How AI Is Redefining SaaS Metrics and Forecasting

Usage-based pricing is quickly becoming the standard for monetizing generative artificial intelligence (GenAI) features, promising stronger alignment between product value and revenue. But this shift introduces new complexity.  

With pricing tied to real-time usage rather than up-front contracts, traditional software as a service (SaaS) metrics like annual recurring revenue (ARR) and net revenue retention (NRR) become harder to rely on. Revenue no longer locks in at the point of sale. It unfolds over time, shaped by how and when customers engage with the product.

In Part 1 of this series, we looked at how GenAI is disrupting the fundamentals of SaaS pricing. Part 2 explored how companies are packaging AI features to drive adoption and monetization. In Part 3, we turn to the operational impact: how usage-based pricing is reshaping SaaS metrics, forecasting and internal alignment.

Why ARR is no longer enough

In seat-based pricing models, ARR served as the backbone of SaaS performance tracking. With fixed contracts, revenue was easy to recognize and forecasting followed a relatively stable rhythm.

To understand why usage-based pricing introduces such complexity, it helps to revisit the standard metrics that have defined SaaS performance (see Figure 1).

Usage-based pricing disrupts this structure. ARR is no longer set at the time of contract and instead unfolds based on how and when customers use the product. This variability is especially pronounced with GenAI, where usage patterns can spike or drop with little warning.

As usage-based models scale, they reveal the limits of traditional SaaS metrics. Churn, customer acquisition cost (CAC), customer payback period (CPP), lifetime value/CAC (LTV/CAC) and net revenue retention (NRR) all rely on stable and predictable revenue. When ARR fluctuates, these metrics lose precision. Forecasting becomes harder. Churn and CAC are less reliable. NRR is often inflated in the early months of a new customer ramp.

As covered in Part 1, GenAI is accelerating this shift. It challenges legacy assumptions and pushes companies to rethink how they define, measure and project revenue.

How ARR is evolving

In response to these limitations, ARR itself is evolving. It has become a catchall label for everything from fixed subscriptions to variable usage and AI-driven pricing. As companies adapt the definition to reflect different combinations of committed, usage-based and forecasted revenue, the term has become more useful but also less consistent.

To adapt, many companies now separate baseline and variable revenue. Committed revenue is often defined as contracted ARR (CARR), where traditional metrics still apply. For variable revenue, some annualize recent usage, while others adopt usage-based ARR (UARR) to reflect account-level ramp patterns, especially when usage takes six to 12 months to stabilize.

A growing number of SaaS companies now break ARR into multiple components. Some distinguish between contracted and variable revenue, while others use blended metrics that reflect real usage patterns. We have included the most common of these evolving definitions (see Figure 2). 

For many SaaS companies, AI-driven features already account for a growing share of revenue. In some cases, AI makes up as much as 50% of total subscription revenue. To give investors and internal teams better visibility, some companies are introducing a new metric: AI ARR. This helps track more clearly how usage of GenAI features is contributing to recurring revenue growth.

Public SaaS companies are already applying different interpretations of ARR (see Figure 3). 

These shifts signal a broader trend. ARR is still the headline number, but its meaning is changing. Understanding what sits underneath the label is now just as important as the number itself.

Beyond ARR: The new metrics supporting usage-based growth

Even with new definitions like CARR and UARR, headline metrics rarely tell the full story. To forecast growth more reliably, companies need granular, behavior-based indicators that show how customers ramp, expand and stabilize over time. These include:

• Realized versus predicted revenue: A feedback loop for forecasting accuracy
• Time to usage: Onboarding speed as a proxy for time to revenue
• Usage ramp rate: Consumption growth during the first six to 12 months
• Usage volatility: Stability and predictability after ramp-up
• Cohort and customer-level tracking: Segmenting usage and retention trends to improve forecasting precision

These metrics help companies navigate usage-led pricing and give investors clearer visibility into AI-driven growth. To support this shift, many teams are rethinking how they use traditional revenue frameworks. NRR and GRR still matter, but both can mislead: NRR often appears inflated during ramp periods, while GRR understates growth by excluding expansion. Some investors view NRR below 90% as a red flag, prompting deeper scrutiny of ARR. That makes it even more important to pair headline metrics with usage-based indicators and customer-level forecasting.

Together, these new metrics reflect a more dynamic way of measuring performance that emphasizes not just what is sold, but also how customers grow and engage over time.

The cross-functional impact

Adopting new metrics is just the beginning. Turning usage-based pricing into a scalable, companywide strategy requires deep coordination across every function. This includes how deals are structured, how revenue is tracked and how performance is evaluated.

• Sales teams need compensation plans that reward consumption and long-term growth, not just up-front deal size
• Customer service and product must monitor and drive usage milestones that lead to expansion
• Finance and operations are reworking billing systems, revenue recognition processes and cash flow models
• Leadership and investors are looking for new ways to evaluate performance, especially during ramp periods when metrics like NRR can appear inflated

Monte Carlo, a data and AI observability company, took this challenge head-on. Their team described AI-native usage as “spiky,” with unpredictable patterns that made traditional monthly forecasts unreliable. In response, they shifted to daily revenue tracking. Pricing ownership moved to product, and go-to-market roles were rebuilt to better align with real-time customer behavior. Daily revenue became the shared performance metric across teams and board reporting, creating tighter alignment between product value and customer outcomes.

The rise of enablers

To support transformations like this, a new category of enablement tools has emerged. These platforms help SaaS companies meter usage, automate billing, recognize revenue and manage flexible pricing models (see Figure 4). 

Vendors like Metronome, Maxio and Orb are enabling more flexible usage tracking and billing infrastructure. Others like Pocus and Subskribe are helping go-to-market teams align pricing with product adoption and customer success. These tools are quickly becoming essential infrastructure for AI-native and usage-led SaaS companies.

The path forward

AI is accelerating the shift to flexible, performance-based pricing. The bigger challenge now is scaling it across sales, product and finance in a way that sustains growth.

As usage-based pricing becomes the norm, SaaS companies need to adapt not just their pricing strategies, but also the systems they use to measure and manage performance. Real-time value delivery demands real-time visibility into revenue.

In Part 4 of this series, we’ll explore how these shifts are influencing SaaS company valuations and reshaping how investors assess growth, predictability and profitability in an AI-powered world.

L.E.K. Consulting works with leading SaaS companies to adapt pricing strategies, build more resilient forecasting models and align operations with usage-based revenue. If your team is navigating this transition, we can help. Contact us to learn more. 

L.E.K. Consulting is a registered trademark of L.E.K. Consulting LLC. All other products and brands mentioned in this document are properties of their respective owners. © 2025 L.E.K. Consulting LLC