The $24M ARR That Isn't
Your board deck says you have $24M ARR. The number is clean, computed the way every SaaS company computes it: take last month's revenue, multiply by twelve. It fits neatly into the growth chart, draws a satisfying upward line, and makes the Series C narrative feel solid.
There is one problem. Sixty percent of that number — $14.4M — is usage revenue. It is not contracted. It is not committed. It is a trailing-twelve-month extrapolation of consumption that fluctuated between 15% and 25% month-over-month for the entire year. Some months it spiked because a large customer ran an end-of-quarter data migration. Some months it dipped because another customer's engineering team was on holiday and their API call volume dropped by a third.
Your investors are modeling linear growth on a number that behaves like a random walk. Your FP&A team is building quarterly forecasts on trailing averages that obscure the underlying volatility. And your board is making capitalization, hiring, and expansion decisions based on a revenue projection whose confidence interval — if anyone bothered to calculate one — would span $19M to $29M.
This is not a rounding error. It is a structural gap between how hybrid-billing companies report revenue and how that revenue actually behaves. And closing that gap requires more than a better spreadsheet. It requires a fundamentally different forecasting framework.
The Confidence Interval Nobody Calculates
Your board deck says $24M ARR. If 60% is usage revenue, the real confidence interval spans $19M to $29M. Your investors are modeling linear growth on a number that behaves like a random walk.
MRR Is a Vanity Metric in a Usage-Based World
Monthly recurring revenue was the master metric of the first era of SaaS. When every customer paid a fixed amount per seat per month, MRR was a perfect proxy for the health of the business. It was predictable, stackable, and directly tied to customer count. You could forecast it with a simple formula: existing MRR + new MRR – churned MRR = next month's MRR.
That formula breaks the moment a significant portion of your revenue becomes variable. Usage revenue does not recur. It re-accrues. Each month, every customer's consumption starts at zero, and the revenue materializes as usage events accumulate over the billing period. The "recurring" part of MRR is an assumption that next month's usage will resemble last month's. Sometimes it does. Often it does not.
If you try to force variable consumption into a traditional MRR spreadsheet, you are not forecasting. You are guessing with a veneer of mathematical respectability. The trailing average smooths out the signal you actually need to see: which customers are accelerating, which are decelerating, which are about to breach their committed tier, and which are trending toward a usage cliff that will show up as a revenue miss in 60 days.
The CFOs who are navigating this transition successfully have stopped trying to make MRR work for hybrid models. Instead, they have decomposed their revenue into streams with distinct behavioral properties and forecast each stream independently. The composite forecast is more accurate, more actionable, and — critically — more defensible to a board that is used to SaaS-era certainty.
CFO Reality Check
When a board member asks for the projected revenue for Q3, how much of your answer relies on "historical averages" versus actual, real-time consumption velocity from your metering pipeline?
If your forecast is built on last quarter's invoices rather than this week's usage trends, you are driving forward while looking in the rearview mirror. The latency between consumption and invoice is where forecast error compounds.
The Four-Stream Revenue Decomposition Framework
The core insight of this framework is that hybrid revenue is not one number with one behavior. It is four distinct streams, each with its own predictability profile, data source, and forecast methodology. Treating them as a single aggregate is why traditional MRR forecasting fails. Decomposing them is how you build a forecast that is both accurate and auditable.
Figure 1: The Four-Stream Revenue Decomposition Model
| Revenue Stream | % of Rev | Predictability | Forecast Method | Confidence |
|---|---|---|---|---|
| Committed Recurring | 35–45% | Very High | Contract waterfall | ±2–3% variance |
| Overage Recurring | 15–25% | Moderate | Cohort overage curves | ±8–12% variance |
| Pure Usage | 20–35% | Low | 90-day velocity + decay | ±15–25% variance |
| Expansion Signals | 5–15% | Directional | Threshold triggers + intent | Leading indicator |
Stream 1: Committed Recurring Revenue
Predictability: Very High (±2–3%)
This is the bedrock. Committed recurring revenue comes from the fixed components of your contracts: platform access fees, seat-based minimums, annual committed-spend guarantees. It is governed by signed contracts with defined start dates, end dates, and renewal terms. Forecasting it is a contract waterfall exercise — the same methodology that on-premise software companies have used for decades.
The forecast inputs are: active contracts (start date, end date, annual committed value), renewal probability by cohort (typically 85–95% for enterprise, 70–80% for SMB), and known new bookings in the pipeline weighted by stage. This stream should be forecastable to within ±2–3% for the current quarter and ±5–8% for the following quarter. If your committed stream has higher variance than this, the problem is contract data quality, not methodology.
Stream 2: Overage Recurring Revenue
Predictability: Moderate (±8–12%)
Overage revenue comes from customers who consume beyond their committed tier. It is not random — it follows cohort-specific patterns that become visible after two or three billing cycles. The key insight is that overage behavior is sticky at the cohort level even when it is volatile at the individual customer level.
A customer who exceeded their committed minimum by 30% last quarter may exceed it by 20% or 40% this quarter, but the cohort of all customers on that pricing tier will produce overage revenue within a tighter band. The forecast methodology is a cohort overage curve: group customers by plan tier and vintage, compute the historical overage percentage per cohort per billing period, apply seasonal adjustments, and project forward.
The critical data point is the mid-cycle consumption rate. If you can measure how much of their committed tier a customer has consumed by day 15 of a 30-day billing period, you can predict end-of-month overage with significantly higher confidence than a trailing average. This is where real-time metering data becomes a financial forecasting asset, not just a billing input.
Stream 3: Pure Usage Revenue
Predictability: Low (±15–25%)
Pure usage revenue comes from customers on fully consumption-based plans with no committed minimum. This is the stream that gives CFOs sleepless nights, and understandably so. A single large customer reducing their API call volume by 40% in a given month can move the entire revenue number.
The forecast methodology for pure usage is a 90-day velocity model. You calculate each customer's daily consumption rate (events per day, API calls per day, tokens per day) over a rolling 90-day window, apply a trend (is the velocity accelerating, flat, or decelerating?), and apply a seasonal decay factor that accounts for known patterns (end-of-quarter spikes, holiday dips, fiscal-year-end bursts).
The key discipline is to express this forecast as a range with a confidence interval, not a point estimate. A pure usage forecast of "$1.8M ± $400K at 80% confidence" is dramatically more useful to a board than "$1.8M" presented as if it were a fact. The confidence interval makes the volatility visible and allows the board to make capital allocation decisions with clear-eyed awareness of the variance.
Stream 4: Expansion Signals
Predictability: Directional (leading indicator)
Expansion signals are not revenue. They are precursors to revenue — usage patterns that predict an upcoming upgrade, tier breach, or expansion conversation. A customer whose API call volume has grown 15% month-over-month for three consecutive months is signaling that their current plan is becoming undersized. A customer who has activated a new product module and is ramping usage is signaling multi-product expansion.
These signals do not belong in the revenue forecast directly, but they belong in the forecast narrative — the qualitative context that surrounds the quantitative projection. When you tell your board, "Our pure usage stream is forecasted at $1.8M ± $400K, and we see 14 customers on trajectory to breach their committed tier next quarter," you are painting a picture of forward momentum that a single ARR number cannot convey.
The data inputs for expansion signals are: consumption velocity trends (30/60/90-day acceleration), tier proximity (percentage of committed quota consumed by mid-cycle), product adoption breadth (number of distinct product modules or API endpoints in active use), and new user or new team activations within an existing account.
Building the Composite: From Four Streams to One Defensible Number
The four-stream framework does not replace the need for a single revenue projection. Boards want a number. Investors want a number. Your FP&A model needs a number to drive headcount planning, marketing budgets, and cash flow projections. The difference is how you arrive at that number.
The composite forecast is a weighted sum of the four stream projections, where each stream's weight reflects both its revenue contribution and its confidence level. In practice, this means:
- Sum the committed recurring forecast (±2–3% variance, high weight)
- Add the overage recurring forecast (±8–12% variance, medium weight)
- Add the pure usage forecast (±15–25% variance, expressed as a range)
- Apply expansion signal adjustments (directional, not arithmetic)
The result is a composite forecast with a weighted confidence interval of ±5–8% for the current quarter, compared to the ±18–30% variance that most hybrid-billing companies experience when they force everything through a monolithic MRR formula. The improvement comes not from better math, but from better decomposition: acknowledging that different revenue streams have different volatility profiles and forecasting each one with the appropriate methodology.
Figure 2: Traditional MRR vs. Four-Stream Decomposition
| Dimension | Traditional MRR | 4-Stream Decomposition |
|---|---|---|
| What it measures | Monthly recurring revenue (single number) | Four distinct streams with separate behaviors |
| Assumption | All revenue is equally predictable | Each stream has its own volatility profile |
| Forecast accuracy | ±18–30% for hybrid models | ±5–8% (weighted composite) |
| Actionability | Tells you the total; not why it moved | Tells you which stream changed and why |
| Investor confidence | Low (hidden volatility) | High (transparent methodology) |
| Data source | End-of-month invoice total | Real-time metering + contract data |
Why This Framework Requires a Real-Time Metering Pipeline
A CFO cannot run this framework manually. The four-stream decomposition is not a quarterly exercise performed on exported spreadsheets. It is a continuous process that requires live data flowing from the metering layer into the financial model. Without real-time consumption visibility, the overage forecast is a guess, the pure usage forecast is a trailing average, and the expansion signals are invisible.
This is where billing infrastructure becomes financial infrastructure.
The Aforo FP&A Layer
Aforo exposes the four revenue streams as separate, real-time data feeds. The FP&A team sees committed revenue from the contract waterfall, overage accruals computed daily from mid-cycle consumption data, pure usage revenue calculated from per-customer daily velocity, and expansion signals derived from consumption trend analysis.
Every metric is available at the customer level and the cohort level, updated daily. Month-end projections carry actual confidence intervals computed from velocity variance, not historical averages assumed to be normal. When your board asks "What is the projected revenue for Q3?" the answer is a defensible range with a transparent methodology, not a single number backed by hope.
Daily Consumption Velocity: The CFO's New Leading Indicator
The most powerful data point in the four-stream framework is daily consumption velocity per customer: the rate at which each customer is consuming their metered resources, measured and updated every 24 hours.
Consumption velocity turns the revenue forecast from a backward-looking extrapolation into a forward-looking projection. On day 10 of a 30-day billing period, if a cohort's aggregate velocity is 12% above their trailing 90-day average, your overage forecast adjusts upward in real time. If a large customer's velocity drops by 30% mid-month, the pure usage projection adjusts downward before the invoice is even generated. The finance team sees the revenue forming, not just the revenue that already formed.
This eliminates the most dangerous dynamic in hybrid-billing FP&A: the end-of-month surprise. In a traditional model, the FP&A team does not know the actual usage revenue until the billing period closes and invoices are generated. By then, the quarter is already underway and the numbers are locked. With daily velocity data, the FP&A team has a continuously updated view of where the month is trending, with enough lead time to communicate adjustments to the board and to the business before they become surprises.
Investor Communication: Turning Volatility into a Narrative Advantage
The four-stream framework also transforms how you communicate with investors. Instead of presenting a single ARR number that obscures the underlying dynamics, you present a revenue composition that demonstrates operational maturity.
A board slide that says "$24M ARR" invites the question "Is that reliable?" A board slide that says "$9.6M committed recurring (±2%), $4.8M overage recurring (±10%), $7.2M pure usage (±20% — trending upward at 8% QoQ), plus 14 expansion signals in pipeline" tells a story of a finance team that understands its revenue dynamics at a granular level. The volatility is the same. The investor's confidence in the number is fundamentally different.
The "Audit Yourself" Checklist
Before the next board meeting, your finance team should be able to answer these three questions with data, not assumptions. If the answers reveal a reliance on trailing averages and invoice-based accounting, the gap between your reported ARR and your actual revenue trajectory is wider than you think.
1. The Trailing Average Dependency
Decompose your current quarterly revenue forecast into its data sources. For each line item, ask: is this number derived from a signed contract, a real-time consumption measurement, or a trailing average of past invoices? If more than 30% of your forecast relies on trailing averages, your forecast is structurally lagging the business by 60–90 days. In a hybrid model, that latency is where misses originate.
2. The Mid-Cycle Visibility Test
On day 15 of the current billing period, can your finance team tell you — to within ±10% — what this month's usage revenue will be when the period closes? Can they tell you which specific customers are on pace to exceed their committed tier? If the answer is no, your billing system is generating invoices but not generating the consumption intelligence your forecast requires.
3. The Confidence Interval Honesty Test
When you present your quarterly revenue projection to the board, do you present it as a single number or as a range with a confidence interval? If it is a single number, ask your FP&A team to retroactively compute the variance between their projections and actuals for the last four quarters. If the variance exceeds ±10%, you are presenting false precision. A range with a transparent methodology is more credible than a point estimate that is consistently wrong.
The Bottom Line
The transition to hybrid billing does not just change how you invoice customers. It changes how you understand your own revenue. The MRR formula that served the subscription era is not equipped for a world where 60% of revenue is variable, and pretending otherwise produces forecasts that mislead the board, misinform capital allocation, and erode investor confidence.
The four-stream decomposition framework is not complicated. It is simply honest about what traditional MRR refuses to acknowledge: that different types of revenue behave differently, and forecasting them requires different methods, different data sources, and different confidence intervals.
The companies that will earn investor confidence in the hybrid-billing era are not the ones that produce the most precise single-number forecasts. They are the ones that present revenue projections with transparent methodology, granular decomposition, and honest confidence intervals. That requires a billing system that is not just an invoicing engine, but a real-time financial intelligence layer that surfaces the consumption data your FP&A model needs to function.
If your current billing system cannot tell you today's consumption velocity for every customer, it is not a billing system. It is a receipt printer. And you cannot forecast from receipts.
Ready to see your revenue through four lenses instead of one?