Benchmark Update — September 07, 2026: NRR Cracks Below 102%, CAC Climbs 14%, and the Growth Model Everyone Built in 2021 Officially Expires

Here is the uncomfortable version of today’s market: the index recovered, your unit economics did not.

IGV — the iShares Expanded Tech-Software ETF, the closest thing B2B SaaS has to a public mood ring — closed September 3 at $106.81, roughly 40% off its lows. That reclaims most of the ~$2 trillion in market cap that evacuated B2B software eight months ago during what the market has now permanently nicknamed the SaaSpocalypse. Cap tables feel better. Boards feel better. Bankers are calling again.

But index-level recovery is not company-level recovery, and the operating benchmarks published this quarter make that distinction brutally clear. Median private B2B SaaS revenue growth is sitting at 26%, down from 30% in 2022. Net revenue retention across all private B2B has compressed to 101%. Customer acquisition cost is up 14% year over year. The public market re-rated on AI narrative. The private operating reality re-rated on math.

This is a benchmark update, so let’s go metric by metric — and then talk about what actually changes in your planning model on Monday.

NRR at 101%: The Expansion Engine Stalled

The single most consequential number in this update is net revenue retention at 101%.

For context on how far that has fallen: the 2021-era benchmark for a healthy private B2B SaaS business was 110-120% NRR, and best-in-class enterprise players routinely printed 130%+. At 101%, the median SaaS company is now essentially treading water on its installed base. Expansion revenue is barely outrunning contraction and churn. Every dollar of net new growth has to be bought from the market at full price.

That is a structural change, not a cyclical dip. Three forces are driving it:

• Seat compression. AI-assisted workflows mean customers do the same work with fewer licensed humans. If your pricing is per-seat, your customer’s productivity gain is your revenue contraction. That is the defining pricing conflict of 2026.
• Procurement maturity. SaaS management platforms are now standard tooling — Europe’s SaaS management market alone is compounding at 14.7% CAGR through 2030 — and their entire value proposition is finding and killing your unused seats at renewal.
• Consolidation pressure. Buyers are collapsing point solutions into suites. Net-neutral for the suite. Net-negative for everyone else.

The tactical read: if your NRR is between 98% and 104%, you are median. You are not broken, but you no longer have a compounding asset. Stop modeling expansion as a passive tailwind and start treating it as a funded motion with an owner, a quota, and a pipeline.

CAC Up 14% YoY, Payback Windows Tightening Below 12 Months

Acquisition got more expensive while retention got weaker. That combination is why so many 2026 board decks look grim despite decent top-line optics.

CAC rose 14% year over year. Meanwhile, the Series A+ investor expectation on CAC payback has tightened to under 12 months — a standard that was considered aggressive when the benchmark sat at 18-24 months during the growth-at-all-costs era. So the cost of a customer went up and the allowed recovery window shrank. Both jaws of the vise closed simultaneously.

On LTV:CAC, the current benchmark hierarchy is worth memorizing because it has genuinely shifted:

• 3:1 — minimum viable. You are technically making money. This is the floor, not a target.
• 4:1 — healthy. This is where industry benchmarks now place a fundable, sustainable growth business.
• 5:1+ — either excellent, or you are underinvesting in growth. Diligence teams increasingly probe whether a 6:1 ratio means efficiency or timidity.

Critically, LTV:CAC is computed with your current NRR, not your 2022 NRR. Drop retention from 115% to 101% and a company that looked like 4.5:1 quietly becomes 3.1:1. Nothing about the go-to-market changed. The denominator of the customer lifetime did.

Churn Flat at 4.1% — The One Piece of Good News

Average annual churn held at 4.1%, essentially flat. This matters more than it sounds. It means the NRR compression is primarily a downsell and seat-reduction story, not a logo-loss story. Customers are staying and spending less rather than leaving.

That is a fundamentally more fixable problem. Logo churn requires rebuilding product-market fit. Downsell requires rebuilding pricing architecture — which explains the sharpest-moving benchmark on the board.

Pricing Is Where the Benchmarks Are Actually Moving

Usage-based pricing adoption hit 38% in 2026, up from 34% in 2023 and 27% in 2021. Hybrid models — platform fee plus consumption plus some seat component — now account for 61% of the market. Pure per-seat is becoming a legacy architecture.

The logic is direct: if AI reduces the number of humans touching your product, you need a value metric that grows when work grows, not when headcount grows. Snowflake and Twilio proved the consumption model at scale years ago. What’s new in 2026 is mid-market SaaS retrofitting it under retention duress. Companies migrating to hybrid pricing are reporting NRR recovery into the 108-112% band — which is roughly the entire delta between a median business and a fundable one.

The AI Margin Trap Nobody Benchmarked For

One more variable is now contaminating unit economics: inference cost volatility. This week alone, OpenAI shipped GPT-6 Astra, Microsoft AI released MAI-Transcribe-2 at aggressive undercuts to OpenAI, Google, and ElevenLabs on both price and latency, and Meta pushed Muse Spark 1.3.

The model layer is now a commodity price war. If your gross margin thesis depends on being a thin wrapper around someone else’s LLM, that thesis has a