Is ARR Dead – Or Just Evolving? A Look At Consumption Based Pricing
May 29, 2025
“It’s hard to think of a software that could lead to such great productivity improvements that company-wide headcount needs to be reduced.” - Deloitte's 2023 CIO Pricing Report
In 2023, that quote reflected a consensus: most software enhanced productivity, but it didn’t fundamentally eliminate labor. Fast forward to today, and that belief feels outdated. AI models are writing code, resolving customer tickets, and building applications. Infrastructure products are spinning up and down dynamically based on developer and agent behavior. As technology accelerates, it’s not just software delivery that is evolving. This impact on productivity is causing pricing models to undergo their most significant transformation since the rise of SaaS .
Pricing strategy is a core input to how companies are operating, shaping everything from customer onboarding to revenue recognition to how teams are compensated. Over the past three decades, we at Notable Capital, have had the opportunity to invest in companies through each wave of pricing innovation. From the early days of perpetual licenses to the rise of SaaS, we’ve seen how pricing models influence everything from product design to enterprise value. We’ve partnered with companies like HashiCorp*, which pioneered self-managed infrastructure-as-code solutions and the subscription pricing associated with this model, and Vercel*, which leverages usage-based billing to support its frontend cloud customers. Each company’s pricing model reflects a deeper philosophy about how to deliver and capture value.
Software pricing models have always evolved to reflect changing customer needs, technological paradigms, and cost structures. Consumption based pricing is becoming more and more popular. While outcome based pricing is emerging, it's still fledgling in popularity. Let's take a closer look at consumption based pricing.
The shift to consumption-based pricing is being driven by both customer expectations and vendor incentives. On the customer side, this model offers immediate value realization. Users pay only for what they use, which makes the cost-to-value ratio transparent and allows for low-friction experimentation. The promise of this approach is, therefore, that it promotes flexibility: customers can scale up as they grow and reduce spend during low-utilization periods, which ultimately reduces churn. Enterprise customers, whose procurement departments are trained to purchase for predictability and economies of scale, need some re-training to become convinced that consumption-based pricing is a tenable solution for them. We expect this friction to decline as the cost-to-value ratio becomes more aligned.
From the vendor perspective, consumption pricing aligns business outcomes with product quality. If the product works well, usage grows, and revenue scales alongside it. Jordan Van Horn, Monte Carlo’s* COO, suggests that “the most surprising part of our move to consumption was the impact on customer health…the bar for delivering great outcomes rose significantly and it clarified the focus of our entire company to building and delivering value for our customers… (now) customer satisfaction score has risen significantly and churn has dropped to near zero.”
For infrastructure and AI companies, this model also creates a direct relationship between delivery cost and pricing. Cloud compute, inference, and storage costs fluctuate based on usage, and consumption pricing allows vendors to recoup those costs more reliably. The emergence of token-based pricing, particularly in AI, has become a standard part of the consumption model. Charging by the token—a measurable proxy for compute usage in large language models and other generative systems, extends this predictable model of recouping costs. The cost per thousand tokens varies depending on model sophistication, ranging from fractions of a cent to several cents. This structure allows for precise alignment between user output and vendor cost.
Token pricing brings a unique set of challenges: high variability in usage, pressure to deliver cost-efficiency at scale, and a greater demand for transparency and monitoring. As compute-heavy AI companies pursue adoption, they must master the balance between usage growth and unit economics. That reality is driving a new generation of pricing, product, and financial leaders to rethink what scalable, sustainable business models look like.
While consumption pricing is gaining traction, it isn’t without its challenges. For example, companies must decide how to balance the emphasis on usage growth versus driving new commitments in their sales compensation models. Some companies who’ve adopted consumption models, such as Snowflake and Databricks, have shifted back and forth between emphasizing usage and commitments in their sales compensation plans. These adaptations reflect a key reality: no pricing model is perfect. It’s not possible to build one model that works for every customer segment or sales motion. The best companies adapt and modify to continually drive toward their goals.
In AI specifically, the pricing model becomes even more critical. Some generative AI and LLM-based platforms are promising efficiency and shifting to an outcome-based pricing model, offering the opportunity to replace human labor, charging for output units, freeing up time to spend on other tasks. Pricing based on output requires vendors to build systems that are efficient, observable, and responsive to customer usage patterns—with significant implications for both technical and business teams. We’re starting to see more innovation in this area and expect vendors to move in this direction over the coming years.
The developments in consumption and output based pricing raises a deeper question for this moment in software: Is ARR dead—or just evolving? For companies embracing consumption, predictable, contract-driven ARR may no longer be the north star. According to Marten Abrahamsen, CFO of Vercel, “it's no longer really recurring in the same sense that a seat was and especially for some of the more hyped up AI platforms with high churn rates” But that doesn’t mean predictability is obsolete—it just takes a different form. In this world, usage data becomes the new currency of confidence. Founders must learn to track it, tell their story through it, and ultimately, grow into a new kind of revenue model that aligns with how software is built and consumed today which we’re seeing in companies like Paid.ai, Metronome, and Orb.
Operating under a consumption-based model introduces new complexity, especially in cash flow, revenue planning and go-to-market execution. For one, cash flow may become harder to predict. Many consumption models charge up-front commits; this can make cash flow look much better than profitability. But if the model is on-demand, invoices are often issued and payable monthly, negatively affecting cash flow. From a revenue recognition perspective, unlike subscription models where revenue is booked at the beginning of a period and therefore highly predictable for months and even years, consumption businesses often start each quarter at zero and must earn revenue through ongoing usage.
This means companies need a new way to forecast. Leading indicators such as activation rate, average daily usage, and product expansion within accounts take precedence over traditional contract-based KPIs. Some companies forecast at the rep level on a weekly basis, tracking daily revenue targets and usage cohorts in dashboards.
The complexity extends into the product itself. Assigning which specific metrics are monetized, and which aren’t, is an art. Sophisticated end-users understand what makes the meter run, and will find holes in pricing models to optimize costs. Companies must be thoughtful when assigning consumption rates to various features, in order to balance revenue generation with low-cost feature adoption. Certain use-cases of SaaS and PaaS can undermine a well-designed pricing structure, letting customers derive great value while not driving monetized consumption.
Sales compensation must also evolve. In the past, compensation was tied to upfront bookings, but this structure risks misalignment in a consumption context. Instead, companies are shifting comp plans to reward revenue generation over time. Reps may be incentivized based on incremental consumption growth within an account, ensuring alignment with long-term customer success. Hybrid models where new logo reps are paid on bookings and account managers are paid on usage are also becoming more common.
The change in pricing brings an evolution to the entire go-to-market motion. Rather than a binary "closed/won" milestone, teams now support customers through activation, usage expansion, and success enablement. Sales executives often have new skills to hone in this model. Per Dan Nemo, CFO of Monte Carlo, “the sales process doesn’t end at signature; it begins there and continues throughout the customer journey post-signature. Account Executives are incentivized to ensure customer success every step of the way.”
Consumption-based companies must redefine what performance looks like. Here are a few critical metrics to track:
Early-stage consumption businesses often present erratic revenue patterns. As we see with many AI companies now, there is high adoption, steep incline in growth, but no way to know what will happen in 12, 18 or 24 months. With how dramatically monthly numbers can swing, it can make conversations with investors challenging, especially if they are used to linear ARR-style growth.
Instead of focusing on traditional revenue benchmarks, founders should highlight:
Educating your board and investors on this model is part of the job. The best founders proactively build dashboards and metrics that reflect the reality of their business, even if it looks different from a standard SaaS motion.
The future of software isn’t priced in seats, it’s priced in seconds, queries, credits and tokens. In a world where output is immediate and infrastructure is ephemeral, pricing must align tightly with value delivered.
This new reality rewards companies that can build with flexibility, operate with rigor, and design incentives that scale with usage. It’s not just a shift in how revenue is booked; it’s a transformation in how success is built.
Founders who embrace this complexity and design for volatility, rather than trying to avoid it, will be best positioned to lead. Because when value is consumption-driven, the best products win faster than ever.
Thank you to Jordan Van Horn, Dan Nemo, Marten Abrahamsen, Arjun Rajeswaran, and Luke Flanagan for sharing their insights and experiences as operators helping shape how we measure success in usage-based businesses.
*Represents a Notable Capital portfolio company.
