Top Decile |
|---|---|---|---|
| Mobile Gaming | 3% | <1% | 3% |
| Fintech (Neobank) | 8% | 4% | 12% |
| Social / Messaging | 3% | 1.5% | 8% |
| E-commerce / Retail | 3% | 1% | 5% |
| Health & Fitness | 2% | <1% | 4% |
| Productivity / Utility | 6% | 3% | 10% |
"Smile curve" note: Best-in-class consumer apps show D365 stabilize or rise vs D90 (resurrection + power-user concentration). Losing 80% of users by D30 is normal; the question is what the flat tail looks like.
Consumer Subscription (Streaming & Media)
Monthly Churn — Subscription Services
Definition: Subscribers who cancel in month M / subscribers active at start of month M. Excludes involuntary churn (payment failure) unless labeled.
| Category | Median Monthly Churn | Best-in-Class | Alarm Threshold |
|---|---|---|---|
| Audio Streaming (Spotify-tier) | 1.5% | <1% (Spotify ~0.9%) | >3% |
| Video Streaming (Netflix-tier) | 2.5% | ~2% (Netflix 1.8–2.0%) | >5% |
| Video Streaming (Mid-tier) | 5.5% | 3.5% | >8% |
| News / Digital Media Subscriptions | 4% | 2% | >7% |
| Subscription Box (Physical) | 8% | 5% | >12% |
| Niche Consumer (Duolingo, Calm) | 3% | 1.5% | >6% |
Annualized: Audio ~12% annual churn; video ~40% annual across mid-tier, ~22% for Netflix-tier; subscription boxes ~60%+ annual.
Trend: Overall consumer subscription monthly churn rose from 2% (2019) to 5.5% (2025). 23% of streaming subscribers are "serial churners" rotating between services.
Resurrection / Win-Back Rate
Definition: Former subscribers who return within N months of canceling.
| Category | 6-Month Win-Back | 12-Month Win-Back |
|---|---|---|
| Video Streaming (Netflix reference) | 50% | 61% |
| Audio Streaming | 30% | 42% |
| News / Digital Media | 20% | 28% |
| Subscription Box | 15% | 22% |
Marketplace Retention (Two-Sided)
Demand-Side (Buyer) GMV Retention
Definition: For a monthly cohort of buyers, sum of GMV they transact in month M+N / GMV they transacted in month 0. Tracks whether buyers come back and spend more (or less).
| Month | a16z Median | Top Quartile | Great Marketplace |
|---|---|---|---|
| Month 1 | 66% | 75% | 85%+ |
| Month 3 | 57% | 68% | 80%+ |
| Month 6 | 50% | 62% | 75%+ |
| Month 12 | 30% | 45% | 60%+ (smile curve) |
| Month 24 | 22% | 38% | 60%+ (flat or rising) |
Curve shape matters more than level: Great marketplaces exhibit a "smile" — retention dips through month 6–9 then rises as power buyers concentrate. Merely-OK marketplaces show monotonic decay.
Supply-Side (Seller) Retention
Definition: Sellers active in month M+N / sellers active in month 0 (same cohort).
| Month | Median | Top Quartile |
|---|---|---|
| Month 1 | 75% | 85% |
| Month 3 | 65% | 78% |
| Month 6 | 55% | 70% |
| Month 12 | 38% | 55% |
Note: Supply retention typically runs 5–15pp higher than demand retention because sellers invest in onboarding (listings, brand, tooling).
Fintech (Neobank & Payments App) Retention
Neobank Cohort Retention
Definition: Users who signed up for a neobank and have at least one active transaction in the given month.
| Month | Median | Top Quartile |
|---|---|---|
| Month 1 | 62% | 78% |
| Month 3 | 45% | 65% |
| Month 6 | 32% | 52% |
| Month 12 | 22% | 42% |
Key behavior: Primary-account neobanks (where user deposits salary) retain 3–5x better than secondary/experimentation accounts. Direct-deposit activation is the leading indicator of Month-12 retention.
Red flag threshold: Month-6 retention <25% = users treating account as trial; acquisition spend likely unrecoverable.
Composite Metrics & Rules of Thumb
| Rule | Formula / Threshold | Interpretation |
|---|---|---|
| Quick Ratio (SaaS) | (New MRR + Expansion) / (Churn + Contraction) ≥ 4 | Healthy growth engine — adding $4 for every $1 lost |
| NRR − GRR Spread | NRR minus GRR, compared to 20pp | Gap >25pp = expansion masking churn; gap <10pp = under-monetizing existing customers |
| Andrew Chen 80/20 | ~80% of mobile users churn by D30 even for great apps | Focus on the flat tail (D30+) — that's where LTV lives |
| Marketplace Smile Test | Month-12 demand retention > Month-6 demand retention | Indicates power-user concentration and genuine habituation |
| Streaming Annual Churn Ceiling | Monthly × 12 × 0.9 (compounding) | Monthly churn ≤ annual churn / 10 for a healthy service |
| Fintech Direct-Deposit Lift | Direct-deposit users retain 3–5x primary baseline | If <20% of accounts have direct deposit by M3, value prop is weak |
Constraint: Rules of thumb break down for early-stage companies (<1K monthly cohort) and for products with long natural purchase cycles (tax software, wedding apps).
Segment Definitions
| Segment | Definition | Typical Characteristics |
|---|---|---|
| SMB SaaS | ACV <$15K, <100-employee buyer, self-serve or inside-sales motion | Higher churn (10–15%/yr), fast sales cycles, seat-based contraction risk |
| Mid-Market SaaS | ACV $15K–$100K, 100–1,000-employee buyer | Hybrid motion, 5–10%/yr churn, meaningful expansion possible |
| Enterprise SaaS | ACV >$100K, >1,000-employee buyer, field sales | <5%/yr logo churn, heavy expansion, multi-year contracts |
| Consumer Mobile App | App-store distributed, cohort measured n-days post-install | Steep D1–D30 curve; retention economics depend on monetization mix |
| Consumer Subscription | Direct-bill monthly/annual, non-app primary channel | Lower D1 churn but higher annual churn via cancel events |
| Marketplace (Buyer) | Buyer cohort tracked by GMV retention — dollars spent, not just sessions | Smile curves indicate PMF; decay curves indicate transactional |
| Marketplace (Seller) | Seller cohort tracked by GMV generated or listings active | Usually 5–15pp higher than buyer retention due to switching cost |
| Fintech Neobank | Checking/deposit account with active transactions in month | Direct deposit is leading retention indicator |
| Fintech Payments App | P2P or bill-pay app with ≥1 transaction in month | Habit-loop driven; Month-3 retention most predictive |
Year-over-Year Trend Summary
| Metric | 2024 | 2025 | 2026 | Direction |
|---|---|---|---|---|
| SMB SaaS NRR (median) | 103% | 100% | 97% | ↓ 3pp (expansion compression) |
| Enterprise SaaS NRR (median) | 113% | 112% | 112% | → Stable |
| Consumer Mobile D30 (aggregate) | 8% | 7.5% | 7% | ↓ 0.5–1pp (CAC saturation) |
| Streaming Video Monthly Churn (mid-tier) | 4.5% | 5.0% | 5.5% | ↑ 0.5pp (fatigue) |
| Marketplace M12 Demand Retention | 32% | 31% | 30% | → Roughly stable |
| Neobank M12 Retention (median) | 20% | 21% | 22% | ↑ (growth-to-profitability refocus) |
Common Misinterpretations
- Using snapshot retention as cohort retention: Dashboards often show "retention = MAU / installs-last-30-days," which overstates true cohort retention by 2–4x. Always label methodology.
- Applying SaaS NRR thinking to consumer apps: A consumer app has no "expansion revenue" unless it has tiered subscriptions. Don't chase NRR-style metrics for cohort engagement analysis.
- Comparing D30 across verticals: A 5% D30 retention rate is disastrous for fintech (median 11%) but healthy for ecommerce (median 5%). The vertical defines the threshold.
- Treating median as a target: Median means 50% of companies are below it. For ambition, use 75th percentile. For sanity checks, use median.
- Ignoring the curve shape: Two products with identical M12 retention can have very different long-term value if one shows a smile (power users concentrating) and the other shows monotonic decay.
- Forgetting involuntary churn: In subscription businesses, 20–40% of observed monthly churn is involuntary (card failures). A 5% reported monthly churn may be 3% voluntary + 2% involuntary — fixable separately.
- Projecting LTV from <6 months of data: Cohort curves need at least 6 months (ideally 12) to be reliable. Early-cohort LTV extrapolations are routinely off by 2–3x.
When This Matters
Fetch this card when an agent is benchmarking retention for a specific vertical, validating a financial model's retention assumptions, diagnosing whether a company's cohort curves are healthy, or setting KPI targets that reference industry norms. Especially relevant during board prep, diligence, or strategy reviews where cross-vertical comparisons come up.
Sources
- Optifai — B2B SaaS NRR Benchmarks, 939 Companies by Segment & ACV Tier (2025-11-15)
- Enable3 — App Retention Benchmarks for 2026: How Your App Stacks Up by Industry (2026-01-20)
- GameAnalytics — Mobile Retention Benchmarks 2026 (2026-01-20)
- a16z — GMV Retention: The Marketplace Metric Most Ignore (2025-09-10)
- Andrew Chen — New Data Shows Losing 80% of Mobile Users Is Normal (2025-06-01)
- Churnkey — Churn Rates for Streaming Services, 2026 Market Analysis (2026-01-10)
- Recurly — 2026 State of Subscriptions (2026-02-01)
- a16z — Retention Is All You Need (2025-10-20)