From Aryan Vatsa | Product & Market Analysis
Hybrid Pricing Hit 37% of Software Companies. The NRR Edge Is 3 Points
On this page
Hybrid pricing is now the most common structure in B2B software, at 37% of 230 companies surveyed in 2026. It also carries the highest median net revenue retention of any model, at 105%. That edge is 3 points over plain subscription, and it comes from a different survey with a different sample. The correlation is real. It is far smaller than the headline implies, and it is not free.
Key takeaways
- Hybrid leads the pricing mix by plurality, not majority. 37% of 230 software and AI companies named it their primary structure in April and May 2026, up from 25% twelve months earlier.
- The whole retention ranking spans 6 points. Hybrid sits at 105% median NRR, subscription at 102%, outcome-based at 100% and consumption at 99%. Nobody is winning a category here.
- Growth and retention rank the four models differently. Outcome-based pricing grows fastest at 65% year over year and finishes third on retention. Subscription grows slowest at 34% and finishes second.
- The usage-pricing premium reported in 2021 has closed. A benchmark of nearly 600 companies then put usage-based net dollar retention at 120% against 110% for subscription peers. The 2025 data reverses that order.
What hybrid pricing actually means
Hybrid pricing charges a fixed base plus a variable component. The base buys access, seats or a tier. The meter charges for what gets consumed on top of it.
That is the whole definition. Everything interesting sits in where the line between the two is drawn, and who carries the risk when consumption moves.
Where the base ends and the meter starts
Most hybrid designs include an allowance inside the subscription and bill overage beyond it. The customer gets a predictable floor. The vendor keeps the upside when usage grows.
The allowance size is the real pricing decision. Set it high and you have subscription pricing with a rarely triggered meter. Set it low and you have consumption pricing with a minimum commitment attached.
Both extremes get reported as hybrid on a survey form. That is worth remembering before you treat the 37% as a single coherent group.
Credits are hybrid with a different label
Credit models are the fastest growing version of this. PricingSaaS tracked companies offering credits rising from 35 to 79 across 2025, a 126% increase within the top 500 B2B and AI companies.
A credit is a unit of prepaid consumption sold alongside a subscription. Structurally it is a base plus a meter with a friendlier name. Buyers accept it more readily than raw per-call billing because the spend is capped at purchase.
That framing advantage is real and it is not a pricing innovation. It is the same economics presented in a form that survives procurement.
The headline joins two surveys that never met
The claim circulating this year is that survey data across 230 software companies shows hybrid winning on net revenue retention. Half of that is right.
The 230-company survey establishes the mix. It does not publish NRR by pricing model. The retention ranking comes from a separate benchmark with a sample more than three times larger.
What the 230-company survey measured
Kyle Poyar's 2026 State of B2B SaaS and AI Monetization Report surveyed 230 software and AI companies in April and May 2026. Its finding on structure is clear: about two in five participants said they run a hybrid model.
The same survey found that only 10% of investors named flat-fee subscriptions as their preferred model, and 5% named seat-based. Hybrid took 35%, outcome-based 26% and usage-based 24%.
Note what that second finding is. It measures investor taste, not company performance. It is a sentiment reading, and it is being quoted as though it were an outcome.
What the 800-company benchmark measured
The NRR figures come from the 2025 SaaS Benchmarks Report, published by High Alpha with more than 800 participating companies. It is the ninth edition of the survey that OpenView started.
High Alpha's own analysis of that data reports the ranking directly. Hybrid pricing companies posted a median NRR of 105%. Subscription came in at 102%, outcome-based at 100% and consumption at 99%.
| 2026 State of B2B Monetization | 2025 SaaS Benchmarks | |
|---|---|---|
| Publisher | Kyle Poyar, Growth Unhinged | High Alpha, formerly OpenView |
| Sample | 230 software and AI companies | More than 800 SaaS companies |
| Fielded | April to May 2026 | 2025, analysis published February 2026 |
| Supports | Pricing model mix, including 37% hybrid | Median NRR and growth by pricing model |
| Does not support | Any NRR figure by model | The current-year mix or the 25% to 37% shift |
Both are self-reported surveys of mostly private companies. Neither audits the figures respondents submit. Combining them is legitimate as long as the join is stated, which is the part that usually goes missing.
The retention spread is 6 points across every model
Draw the four medians to scale and the story changes shape. The bars are nearly identical lengths.
| Pricing model | Median NRR | Median growth | What the pair says |
|---|---|---|---|
| Hybrid, subscription plus usage | 105% | 40% year over year | Best retention, third on growth |
| Subscription | 102% | 34% year over year | Second on retention, slowest growth |
| Outcome-based | 100% | 65% year over year | Fastest growth, third on retention |
| Consumption | 99% | 43% year over year | Second on growth, lowest retention |
Source: High Alpha's analysis of the 2025 SaaS Benchmarks Report, published February 2026. Medians across all reporting companies, not matched cohorts.
Six points is inside the noise
These are self-reported medians from private companies across every revenue band. There is no published confidence interval, no matched-cohort control and no audit.
A 3-point gap between hybrid and subscription is not a finding you would act on if it appeared in a product experiment. It would be labelled directional and sent back for more data.
My position is that the ordering here is close to arbitrary. The useful signal is that no pricing model is currently delivering the retention that private software companies were reporting three years ago.
Growth and retention rank the models in opposite order
The same dataset shows outcome-based pricing growing at 65% a year, the fastest of the four. It also puts outcome-based third on retention at 100%.
Subscription is the mirror image. Slowest growth at 34%, second-best retention at 102%. Whatever mechanism is at work, it is not one that a single pricing choice controls.
The premium that closed since 2021
This matters because the industry is still repeating a number from a different era. In November 2021 OpenView surveyed nearly 600 SaaS companies and reported that usage-based companies posted 120% net dollar retention against 110% for subscription peers, with growth of 29.9% against 21.7%.
That 10-point retention premium is still quoted in vendor material today. In the 2025 data from the successor survey, consumption pricing sits at 99% and subscription at 102%. The order has flipped and the levels have collapsed.
Two things changed in between. Software buyers went through a spending correction, and optimisation became a standing discipline rather than a crisis response. Usage-based revenue falls when customers tune their workloads, which is precisely what happened to the cohort that repriced hardest in the selloff.
Why the pricing model is probably not the cause
Correlation between hybrid pricing and retention is easy to find. A causal path from one to the other is harder, and the survey structure hides the obvious confounder.
Expansion revenue tracks company size, and so does pricing choice
High Alpha's data shows expansion revenue growing steadily as a share of total revenue with scale. It reaches 23% of revenue at $1M to $5M ARR, 34% at $5M to $20M, and 40% at $20M to $50M. Above $50M ARR, expansion overtakes revenue from new customers entirely.
Pricing structure follows the same size gradient. The 2026 monetization survey found flat-fee subscriptions concentrated in the earliest stage, at 37% of companies below $5M ARR. Per-seat pricing persists at 29% of companies above $150M ARR.
So the hybrid cohort is drawn disproportionately from the middle and upper bands, where expansion revenue is structurally larger. Mature companies both retain better and adopt meters more often. Any comparison that does not control for ARR band is measuring maturity and calling it pricing.
I would not publish the 105% figure as evidence for a pricing decision without that control. Neither publisher claims it as causal, for what that is worth. The claim gets added in the retelling.
What the overhead actually costs
The case for hybrid is that it captures value as usage grows. The cost is that you now run two billing paradigms against one customer record, and the failure modes are not symmetrical.
A subscription invoice is wrong in a way that is obvious and rare. A metered invoice is wrong in a way that is subtle and monthly.
Metering is a permanent engineering commitment
Every billable event has to be captured, rated, deduplicated and reconciled before it becomes an invoice line. Dropped events leak revenue quietly. Double-counted events produce disputes that cost more than the amount contested.
MGI Research, which tracks monetization platforms, puts revenue leakage at at least 3% to 5% of revenue across the enterprises it studies. That figure covers all industries and the firm does not disclose a sample size, so treat it as directional. The direction is the point: the leak is larger than the retention edge you are chasing.
There is a second-order cost that rarely makes the business case. Sales compensation has to be redesigned around revenue that reps do not control, and forecasting has to absorb a variable line. Neither is a project. Both are permanent.
Your buyer feels the variance before you do
Meters transfer forecasting risk to the customer, and buyers have started to price that in. Zylo's 2026 index reports that 78% of IT leaders hit unexpected charges tied to consumption or AI features in the past year, and that 61% of organisations cut projects because of unplanned SaaS cost increases.
A cancelled project is a lost expansion opportunity that will never appear in your NRR attribution. It shows up as a flat renewal, and the pricing model gets no blame. This is the same procurement reflex driving the app rationalisation cycle running through most software portfolios.
The variance is also why buyers now ask for caps, commitments and true-ups on any meter. Each of those is a contract term that partly undoes the upside the meter was supposed to deliver, and the same dynamic is reshaping how seat counts are being renegotiated at renewal.
When hybrid is worth the overhead
None of this argues for staying on seats. Seat pricing has a specific problem in an agent economy: the software does the work and the headcount does not grow, which is the mechanism behind point tools being absorbed into broader agent platforms.
The argument is narrower. Add a meter when a meter is doing work you can name, and not because 37% of a survey did.
| Condition | The test | If it fails |
|---|---|---|
| The unit tracks value the customer already sees | Can a buyer defend that unit to their own finance team without your deck? | Keep the subscription and raise the price instead |
| Your variable cost moves with usage | Does gross margin fall as your heaviest account grows? | You are adding a meter to protect a margin you already have |
| You can meter to invoice accuracy | Can finance reproduce last month's invoice from raw events? | Fix metering before you touch pricing |
| Buyers can forecast their own spend | Can a customer predict next quarter's bill within 10%? | Ship caps and commitments first, then the meter |
| Sales compensation survives the change | Would a rep be paid on consumption they cannot influence? | Expect quota disputes rather than expansion |
Rows two and three are the ones companies skip. Row two is where AI products differ from classic SaaS, because inference is a real variable cost rather than a rounding error.
Row two deserves the most attention right now. Traditional software had variable costs close to zero, so a meter was purely a value-capture device. AI features change that, and the gross margin arithmetic behind it is covered in the breakdown of how inference costs land on software margins.
Where usage genuinely tracks cost and value together, hybrid is not a retention tactic. It is the only structure that keeps a product solvent as it scales.
Where this argument is weakest
I have argued that the retention edge is small, confounded and expensive to obtain. Here is the strongest case against that reading.
Medians conceal the distribution, and the distribution is where consumption pricing lives. Snowflake reported a net revenue retention rate of 125% for fiscal 2026, on product revenue of $4.47 billion. That is 26 points above the consumption median in the benchmark.
Consumption pricing at scale, with a product whose usage compounds, produces retention no subscription business reaches. The 99% median mostly tells you that many small companies attached meters to products that do not compound. It does not tell you the ceiling.
The second weakness is timing. The retention data describes 2025. The mix data describes mid-2026, after AI features moved into general availability and after credits went mainstream. If hybrid pricing pays off through AI consumption, the payoff would not yet appear in the retention series I am quoting.
What would change my mind is a matched cohort. Companies of similar size and segment, tracked through a pricing change, with retention measured before and after. Nobody has published that, and until someone does, every version of this argument including mine is inference from cross-sectional medians. The same evidentiary gap runs through most claims about where AI spending has produced measurable return.
Frequently asked questions
What is hybrid pricing in SaaS?
Hybrid pricing combines a fixed subscription with a variable usage component. The customer pays a base fee for access, seats or a tier, then pays more when consumption exceeds an included allowance. Credit models are a common form, where prepaid credits cover metered activity. It gives buyers a predictable floor while letting the vendor capture revenue as usage grows across the account.
Does hybrid pricing actually improve net revenue retention?
The correlation is real but small. High Alpha's 2025 benchmark of more than 800 companies put median NRR at 105% for hybrid, 102% for subscription, 100% for outcome-based and 99% for consumption. That is a 6-point spread across every model. The data is self-reported and uncontrolled for company size, so it cannot show that pricing structure caused the difference.
What is a good NRR for a SaaS company in 2026?
Benchmarks cluster near 100% for private companies across all pricing models, which means the typical business roughly replaces what it loses. Anything above 110% puts you well ahead of the median in that dataset. Public consumption businesses run much higher, with Snowflake reporting 125% for fiscal 2026. Compare yourself to your own segment and revenue band rather than to a headline figure.
Is usage-based pricing better than per-seat pricing?
Not by the current retention data. In 2021 OpenView reported usage-based companies at 120% net dollar retention against 110% for subscription peers. The 2025 successor survey reverses that, with consumption at 99% and subscription at 102%. Usage pricing exposes you to customer optimisation cycles. It works best when consumption compounds with the value your product delivers.
When should you not switch to hybrid pricing?
Avoid it when your variable costs do not move with usage, when you cannot reproduce an invoice from raw metering events, or when buyers cannot forecast their own spend within a reasonable range. Also avoid it when the honest reason for the change is that your subscription price is too low. Raising the price is cheaper and faster than building a billing system.
How much does hybrid pricing complicate billing operations?
Enough to matter against the retention gain. Two billing paradigms run against one customer record, so metering, rating, reconciliation and revenue recognition all become continuous work. MGI Research estimates revenue leakage at 3% to 5% of revenue across enterprises. Sales compensation and forecasting also need redesign, because reps are now paid partly on consumption they do not control.
Where to start this quarter
Start with your own data before you touch the pricing page. Split last year's net revenue retention into two numbers: expansion that came from the base, and expansion that came from the meter. If you have no meter yet, split it into seat growth and tier upgrades.
That single split answers the question the benchmarks cannot. It tells you whether your expansion is already tracking usage, in which case a meter formalises something real, or whether it is tracking headcount, in which case a meter is a new billing system in search of a reason.
Then take the five conditions above to whoever owns billing, and get an honest answer on row three only. If finance cannot reproduce an invoice from raw events today, that is your project for the quarter. The pricing change comes after.
Related analysis
Pricing structure is one half of the story. The other half is what happens to seat counts when the software does the work, covered in the piece on seat compression, and what happens to horizontal vendors when buyers consolidate, covered in the analysis of vertical AI.
References
- Kyle Poyar, Growth Unhinged, The 2026 State of B2B SaaS and AI Monetization Report, 13 May 2026. Survey of 230 companies, fielded April to May 2026. Used for the 37% hybrid share, the shift from 25%, the stage-level mix and the investor preference figures.
- High Alpha, How expansion revenue drives sustainable SaaS growth, February 2026. Used for median NRR and median growth by pricing model, and for expansion revenue by ARR band.
- High Alpha, 2025 SaaS Benchmarks Report, ninth edition, 800+ participating companies. The underlying dataset for reference 2.
- OpenView Partners, Usage-based pricing adoption up 32%, 4 November 2021. Nearly 600 companies. Used for the 120% against 110% net dollar retention comparison and the growth figures.
- Zylo, SaaS statistics from the 2026 SaaS Management Index, updated February 2026. Used for the 78% unexpected-charges figure and the 61% cut-projects figure.
- MGI Research, Revenue leakage series part 4, 19 December 2025. Used for the 3% to 5% leakage range.
- Snowflake Inc., Fourth quarter and full year fiscal 2026 results, 25 February 2026. Used for the 125% net revenue retention rate and $4.47 billion product revenue.
- Rob Litterst, Growth Unhinged, What actually works in SaaS pricing right now, 7 January 2026, citing the PricingSaaS 500 Index. Used for credit model growth from 35 to 79 companies and 1,800 pricing changes in 2025.
Weakest thing about this source base: both benchmark surveys are self-reported by private companies, unaudited, and neither publishes confidence intervals or matched cohorts. The retention ranking they support is a correlation across cross-sectional medians, and this post treats it as nothing stronger.
Related reading