.png)
Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Which Pricing Metric Fits Biotech Startups SaaS Best per Seat per Transaction or per Outcome
Biotech SaaS has an unusual pricing problem. The people using the software are often not the thing that scales. A twelve-person computational biology team can move from hundreds of analyses to tens of thousands without adding twelve times as many scientists. At the other extreme, the final business outcome can be enormous - a nominated candidate, a successful trial, an approved drug - but years and dozens of decisions may separate the software's contribution from that result. The pricing metric sits between those two realities.
The choice matters now because the commercial model determines more than the invoice. It decides whether expansion follows scientific adoption, whether a customer can forecast spend, whether gross margin survives heavy users, and whether procurement can explain a bill to finance. Public evidence from SOPHiA GENETICS, AWS HealthOmics, Illumina, DNAnexus and Seven Bridges shows that genomic and bioinformatics workloads already lend themselves to charging as scientific work grows. At the same time, Veeva, Snowflake and Unity show three distinct ways a meter can lose alignment with value.
Monetizely's position is that per transaction should be the primary pricing metric for biotech startup SaaS, with per seat a distant second and per outcome third. We would put an annual commitment around the transaction meter for budget predictability, but the unit that drives expansion should remain a completed, customer-recognised piece of scientific work.
Scientific throughput carries more pricing signal than headcount
The first mistake is treating "transaction" as a synonym for technical usage. A transaction should not be a token, API call, CPU-second or database query unless customers already buy those things directly. For biotech SaaS, the better unit is usually one completed analysis, sample, assay batch, workflow run, simulation or report.
SOPHiA GENETICS offers unusually clean evidence. In its reporting for the year ended 31 December 2025, the company described platform analysis volume as a major revenue driver and reported 391,698 analyses, up 11% from 352,628 in 2024. An earlier 2025 filing was more explicit: SOPHiA said it generated revenue on a pay-per-analysis basis and defined platform analysis volume as analyses that generated revenue.
That is the relationship a biotech startup should want. Revenue expands because customers do more science through the product, rather than because they create another login.
Our ranking follows from that principle.
| Metric | Verdict | What to do in practice |
|---|---|---|
| Per transaction | Rank 1 - primary meter | Charge for a completed analysis, processed sample, workflow run, assay batch or similarly recognisable unit. Use an annual minimum with included volume and stated overages. |
| Per seat | Rank 2 - access meter | Use for products where a named scientist's ongoing workspace is the value, or for tightly controlled admin and approval roles. Do not make seats the main expansion engine for high-throughput analysis products. |
| Per outcome | Rank 3 - narrow supplement | Reserve for results that arrive quickly, can be measured objectively and are substantially controlled by the software. Do not base core ARR on candidate success, clinical success or regulatory approval. |
The ranking is not an argument that every form of usage pricing is good. Per transaction wins only when the transaction is close enough to the customer's value to be meaningful and close enough to the vendor's workload to keep economics under control.
A completed genomic analysis illustrates the balance. Running it creates real infrastructure cost, but the customer also recognises the analysis as productive work. Drug approval sits much closer to ultimate value, yet too many other variables intervene; CPU-hours sit closer to cost, yet the scientist does not care how many CPU-hours the platform needed.
The five-step sequence makes the metric a business choice rather than a billing choice
Monetizely's 5-Step Pricing Framework starts with Goals and Segmentation, which defines what the company wants pricing to accomplish and which customers it intends to serve. Packaging determines what those customers buy. Pricing Metric selects the unit that causes the bill to change as use or value increases. Rate Setting decides how much to charge for that unit, including commitments, tiers and overages. Operationalization then makes the model work in quoting, entitlements, metering, billing, reporting and renewal. The sequence matters here because no clever transaction unit can rescue a package built for the wrong biotech segment, and no attractive rate can rescue a transaction that the product cannot count reliably. The full logic is developed in Monetizing Agentic AI.
For a biotech startup, applying those five steps forces several decisions before anyone writes "$X per analysis" into a price sheet.
| Step | Question a biotech SaaS company must answer | Implication for the meter |
|---|---|---|
| Goals and Segmentation | Are we serving small discovery teams, high-throughput genomics labs, biopharma development teams or regulated clinical operations? | Pick the workflow whose growth matters commercially. |
| Packaging | What capability belongs in the core platform, and what deserves a higher tier? | Do not make every feature or user another billing variable. |
| Pricing Metric | What event increases when customers get more useful work done? | Prefer the completed scientific transaction. |
| Rate Setting | What is the work worth, and what does it cost to deliver at low and high volume? | Price from customer value, then test against compute, storage, support and third-party costs. |
| Operationalization | Can customers see and reconcile every billed event? | Create a durable event record, project attribution, spend view and true-up process. |
The pricing-metric step therefore cannot be separated from rate setting. A genomic analysis may cost a vendor $0.50, $5 or $50 depending on the workload, while the value to the customer could be hundreds of dollars in saved labour or avoided infrastructure. Cost establishes the floor. Delivered value establishes how much room exists above it.
Pricing purely as "compute plus 30%" leaves most of that room with the customer. Pricing against an eventual $500 million drug opportunity goes too far in the other direction because the software did not create that value alone. Per transaction gives the company a unit that can be priced from value without pretending that the vendor controls biology.
Genomics platforms already show that scientific work can carry the bill
The public market evidence does not converge on one exact billing system. Some vendors charge directly per analysis or run; others use credits, compute or task consumption. What matters is the common departure from pure headcount: payment expands when scientific workload expands.
Five current examples make the pattern visible.
| Vendor | Published meter | Date and primary source | What it tells a biotech startup |
|---|---|---|---|
| SOPHiA GENETICS | Pay per analysis; platform analysis volume is a major revenue driver | FY2025 filing, reported 2026 | The closest public analogue to the model we recommend. More chargeable analyses create expansion. |
| AWS HealthOmics Ready2Run | Fixed price per workflow run; AWS gives a current GATK example at $10 per run | Accessed 13 August 2026 | A customer-recognised analysis can abstract the compute beneath it. |
| Illumina Connected Analytics | Annual licence plus iCredits for analysis, compute and storage; monthly consumption billing is also offered | Accessed 13 August 2026 | Commitment and variable scientific usage can coexist in one contract. |
| DNAnexus | Charges arise from running analyses, storing data and egressing data | Accessed 13 August 2026 | Analysis activity can be billed to an organisation rather than to each scientist using the platform. |
| Seven Bridges | Licence price reflects platform use and account count; compute and storage are billed through project billing groups | Accessed 13 August 2026 | Access can remain part of the contract while workload carries a separate economic signal. |
SOPHiA is the strongest proof because its public disclosures connect the unit to revenue rather than merely publishing a price page. For the first quarter of 2026, analysis volume reached 107,576, 16% above the prior-year quarter, and the company again called platform analysis volume a primary driver of overall revenue.
AWS HealthOmics demonstrates the design distinction inside a single vendor. Its private workflows expose infrastructure economics through compute and file-system resources, while Ready2Run workflows simplify the customer's decision to a fixed per-run price when a run successfully completes. As accessed on 13 August 2026, AWS's own example prices three GATK-BP Germline fq2vcf 30x genome runs at $10 each, or $30 in total.
That is closer to how a startup should package its own differentiated workflow. Engineers still need to know what storage, compute and third-party licences cost. Customers should usually see the analysis they bought.
Illumina shows how to make variable spend easier to procure. As accessed on 13 August 2026, Connected Analytics requires an annual licence and sufficient iCredits for storage and analysis; Illumina also offers monthly consumption billing with a quoted not-to-exceed amount. Its current platform documentation shows the iCredit cost of an individual analysis and tracks that usage through the product.
The synthesis is important: transaction pricing does not require abandoning annual contracts. For biotech startups, the stronger architecture is recurring commitment to transaction volume, not recurring payment instead of transaction volume.
Seats fit scientific workspaces but underprice automated throughput
Per-seat pricing remains attractive because everyone understands it. Sales can quote it, finance can multiply it by headcount, and billing requires little instrumentation. A collaborative notebook can legitimately be worth more when another scientist receives a persistent workspace.
LabArchives is a good example. As accessed on 13 August 2026, its Professional ELN lists corporate pricing at $575 per user per year, while its ELN plus Inventory bundle lists $675 per user per year. The product also assigns storage and notebook capabilities at user level.
For that type of product, a scientist is a plausible unit of value. A new researcher gets a new workspace, creates records and uses collaborative functions. Seat count also tends to move with organisational scale.
The trouble starts once software performs large amounts of work independently of headcount. A computational team can increase sequencing runs, compounds screened or pipeline executions several-fold with the same scientists. Charging another £2,000 only when another employee joins means the vendor absorbs the economics of that expansion.
Veeva's 2026 Form 10-K exposes the broader weakness even in one of life sciences software's most successful businesses. Veeva generated $2.684 billion of subscription revenue in its fiscal year ended 31 January 2026, but it also states that when customers reduce sales representatives they may need fewer user subscriptions, lowering aggregate renewal fees for the affected solutions. The filing separately notes that computing infrastructure costs rose with both the number of end users and their volume of activity.
The lesson is not that seats failed Veeva. The company is far too successful for that claim. The filing shows something more useful for an early-stage vendor: headcount and activity are different variables, while a seat model primarily captures the first.
Two other cases reveal what happens when companies choose the wrong kind of variable meter.
| Example | What the primary evidence says | What broke in the pricing logic |
|---|---|---|
| Veeva, FY2026 | Customer headcount reductions can lower required user subscriptions and aggregate renewal fees. | A seat follows staffing even when the platform may retain broader strategic value. |
| Snowflake, FY2026 | Customers can optimise consumption, and better software or hardware can let them perform the same workload with fewer compute, storage and transfer resources. | Raw infrastructure consumption can fall while the business task stays constant. |
| Unity, 2023-2024 | Unity said its 2023 Runtime Fee changes produced negative customer feedback, a boycott and slower contracts and renewals; it cancelled the changes in Q3 2024 and reverted gaming customers to subscription pricing. | A transaction-like meter fails when customers reject the event being charged and cannot comfortably predict its economics. |
Snowflake deserves careful reading because consumption pricing also powers a very large business. Its fiscal 2026 filing reports a 125% net revenue retention rate and 13,328 customers, demonstrating substantial expansion. Yet the same 10-K warns that improved efficiency can let customers accomplish the same workloads with fewer billable resources and that consumption timing creates revenue variability.
For biotech founders, those two facts belong together. Variable pricing is powerful, but the closer the invoice gets to CPUs, storage operations or similar inputs, the more vendor efficiency can reduce revenue without reducing customer value.
Unity provides the opposite warning. Its 2024 Form 10-K says the Runtime Fee pricing changes announced in the third quarter of 2023 generated a high volume of negative customer feedback, including a boycott and slower new contracts and renewals. Unity cancelled the changes before they took effect and returned gaming customers to a subscription model in the third quarter of 2024.
A biotech transaction therefore needs to pass a simple test:
The customer already thinks in that unit.
More units normally mean more useful scientific work.
The software can count the unit consistently and explain each charge.
The vendor has enough influence over completion that customers do not feel they are being billed for failure.
Those conditions point towards an analysis, sample, workflow or report and away from a raw API request.
Outcome pricing is seductive in biotech because the ultimate value pool is huge. If a computational platform improves molecule selection, why not charge when a molecule becomes a clinical candidate? If the software helps choose trial sites, why not share in the value of a successful trial?
The first problem is time. FDA describes drug development as a sequence running from discovery and development through preclinical research, clinical research, FDA review and post-market monitoring. A discovery SaaS product used in the first stage therefore sits several major decision points away from approval.
The second problem is probability. A peer-reviewed Biostatistics study published in 2018 and using data from 2000 through 2015 estimated that 13.8% of programmes entering Phase I eventually reached approval across its dataset; for oncology, the estimated probability was 3.4%. Those figures are historical evidence rather than a 2026 forecast, but they show how much uncertainty sits between an early scientific decision and approval.
Pricing a discovery platform mainly on approval would make the software vendor absorb clinical design, patient recruitment, safety, efficacy, manufacturing, financing, regulatory and portfolio risks it does not control. Even candidate nomination is influenced by the biotech company's internal thresholds, capital position and strategic priorities.
Experience with outcomes-based pharmaceutical contracts makes the administration problem concrete. A peer-reviewed 2021 scoping review of 24 studies on outcomes-based medicine contracts found barriers across five areas: negotiation, defining outcomes, data, administration and implementation, and law and regulation. It also noted that suitable outcome periods must be long enough for reliable assessment but not so long that contracts become hard to operate.
Outcome pricing is therefore strongest much closer to the software. A laboratory optimisation product might attach a performance payment to a validated report completed within an agreed turnaround time. A data-quality product could potentially price around a completed, accepted dataset if acceptance criteria are objective.
Those are operational outcomes, not drug outcomes. More importantly, even there Monetizely's position remains transaction-first: the recurring business should be paid for completed work, with any performance payment kept subordinate to that primary meter.
A transaction commitment gives buyers predictability without flattening expansion
CFO objections to usage pricing are often objections to surprise, not to variable pricing itself. A biotech company funded against milestones needs to know whether next year's software bill is likely to be $80,000 or $800,000. A well-designed transaction contract can answer that question without retreating to flat seats.
The contract should therefore separate what is fixed from what expands. The annual commitment pays for a known volume of transactions plus continued platform access, security, integrations and support. Once included volume is used, transparent rates apply to further transactions, preferably with lower unit prices at meaningful scale.
Regulated environments make traceability especially valuable. FDA's October 2024 final guidance on electronic systems in clinical investigations emphasises trustworthy and reliable electronic records; longstanding FDA guidance defines an audit trail as a secure, computer-generated, time-stamped record that can reconstruct events. We infer a commercial design principle from that regulatory practice: even when the billing record itself is not a regulated record, a biotech buyer will trust a transaction meter more readily when each charge can be traced to a project, workflow, timestamp and status.
A simple three-year scenario illustrates why the primary meter matters more than the wrapper around it.
Operating year Active scientists Completed analyses Candidate nominations Seat model Transaction-first model Outcome-first model Discovery year 12 8,000 0 $18,000 $24,000 $25,000 Scale-up year 18 30,000 1 $27,000 $54,000 $125,000 High-throughput year 20 80,000 0 $30,000 $129,000 $25,000 The operating story changes dramatically while headcount barely does. From the first to the third year, analyses rise tenfold, while the seat bill increases only 67%. The outcome model moves in the opposite direction to workload in the final year because no candidate happens to be nominated. Transaction pricing is the only one of the three that rises consistently with the work being delivered.
Rate setting still matters. A founder who calculates an analysis costs $3 and automatically charges $4.50 has merely converted cloud resale into SaaS. The relevant question is what the customer avoids or gains: bioinformatician time, infrastructure work, workflow configuration, turnaround delay, quality failures or the cost of maintaining an internal pipeline.
AWS illustrates why the distinction is practical. Its current HealthOmics page exposes compute-based pricing for private workflows but offers fixed per-run pricing for Ready2Run workflows. The underlying cost remains variable, yet the customer can buy a recognisable piece of work.
SOPHiA takes the logic further commercially. Its 2025 disclosure links pay-per-analysis revenue to the number of chargeable analyses, while its first-quarter 2026 filing attributes volume growth both to existing-customer usage and newly onboarded customers. Expansion can therefore come from deeper adoption without waiting for the customer's payroll to expand.
That is the architecture we would build for a biotech SaaS startup: annual committed transaction volume as the commercial base, a completed scientific unit as the primary meter, and explicit overage bands as the expansion mechanism. Seats remain an entitlement where necessary. Outcomes remain an optional performance layer where attribution is genuinely strong.
The practical decisions now sit above individual contract clauses:
Choose the scientific event around which the company wants to scale. Product strategy and pricing strategy should agree on whether the core business is processing samples, completing analyses, executing workflows or producing validated outputs. A company that cannot name that event probably has not yet defined what customers repeatedly buy.
Build the company plan around transaction expansion, not licence expansion. Board forecasts should distinguish new-logo growth from more scientific work inside existing accounts. For a transaction-first model, increasing throughput at an existing twelve-scientist customer should be visible as a first-class growth path.
Require product, engineering and finance to own unit economics together. Product determines which completed event customers recognise, engineering measures its delivery cost, and finance tests margin across workload types. No one function should be allowed to optimise its piece while breaking the whole price.
Earn the right to charge closer to outcomes over time. Start by proving that the software reliably completes valuable work. As data accumulates showing a repeatable relationship between that work and near-term customer results, a performance component may become defensible. Drug approval should not be the starting point.
Treat pricing architecture as part of product strategy before scale. It is easier to instrument one analysis or workflow today than to migrate hundreds of enterprise contracts after the company discovers that seats no longer track usage. The meter chosen at $1 million ARR can determine whether customer adoption translates into expansion at $20 million ARR.
Per-seat pricing helped build SaaS because people operated the software and headcount often approximated value. In biotech, that relationship is already weaker whenever software processes scientific work at machine scale. Outcome pricing reaches towards the customer's ultimate value but reaches too far for a dependable core business.
Per transaction occupies the commercially useful middle. It allows revenue to grow when customers do more science, gives finance a unit it can forecast, gives engineering a cost it can measure, and gives the customer an invoice tied to work they recognise. For biotech startup SaaS, that is not merely the least problematic option. It is the strongest primary pricing metric.
Assumptions
The scenario model uses $1,500 per active scientist per year for the seat model; a $24,000 annual transaction commitment including 10,000 analyses and $1.50 for each additional analysis; and a $25,000 annual base plus $100,000 per candidate nomination for the outcome model. The figures are used only to compare how the three meters behave as headcount, throughput and outcomes diverge; they are not market benchmarks or recommended rate cards.
Footnotes
Ajit Ghuman and Akhil Gupta, Monetizing Agentic AI, Monetizely, 2026. https://www.amazon.com/Monetizing-Agentic-AI-Handbook-Transformation/dp/B0H7Z13VKJ/
SOPHiA GENETICS, Annual Report for the year ended 31 December 2025, filed with the U.S. Securities and Exchange Commission in 2026; and Q2 2025 filing describing pay-per-analysis revenue. https://www.sec.gov/Archives/edgar/data/1840706/000184070626000006/soph-20251231.htm https://www.sec.gov/Archives/edgar/data/1840706/000184070625000026/sophiageneticssaex992q22025.htm
Amazon Web Services, "AWS HealthOmics Pricing", accessed 13 August 2026. https://aws.amazon.com/healthomics/pricing/
Illumina, "Illumina Connected Analytics ordering", accessed 13 August 2026. https://assets.illumina.com/products/by-type/informatics-products/connected-analytics/order.html
Illumina, "Illumina Connected Analytics Pricing", accessed 13 August 2026. https://help.ica.illumina.com/reference/r-pricing
DNAnexus, "Billing", official platform documentation, accessed 13 August 2026. https://documentation.dnanexus.com/admin/billing-and-account-management
Seven Bridges, "Manage your subscription", official platform documentation, accessed 13 August 2026. https://docs.sevenbridges.com/docs/manage-your-subscription
Seven Bridges, "Cloud infrastructure pricing", official platform documentation, accessed 13 August 2026. https://docs.sevenbridges.com/docs/about-pricing
Veeva Systems Inc., Form 10-K for the fiscal year ended 31 January 2026, filed with the U.S. Securities and Exchange Commission in 2026. https://www.sec.gov/Archives/edgar/data/1393052/000139305226000014/veev-20260131.htm
Snowflake Inc., Form 10-K for the fiscal year ended 31 January 2026, filed 20 March 2026. https://www.sec.gov/Archives/edgar/data/1640147/000164014726000008/snow-20260131.htm
Unity Software Inc., Form 10-K for the year ended 31 December 2024, filed in 2025. https://www.sec.gov/Archives/edgar/data/1810806/000181080625000026/unity-20241231.htm
Unity Software Inc., Q3 2024 shareholder letter filed with the U.S. Securities and Exchange Commission, 2024. https://www.sec.gov/Archives/edgar/data/1810806/000181080624000225/a2024q3shletter.htm
LabArchives, "Pricing", accessed 13 August 2026. https://www.labarchives.com/pricing
U.S. Food and Drug Administration, "The Drug Development Process", accessed 13 August 2026. https://www.fda.gov/patients/learn-about-drug-and-device-approvals/drug-development-process
Wong, C. H., Siah, K. W. and Lo, A. W., "Estimation of clinical trial success rates and related parameters", Biostatistics, published 31 January 2018, volume 20, issue 2, 2019. https://academic.oup.com/biostatistics/article/20/2/273/4817524
Bohm, N. et al., "The Challenges of Outcomes-Based Contract Implementation for Medicines in Europe", PharmacoEconomics, published 4 September 2021, volume 40, 2022. https://link.springer.com/article/10.1007/s40273-021-01070-1
U.S. Food and Drug Administration, "Electronic Systems, Electronic Records, and Electronic Signatures in Clinical Investigations: Questions and Answers", final guidance, October 2024. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/electronic-systems-electronic-records-and-electronic-signatures-clinical-investigations-questions
U.S. Food and Drug Administration, "Guidance for Industry - Computerized Systems Used in Clinical Trials", April 1999. https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/fda-bioresearch-monitoring-information/guidance-industry-computerized-systems-used-clinical-trials