Three distinct shifts are hitting B2B SaaS revenue right now, and each one breaks a rule that held for the past decade. Growth is no longer a straight upward line that you can force higher simply by throwing more headcount or venture capital at it. Revenue generated from the customers you already have has shifted from a nice bonus into the single factor that decides whether your business survives. At the same time, AI has quietly rewritten the cost structure underlying the product itself, forcing companies to tear up pricing models that worked fine back when the only variable was counting how many people logged in.
The data behind those realities comes from a handful of primary datasets that are worth naming right away because they disagree in instructive ways. Maxio, which analyzes more than $40 billion in real billings across 2,000-plus private companies, put average year-over-year growth at 18 percent for 2024 to 2025, with more than a third of those companies actually shrinking. SaaSRise, synthesizing over a dozen datasets covering roughly 2,500 companies, reports median private growth closer to 26 percent, while Benchmarkit’s data puts median annual net revenue retention right at 101 percent. Those numbers are not in conflict. They measure different populations using different methods, with an average pulled down by large mature companies versus a median across a broader private set. Knowing which lens you are looking through is the first rule of reading any 2026 benchmark.
The 2026 picture, and why the datasets disagree
There is no single number that captures how B2B SaaS is performing globally, and anyone who tries to quote one is flattening a market that now operates across several distinct tiers. Public companies, venture-backed private startups, bootstrapped shops, and AI-native companies are living in entirely different economic realities, priced by different buyers, and tracked with different accounting standards.
Here is where the credible medians land as of mid-2026.
Every figure below carries its population and source because the number is meaningless without them. An average and a median are not the same claim, and a private mark versus a public mark represents entirely different markets.
| Metric | 2026 benchmark | Population and source |
| Growth, average | 18 percent YoY (aggregate closer to 13 percent) | Private, all sizes; Maxio billings data |
| Growth, median | 26 percent | Private; SaaSRise / Benchmarkit |
| Net revenue retention, median | 101 percent (SMB 97 percent, enterprise exceeding 118 percent) | Private, by ACV; Benchmarkit / ChurnZero |
| Gross revenue retention, median | Roughly 88 percent (down from roughly 90 percent in 2022) | Private; Benchmarkit |
| Gross margin, AI-heavy | Roughly 50 percent reported versus 70 to 80 percent plus classic SaaS | Survey-stated; Growth Unhinged |
| Rule of 40, median | Roughly 31 (up from roughly 21 in 2023) | Private; KeyBanc / SaaSRise |
| EV/Revenue, private median | 5.9x (down from 6.7x in 2024) | Private M&A; Finerva |
| EV/ARR, public median | Roughly 3.2x, a decade low | Public index; SaaS Capital |
The last two rows are not directly comparable to each other, and it matters that nobody reads them that way. Private revenue multiples and public ARR multiples are built from different deal populations, different revenue definitions, and different timing. A 5.9x private mark sitting next to a 3.2x public mark does not describe a simple premium or discount between them. They are two separate markets that happen to use a similar-looking number.
Two of these figures set the tone for the rest. Retention is compressing, with median NRR sliding from roughly 105 percent in 2021 toward 101 percent per Benchmarkit, and gross retention drifting down with it. Meanwhile, public multiples have fallen to around 3.2x ARR, a level SaaS Capital calls a decade low, driven in Q1 2026 by investor fear that AI will commoditize traditional software. A market that once paid blindly for growth is now paying strictly for durability.
Growth gets harder to earn as companies scale
The comfortable story about scaling is that momentum compounds naturally over time. The data says the opposite. Maxio found growth compression showing up earlier than traditional textbook thresholds, with the sharpest slowdowns clustering around $5 million in billings and again past $25 million, rather than the $1 million, $10 million, and $50 million milestones operators are usually told to watch. Alan Taylor, Maxio’s COO, summed up the year in a line worth keeping: growth didn’t disappear in 2025; it just became a lot harder to earn.
The reason those inflection points exist is that the business machine fundamentally changes shape as it grows. A company clears its first few million on founder-led selling and warm organic demand. Somewhere around $5 million that initial motion runs out of steam, requiring the business to build a real sales organization, broaden its market positioning as the obvious early buyers saturate, and absorb longer enterprise sales cycles heavy with security reviews and procurement red tape. By $25 million, the addressable market narrows, the easy logos are gone, and pricing structures that felt fine at $1 million ARR start leaking margin at scale. The core revenue question changes right along with it. Early on, it is whether you can attract buyers at all. Later, it is whether your retention and expansion economics can carry the growth rate you need without eroding your unit margins.
None of this has actually dented operator optimism, which remains its own signal. Maxio reports that 72 percent of companies expect to grow faster in 2026 than they did in 2025. The gap between that confidence and the compression visible in the hard data is precisely where operational execution now decides the winners.
The installed base is the growth engine now
When acquiring a new logo costs more and takes longer to close, the customers you already have become the absolute cheapest growth you can buy. That shifts the center of gravity in revenue management away from pure acquisition metrics and onto two retention numbers that must be read together.
Net revenue retention answers a blunt question: how much of this year’s growth can be funded entirely by accounts you already own? At the 101 percent median, the answer for a typical private company is a modest amount. At the enterprise tier, where annual contract values clear $100,000 and NRR routinely runs above 118 percent, the answer is most of it, which explains why enterprise-heavy businesses can grow respectably even when new-logo acquisition slows to a crawl.
The major trap is reading NRR in isolation. A company can post a healthy 110 percent NRR while quietly bleeding customers in the background, because a small cluster of enterprise accounts expanding aggressively can paper over real churn underneath. Gross revenue retention is the honest check. It strips out expansion entirely to show whether the core customer base is actually stable. At a median around 88 percent, GRR reveals a lot more leakage than most glowing NRR headlines care to admit. Operators who truly understand their revenue treat GRR as a hard floor to defend and NRR as the lever to grow, never the other way around.
Expansion revenue is exposing weak pricing
Ask software leaders what hurts most about their monetization right now, and the answer is remarkably consistent. In Kyle Poyar’s 2026 State of B2B Monetization survey of 230 companies at Growth Unhinged, the single biggest complaint was not having enough expansion revenue. That is fundamentally a pricing-architecture problem wearing a growth costume.
Flat-fee pricing is the clearest offender. Poyar describes it as structurally broken for any company that needs to grow beyond new logos, because it caps a contract at a fixed number no matter how much immense value the customer extracts from the product. Per-seat pricing carries the opposite failure mode. It taxes the exact behavior you actually want to encourage, since a customer who rolls the tool out across a wider team gets punished with a much larger bill. As a result, adoption and revenue end up working directly against each other. Both models leave money on the table the moment usage and value stop tracking neatly with seat counts, which is precisely what AI-driven products have made routine.
The fix that actually works in the field is tying billing to something that scales directly with value. Intercom’s Fin AI agent provides the cleanest proof point: after moving to per-resolution pricing, where the customer pays only when the AI successfully resolves a customer support ticket, its net revenue retention reportedly climbed from 112 percent to 146 percent following the switch. Operator analyses show that usage-aligned models run roughly 20 points of NRR above comparable seat-based alternatives, representing the difference between a business that has to run frantically just to stand still and one that grows organically from its own base.
Hybrid pricing has effectively won
The market has already converged on an answer, and it is not pure usage billing. Hybrid pricing, combining a committed base subscription with a consumption meter on top, reached 37 percent adoption in the Growth Unhinged data, up from 25 percent a year prior, making it the most common commercial model in B2B software. It won because it resolves a genuine standoff between the two sides of the contract.
Corporate buyers want a predictable figure their CFO can budget for, multi-year price caps, and invoices free of nasty end-of-month financial surprises. Vendors want revenue streams that scale with the actual value a customer extracts, upside from power users, and direct protection against volatile AI compute and token costs eating away at their margins. A pure usage model gives the vendor what it wants while terrifying corporate procurement. A pure subscription does the exact reverse. Hybrid pricing holds both demands in balance, which is why institutional investors now favor it heavily over flat fees and why it has become the default recommendation for anything carrying meaningful AI compute costs underneath.
The catch is entirely operational. Every meter you add multiplies complexity across billing infrastructure, forecasting pipelines, and customer communication. Multi-variable contracts force finance teams to maintain strict metering accuracy, clear usage alerts, and dynamic invoicing, because a single surprise bill destroys the hard-earned trust the entire model depends on. Outcome-based pricing, where you charge only for a delivered business result, is widely talked about as the ultimate endgame, but Poyar’s data shows it remains out of reach for roughly 95 percent of the market. The underlying tooling and mutual trust just are not there yet.
AI changed the economics underneath the product
The reason software pricing is being aggressively rebuilt is that the cost of delivering software has stopped being nearly free. Classic SaaS ran on 70 to 80 percent-plus gross margins because serving one extra user cost almost nothing. AI-heavy products do not operate that way. Growth Unhinged’s 2026 survey indicates that AI-heavy businesses are running or targeting gross margins closer to 50 percent, materially below traditional benchmarks, because every inference call, every background agent firing off a chain of queries, and every power user consuming heavy compute carries a real variable cost that lands whether the customer pays extra for it or not.
That single economic fact ripples through every layer of revenue management. Cost-to-serve becomes difficult to forecast when autonomous workflows run continuous background queries. Token-based, credit-based, and task-based units must be tracked with absolute precision, or a handful of heavy power users will quietly erode the margin on the entire customer book. The defining question for any AI-heavy business is whether it can maintain the operating leverage that made software attractive to investors in the first place, or whether it has quietly morphed into a services business hiding behind a software logo. Maxio’s data offers a vital piece of good news here: AI shows up as a true competitive advantage when it is built natively into the core product, and as expensive noise when it is merely bolted on as a superficial feature checkbox. Investors are rapidly learning to tell the difference.
Seat-based pricing is not dying, it is being unbundled
The premature obituaries for per-seat pricing miss the mark. Growth Unhinged found that the largest companies, those clearing $150 million ARR, still lean heavily on per-seat structures at meaningful rates, and mature enterprise applications keep the seat as a foundational baseline because corporate buyers understand it and budget around it easily. What is actually happening is far more interesting than wholesale extinction. The seat is being unbundled wherever it stops reflecting true customer value, with vendors layering new consumption meters right beside it.
The alternative value vectors remain consistent across the companies leading this market shift. Some charge for platform access, opening up user accounts freely while metering a different operational axis. Others charge for compute and workload volume, billing directly on records processed, storage utilized, or data pipelines executed. Some tie price directly to business outcomes, charging for revenue generated or hours of human work saved. A growing set of platforms charges specifically for agentic tasks, representing the discrete unit of work an autonomous AI agent completes entirely on its own. The clear through-line is a structural move away from charging for who logs into the app and toward charging for what the software actually accomplishes.
The real 2026 news: how the biggest vendors repriced
The clearest evidence of this transformation is not found in a benchmark chart. It is visible in what the largest software companies did to their own commercial price lists, in public view, over the past eighteen months.
Salesforce serves as the ultimate case study. Its Agentforce product shipped three different pricing models since late 2024, all of which ended up running concurrently. It launched at $2 per conversation, which read cleanly on a marketing slide but fell apart completely in production because nobody could clearly define what counted as a single conversation when one user query triggered eight complex backend actions. The initial market result was tepid: out of roughly 5,000 early Agentforce deals, only about 3,000 converted to paid agreements. In May 2025 Salesforce pivoted to Flex Credits, an action-level consumption scheme priced at 100,000 credits for $500, with each automated action burning 20 credits. While more granular and honest about the work being done, it remained frustratingly unpredictable for corporate procurement teams. By late 2025, the company added per-user licenses starting around $125 a month under an Agentic Enterprise License Agreement, putting the traditional seat back as the predictable wrapper CFOs know how to budget. A software giant famously known for never discounting ran through an entire pricing philosophy in a year and a half and ended up offering all three models at once, letting customers self-select. That is not corporate indecision; it is a reflection of a market that has not yet universally agreed on how to buy AI, captured directly inside a single order form.
Zendesk went in the opposite direction and committed entirely to outcomes. Its Automated Resolutions model gives each pricing plan a baseline of free AI resolutions, then charges $1.50 per resolution on committed volume or $2.00 on a pay-as-you-go basis, counting only when the AI fully resolves a support ticket without human intervention, verified after 72 hours of customer silence. HubSpot shifted its AI billing toward per-resolution logic as well, reworking tier boundaries to push multi-product adoption rather than basic seat expansion. Read together, these are not isolated experiments. They represent the largest, most conservative vendors in the enterprise category abandoning static, one-size-fits-all subscription pricing in favor of models that track actual value and consumption, signaling precisely where the rest of the market is heading.
Valuation now pays for revenue quality, not raw growth
The repricing of software companies closely mirrors the repricing of software products. Finerva reports that median B2B SaaS revenue multiples recovered to 6.7x in 2024 before contracting back to 5.9x in 2025, while the public SaaS Capital Index sits near 3.2x ARR, a decade low reached as AI-disruption fears re-rated the entire sector in early 2026. The public and private markets stopped paying for unbridled top-line acceleration and started paying heavily for the structural quality of the revenue underneath.
Retention is the single metric doing most of the heavy lifting in that valuation shift, and the financial relationship is deeply nonlinear. Analysis from Aventis Advisors ties a ten-point NRR improvement, such as moving from 100 to 110 percent, directly to a two to three times jump in EV/ARR. M&A deal advisors report that companies sitting below 90 percent NRR trade near 1.2x revenue, companies sitting around 100 to 110 percent fetch near 6x, and elite companies above 120 percent command 8x or more. Growth rate combined with profit margin, known as the Rule of 40, remains the single best summary of overall business health, with the median private company pushing that combined score from about 21 in 2023 toward the low 30s, driven mostly by operational margin discipline rather than reckless growth. Capital markets are no longer buying raw recurring revenue; they are buying its durability, its organic expansion capacity, and its protected margin profile.
Vertical SaaS keeps winning on economics
One dominant pattern shows up in nearly every 2026 dataset: industry-specific software platforms consistently outperform horizontal competitors. Maxio found vertically focused companies consistently beating horizontal peers on growth resilience, and private market valuation data hands them a distinct premium to match. The underlying reason is structural rather than stylistic. Deep market specialization sharpens product-market fit, shortening sales cycles and lowering upfront customer acquisition costs. Software wired directly into the complex, specific workflows of an industry is notoriously painful to rip out, which lifts switching costs and props up gross retention floors. Domain depth supports higher contract values and real pricing power. Furthermore, many vertical platforms layer in embedded payments and financial services, opening up lucrative secondary expansion revenue streams that horizontal tools can never access. Specialization acts as a powerful defense against commoditization, and in a market terrified of exactly that, it is well worth paying for.
The signals to watch through the rest of 2026
If you are tracking your own corporate performance against the broader market, six specific numbers will tell you whether you are keeping pace. Watch whether NRR holds the 101 percent median benchmark or slips below it, since erosion there impacts asset valuation fast. Track the share of new ARR coming directly from expansion, which KeyBanc puts near 38 percent for companies above $25 million ARR and SaaSRise places at 58 to 67 percent once a company clears $50 million. Watch your CAC payback metrics for drift, because a lengthening payback period is always the earliest warning sign that a go-to-market motion is breaking down. Measure how far AI inference and compute costs are pulling your gross margins below the classic 70-plus percent threshold. Track hybrid pricing adoption past the current 37 percent level, recognizing that laggards on pricing architecture are the ones leaving vital expansion revenue uncollected. Finally, watch your own scaling curve for the $5 million and $25 million compression points, allowing you to build your next revenue motion before momentum stalls rather than reacting after the fact.
How to read these benchmarks
The numbers compiled throughout this piece originate from distinct sources measuring different slices of the software economy, and conflating them remains the most common analytical mistake in SaaS. Maxio works from hard historical billings across 2,000-plus private companies and $40 billion in actual transactions, capturing what really happened rather than what operators report in surveys. SaaSRise and Benchmarkit synthesize aggregated operating metrics across thousands of companies, smoothing out the bias of any single dataset. Growth Unhinged executes targeted executive surveys tracking 230 software and AI companies in its 2026 monetization report, meaning those numbers reflect strategic sentiment and stated intent more than audited financial records. Finerva and SaaS Capital track asset valuation. Public-market figures cannot be read as direct proxies for private startup performance, medians diverge sharply from averages, and AI-native businesses must be evaluated completely apart from legacy architectures. Always match the benchmark to your exact tier before drawing a conclusion for your own business.
Frequently asked questions
What are the most important B2B SaaS revenue benchmarks in 2026?
Median private growth sits around 26 percent (18 percent on an average basis per Maxio), median annual NRR is 101 percent scaling above 118 percent at the enterprise tier, gross retention rests near 88 percent, AI-heavy gross margins sit around 50 percent compared to 70 to 80 percent for classic SaaS, a median Rule of 40 score sits in the low 30s, and the median private revenue multiple is 5.9x (a separate figure from the roughly 3.2x public ARR multiple, which is not directly comparable).
What is actually happening to SaaS growth rates?
Growth remains the norm but is no longer universal. Maxio’s average sits at 18 percent, more than a third of companies shrank year over year, and growth compression now hits earliest around $5 million and $25 million in billings rather than traditional scaling thresholds.
Why does expansion revenue matter so much now?
Because acquiring new logos has become significantly more expensive and slower, making the installed customer base the most capital-efficient growth path available. Insufficient expansion is the top monetization complaint among software leaders, usually tracing back to flat-fee or per-seat pricing that caps revenue regardless of the actual value delivered.
How is AI changing SaaS monetization?
It is pushing software pricing away from static per-seat models toward hybrid and consumption structures that meter compute resources, tokens, or completed AI tasks. It also pulls gross margins down toward the roughly 50 percent range reported by Growth Unhinged for AI-heavy businesses, well below the classic 70 to 80 percent standard, forcing the pricing transformation in the first place.
Is hybrid pricing really becoming standard?
Yes. Adoption reached 37 percent in the 2026 Growth Unhinged survey, up from 25 percent a year prior, making a committed base subscription plus a usage meter the most common commercial model in B2B software. It wins by giving corporate buyers budget predictability while giving vendors direct upside on heavy usage.
What do investors reward most in 2026?
Investors prioritize revenue quality over raw top-line growth: durable NRR, defensible gross margins, Rule of 40 discipline, and strong vertical market positioning. The relationship between retention and valuation is steep, with a ten-point NRR improvement linked directly to a two to three times increase in revenue multiples.
