Usage-Based Billing Systems

We're moving away from charging for *access* to software and toward charging for the *work delivered* by software & AI agents. Don't freak out: this doesn't mean everything will become *pay-as-you-go* overnight. I can think of 7 flavors of charging for work: 1️⃣ Pay-as-you-go - No commitment, totally flexible - Enterprise procurement teams usually *hate* this! - Works best when your customers can bill-back the expense or bake it into an operating budget - Otherwise, there's a risk of customers policing their own usage (taximeter effect) 2️⃣ Subscription + pay-as-you-go - Small level of commitment helps 'lock customers in' and give them access to advanced features, support, etc. - Works well when the usage metric is getting commoditized (ex: SMS messages, compute, storage) -- you can advertise a low usage fee & make up for it with the subscription fee - Still not quite loved by enterprise procurement since their bill isn't predictable yet now includes multiple line items... 3️⃣ Three-part tariff (usage subscription + PAYG) - Similar to the above, but with a larger subscription fee that includes some level of usage "included" - Folks usually advertise the initial usage as a gift ("get your first 500 SMS messages for free!") - Including a minimum level of usage helps get the customer hooked & usually incentivizes more overall consumption 4️⃣ Usage-based subscription (high watermark) - Customers commit to a certain level of usage or tier (ex: up to 5,000 API calls per month); this is typically "use it or lose it" - Subscriptions are for a high watermark of usage -- if usage exceeds the plan in a given month, they immediate move into upgrade territory - Fear of overages + usage fluctuations encourages sales to over-sell & customers to over-buy 5️⃣ Usage-based subscription (annual drawdown) - Similar to the above, but the usage allocation can be consumed flexibly over the course of 12 months similar to a gift card - This gives the customer plenty of time to monitor adoption & plan for an early renewal/upgrade if usage is trending above their commit - Great for customers with seasonality or month-to-month usage fluctuations who still want a predictable bill 6️⃣ Roll-overs - If the customer doesn't consume their full allocation, they can "roll it over" to the next year -- typically only if they commit to a flat or increased renewal - More customer friendly, but also more painful to manage! 7️⃣ Adaptive flat rate - The customer commits to a usage-based subscription, but can use the product as much as they want with no overages/upgrades during that period - Their tier resets up/down at renewal based on their actual usage behavior - Much more predictable for customers while encouraging them to increase consumption (downside is that you could be stuck with the costs!) -- I suspect most folks will offer multiple options as they seek to balance lands, expands & tough procurement convos. The downside: complexity.

Welcome to this edition of the Fintech Wrap Up Newsletter! In today’s deep dive—crafted with insights from Activant Capital—we explore why usage-based billing (UBB) is emerging as one of the most impactful pricing models in modern software and services. UBB isn’t exactly new—it’s how we’ve long paid for utilities, rides, and phone bills. But in the era of SaaS, AI, and automation, it's making a powerful comeback. As software eats the world and machine-to-machine consumption surges, the traditional subscription model is losing ground. A recent report shows that 3 out of 5 SaaS companies already use some form of UBB, and that figure is expected to rise to 80% in the next few years. With SaaS growing at 7.3% CAGR and AI accelerating usage complexity, pricing is no longer a static lever—it’s a data challenge demanding flexibility and customer-centricity. The advantages are compelling: better net revenue retention (up to 9% higher than subscription-only peers), reduced churn, expanded total addressable markets, and higher customer satisfaction. But UBB isn’t without pitfalls—monitoring usage metrics accurately, managing billing systems, and ensuring revenue predictability remain key hurdles. That’s why many companies are now embracing hybrid pricing models, combining UBB with subscriptions to align value with customer behavior throughout the user journey. Underpinning this shift is a growing ecosystem of pricing infrastructure—from legacy ERP integrations to nimble UBB platforms like Metronome, M3ter, and Amberflo. Industry giants like Stripe and Zuora have jumped in through acquisitions, signaling confidence in UBB’s strategic value. And with pricing increasingly seen as a tech and data play, companies that nail dynamic, transparent, and usage-aware models will be the ones that thrive. In short, usage-based billing is no longer optional—it’s a strategic imperative. And whether you’re a SaaS founder, pricing strategist, or investor, now is the time to pay attention to how usage is reshaping value. #fintech #payments #billing

Back when the AI boom first kicked off, most startups defaulted to usage-based pricing: charging per token, message, or API call. Simple, familiar (like AWS), easy to ship. But as inference costs plummet this approach is becoming a dangerous race to the bottom. The reality is customers care about outcomes and business value. How you charge is becoming as important as what you build. We’re seeing 4 distinct pricing models as companies move away from pure consumption-based approaches: 1 - Activity-based pricing (pay per use): The default approach we've all seen, charging by tokens or compute usage. It mirrors cloud services but ultimately treats AI as a commodity. 2 - Workflow-based pricing (pay per workflow): Instead of raw usage, you price the completion of structured tasks. An AI drafting and sending an email might cost $0.10 regardless of tokens used. 3 - Outcome-based pricing (pay per result): Customers pay only when a desired outcome is delivered. Companies like Intercom and Zendesk are pioneering this with per-resolution pricing. 4 - Per-agent pricing (pay per "AI employee"): Bill an AI agent like a SaaS seat or virtual hire with a flat monthly fee. This brilliantly taps into headcount budgets, much larger pool than IT budgets (see Joanne’s “Software-as-a-Service”). The further you move from consumption-based pricing toward value-based models, the stickier your product becomes. Pricing strategy IS product strategy. Build it in early, not as a bolt-on later.

“Unlimited usage” is the most expensive lie in AI. Back in 2023, I wrote about emerging AI pricing models and flagged that seat-based pricing "becomes brittle when usage per seat explodes." Cursor just lived this reality in real-time. ICYMI: On June 16th, Cursor abruptly shifted from a request-based pricing model (500 requests/month for $20 Pro users) to compute-based billing measured in token usage. They also introduced a new $200/month Ultra plan with ~20x usage allowance. The problem? Under the old system, users could run $0.01 simple requests alongside $1+ mega prompts with massive context windows - all counting as "one request." Some power users were essentially running 100x more expensive operations while paying the same price as casual users. I would say Cursor's new pricing model is actually MORE fair than the old model. In the previous model, power users were subsidized by casual users who barely touched their limits. Now everyone pays for what they actually consume. But the backlash from this rollout was swift and brutal. Users woke up to unexpected bills with no warning about the new metering system. Cursor had to issue blanket refunds for overages between June 16-July 4 just to contain the damage. When your pricing model has a 100x cost variance hidden beneath a single "request" unit, you don't have a pricing model - you have a time bomb. 💣 What this teaches us: Power users will always find ways to maximize value within your constraints "Requests" as a unit of measurement breaks down in multi-model environments The shift from requests to compute-based billing was inevitable, not optional In AI, usage-based pricing MUST meter actual compute, not just request count. Otherwise, you get invisible margins and pricing blind spots that can kill your economics overnight. What to do if you're still on seat-based pricing: 1. Audit your usage variance immediately - Are your top 10% of users consuming 10x more compute than average? 2. Build usage transparency - Show users their token/compute consumption in real-time, not just at billing 3. Create soft guardrails - Rate limits, usage tiers, or gentle nudges before hard cutoffs 4. Plan your transition - Have a compute-based backup model ready and communicate the possibility of change early For founders: Your pricing model needs to survive contact with your most sophisticated users, not just your average ones. Design for the power user who will push every boundary, because they're the ones who'll either make or break your unit economics. I’ve heard from plenty of developers that they always knew the $20/month plan was too good to be true. So the Cursor backlash isn't really about the $200/month price point - it's about the jarring transition from subsidized usage to more sustainable economics.

In 2021, the first edition of OpenView’s usage-based pricing study reported that 45% of SaaS businesses had implemented ‘some form’ of usage-based pricing. In the second edition of their research, published in February 2023, three out of five SaaS businesses had adopted some form of this pricing model. Usage-based SaaS companies operate differently from pure SaaS companies when it comes to SaaS metrics. The definitions of Billed ARR, upgrades, downgrades, and the like undergo slight changes. Additionally, there are many more data points to consider, with usage/product data becoming a crucial indicator of customer health. At Mantys (YC W23) our experience with numerous usage-based companies has highlighted the immense potential of combining usage data with financial data. Three primary use cases that stand out are: 1. Usage-based Customer Churn Prediction Parameters such as usage trends, past churn data, and upsell trends allow businesses to predict potential churn and identify customers more likely to upgrade. This can be analyzed on both a cohort level and an individual customer level. However, such predictions can be challenging for new products. 2. Customer-wise Gross Margin (GM) Companies can calculate costs based on actual usage, enabling them to track GM at the customer or cohort level. Such insights identify cohorts with superior margins, which businesses can then focus on intensively. 3. Revenue Forecasting Combining usage and financial data for forecasting offers a more accurate picture than relying solely on overarching assumptions. Companies can predict revenues for each customer, taking into account seasonalities of usage and customer types, and can also construct the NRR cohorts accordingly. Such forecasts serve as a more precise indicator of future revenue and, by extension, the current valuation. Many companies have begun actively monitoring these metrics. However, the challenge remains in consolidating data from varied sources to track these metrics accurately and in real-time. That's where Mantys (YC W23) steps in. If you're a usage-based SaaS company seeking to understand best practices, or if you have insights on how to enhance metric tracking, we'd love to engage with you.

You're a usage-based SaaS and not tracking usage in real-time? Stripe Meters API & why you should care... At Sessions this year, Stripe announced the launch of the usage-based billing with the Meters API - I could not tell you how many SaaS businesses I've spoken to that have asked if Stripe could track and automatically update usage events in real-time so that they can offer: 1. Accurate Usage-Based Billing 2. Enhanced Visibility and Transparency -- both customers and merchants increased visibility into consumption patterns and billing 3. Strategising with Flexible Pricing Models 🚀 How it works: Sarah's GPUs (its a fake company for now!) is a company that provides GPU computing resources for tasks like machine learning, rendering, and scientific simulations. They offer different tiers of GPU power, and they want to bill customers based on their actual usage rather than a fixed subscription plan. Define Meter Events: Sarah's GPUs would send meter events to Stripe whenever a customer uses their GPU resources. These events might represent the number of GPU hours used, the amount of GPU memory consumed, or any other relevant usage metric. Configure Meters: Sarah's GPUs would create meters in Stripe to define how the meter events should be aggregated over the billing period. For example, they might have a meter that sums up the total GPU hours used by each customer per month. Set up Prices: Sarah's GPUs would define prices in Stripe for their different GPU tiers. For instance, they might have a price of $0.50 per GPU hour for their entry-level tier, and $1.00 per GPU hour for their high-performance tier. Create Subscriptions: When a new customer signs up, Sarah's GPUs would create a subscription in Stripe and associate it with the appropriate price for the GPU tier the customer selected. Send Meter Events: As customers use Sarah's GPUs' resources, the company would send meter events to Stripe, representing the actual usage for each customer. Billing: At the end of each billing period (e.g., monthly), Stripe would aggregate the meter events for each customer based on the configured meters. It would then calculate the charges for each customer's subscription based on their aggregated usage and the associated prices. For example, if a customer used 100 GPU hours on the entry-level tier in a given month, their bill would be calculated as: 100 GPU hours x $0.50 per GPU hour = $50 Thanks for coming to my TED Talk! Just a passionate Stripe about automating payments instead of spending time crunching numbers. Sarah Useful links: https://lnkd.in/dSBpDdFD https://lnkd.in/dg5RmWZm https://lnkd.in/deS_Kn72 #Stripe #StripeSessions #Sessions2024 #Stripepayments #usagebasedbilling

Customers love Usage-based pricing - but is it profitable for your company? As a consumer, I love usage-based pricing (unless the flat rate is really low, then I love that 😉). With usage-based pricing, the customer pays based on actual consumption: API calls, minutes used, messages sent. No flat monthly rate and no minimums (usually). It feels fair: no overpaying, no commitment. But for the company providing the service, it’s a lot less predictable. Usage might drop for reasons completely out of the company’s control (seasonality, customer churn, economic shifts) But costs like payroll and lease payments remain constant. It’s flexible, but it’s also volatile. To make it viable, here's what I’ve seen work: • Minimums or committed usage – Sets a revenue floor so even if usage dips, your baseline revenue doesn’t. • Tiered pricing – Protects margins as usage scales. • Prepaid credits – Smooths cash flow and builds customer commitment. • Hybrid models – Pair a defined base retainer with usage-based overages  • Margin tracking by cohort – Not all usage is equal. Some customers will always cost more than others. Know where you’re profitable. Usage-based pricing needs tight oversight - especially on margins. Track by customer, by product, and over time. And make sure Finance, Ops, Sales, and CS are tightly aligned. What’s been your biggest win, or struggle, with usage-based pricing? Share in the comments! ___________________________ Need help with your pricing  - or a quick CFO chat about your specific needs? DM me, happy to help you figure out what makes sense and when. #FractionalCFO  #UsageBasedPricing #CFO #StrategicFinance

One of the least-discussed challenges in AI adoption today is pricing. Everyone talks about model performance, benchmarks, or features. But for enterprises, the real sticking point often shows up when the bill discussion starts. The problem: current pricing models don’t align with how enterprises budget and buy. Usage-based pricing makes perfect sense for vendors, but it feels like a blank cheque for buyers. If adoption succeeds, the bill grows in unpredictable ways. No CFO wants to be surprised by a doubling in costs because usage spiked. Flat subscriptions feel safer for buyers, but they put vendors at risk. The underlying compute costs fluctuate, and a heavy customer can easily push margins underwater. Hybrid models try to balance the two, to put in predictability for buyers’ forecast, and vendors try to to defend and improve profitability. This mismatch slows progress. Solution: a new generation of pricing models. Simple enough to understand, predictable enough to budget for, but still sustainable for vendors. It could also mean having periodic reviews instead of fixed term pricing for multi year deals. That could mean outcome-based contracts, tiered usage bands with hard caps, or bundled services that absorb variability in spikes. Until AI economics are solved, adoption will remain slower than the technology itself.

Ok, let's talk about CRM pricing: what's wrong with it and how we're tackling it at Clarify... Seat-based SaaS pricing is broken. You pay for seats, even if they go unused. I saw this firsthand at Amplitude across our customers. - Most paid seats sat inactive. - Vendors don't show clear usage data, so you can't tell who's actually using what. - Collaboration suffers when only half the team has access, and admins scramble to shuffle seats around. 🔁 At Clarify, we’re all-in on pricing that’s simple and customer-friendly: - You’re charged on actual usage. No seat-based tax. - The CRM itself is free. You only pay when our AI agents do real work for you — summarizing deals, updating fields, prepping meetings, and more. - We don't feature gate any of our usage based features. - The cost per credit gets cheaper the more credits you consume. ❤️ This isn’t just a better pricing model — it’s a customer-first philosophy: - Turn features on/off as you need them. - Control usage with a built-in credit calculator. - Get transparency into what’s driving value for your team. Our usage-based pricing gives us the clearest signal for where to invest — we only succeed when we help you get work done. It also pushes us to continuously optimize and drive down costs over time. We’re the first AI-native CRM that truly charges for outcomes, not seats. And we provide full data access and portability — more on that soon. ⚠️ Not everything is perfect: - Predicting usage can be tricky, but we give you the tools to manage it confidently. - In practice, we’ve seen credit consumption stay ~90% stable month over month, so there’s generally not much surprise. Still, we think this is the fairest pricing model in the market. 💬 Curious how this works in practice? Our pricing page spells it out: clarify.ai/pricing Would love to know your thoughts on what they like/dislike about CRM pricing today.

Let's talk a bit about Stripe's usage based billing product from Sessions. The gaps in their new "reporting usage" features highlight why Orb is a *different* product from the ground up (and why companies choose to work with us!) 1️⃣ One fundamental move you need when doing usage-based billing is the decoupling of "events" with "quantities". You report raw usage, and leave it to a different layer of the system to aggregate into quantities. That aggregation layer needs to be flexible. 2️⃣ Stripe's new `Meter` APIs have higher rate limits, but they don't actually change the abstractions - you only get to do a `SUM` or `COUNT`. For anything that requires a real query (even just a `COUNT DISTINCT`), you still have to build and run your own data infra. 3️⃣ Starting today, you can report "more events", but it's a far cry from "raw events" → you get one numeric key, and no grouping or filtering capability. And note that `transform_quantities` operates at wrong layer -- you need help going from events -> quantity, not quantity -> 2*quantity. Engineers, don't settle for `"round": "up"`. You want SQL! This is a data infra problem, not a multiplication problem. 4️⃣ Unsurprising to anyone who's done it, billing on usage requires transparent invoicing - they're inexorably tied. Now with Stripe, you get more precision on your usage ...and then lose it when actually billing -- see screenshot 👇🏽 5️⃣ What happens when you report the wrong usage? Stripe asks you to get the id of the event and you have 24 hours to do so (super limiting!). This means *you need to store each event*. It's the wrong paradigm altogether -- this storage and query is exactly what should offload.