Series B diligence subjects SaaS revenue mechanics to a level of scrutiny most companies haven’t faced before.
At seed and Series A, investors often accept a reasonably credible top-line growth story. By Series B, the diligence team wants to take the ARR number apart entirely, and a model that can’t survive that disassembly becomes the story of the round whether the founder wants it to be or not.
What Series B Investors Actually Look For in ARR Modeling
The ARR waterfall is the central artifact Series B investors expect to see, and it breaks total ARR movement into four components: new ARR from new customers, expansion ARR from existing customers buying more, churned ARR from customers who leave or downgrade, and net new ARR as the sum of all three. A single aggregate ARR growth number tells an investor almost nothing about which of these forces is actually driving the business.
Cohort-level data matters more than aggregate ARR at this stage because aggregate numbers can hide real problems. A company can show healthy year-over-year ARR growth while individual customer cohorts are quietly deteriorating, with new cohorts compensating for weakness in older ones. Series B diligence teams know to ask for the cohort cut specifically because the aggregate number is the easiest place to hide a retention problem.
Modeling New ARR
A credible new ARR forecast comes from two models built independently and then reconciled against each other.
The sales capacity model works from the bottom up: number of reps, multiplied by quota per rep, multiplied by expected attainment. This model reflects what the sales organization is actually capable of producing given its current size and ramp state, and it’s the model that grounds the forecast in operational reality rather than aspiration.
The pipeline model works from a different angle: number of qualified opportunities, multiplied by historical conversion rate, multiplied by average ACV. This model reflects what the current pipeline can actually convert into, based on how the funnel has historically performed rather than how the team hopes it will perform going forward.
Validating the forecast means checking that these two models roughly agree, and investigating carefully when they don’t. A sales capacity model that assumes far more revenue than the pipeline model supports usually means the quota-to-attainment assumptions are too aggressive. A pipeline model that outpaces sales capacity may mean the company needs to hire faster than the current plan assumes. Reconciling these two views against actual historical conversion patterns is what turns a bottoms-up forecast from a guess into a defensible number.
Modeling Churn
Logo churn and revenue churn tell genuinely different stories, and conflating them obscures what’s actually happening in the business. Logo churn counts the number of customers lost, regardless of size. Revenue churn weights that loss by ARR, which means losing ten small accounts and losing one large account can produce wildly different revenue churn numbers despite similar logo churn. A business with high logo churn concentrated in small accounts and low revenue churn overall is in a very different position than one where both numbers move together.
Predictive churn modeling goes a step further by identifying the leading indicators, drawn from product usage data and customer health scores, that show up before a customer actually cancels. Declining login frequency, reduced feature adoption, and unresolved support tickets often predict churn months before it appears in the revenue numbers, and building these signals into the model gives the forecast an early-warning capability that a lagging churn rate never provides.
Modeling churn cohort by cohort, rather than applying a single flat rate across the entire customer base, produces a far more accurate forecast. Newer cohorts typically churn at different rates than mature ones, and a flat blended churn rate averages away exactly the pattern an investor wants to see: whether churn is improving or worsening as the product and onboarding process mature.
Modeling Expansion Revenue
Expansion ARR comes from three sources: seat expansion as customer teams grow, tier upgrades as customers move to higher plans, and add-on modules as customers adopt additional parts of the product. Each of these has a different growth driver and deserves to be modeled separately rather than folded into a single expansion assumption.
The most accurate way to model expansion is as a function of the installed base combined with the product roadmap, rather than as a fixed percentage applied uniformly. A company shipping a new module next quarter should expect an expansion bump tied to that launch among customers who are good fits for it, and modeling expansion this way ties the forecast to specific, identifiable drivers rather than an abstract growth rate.
Presenting expansion as a strategic asset, rather than a secondary revenue stream that happens passively, changes how investors read the business. A company that can show expansion driven deliberately by product strategy and customer success motion looks fundamentally different from one where expansion is simply whatever happens to occur after the initial sale.
Pulling It Together: The Series B ARR Waterfall
The rolling 12-month ARR bridge ties everything together into a single view: starting ARR, plus new ARR, plus expansion ARR, minus churned ARR, equals ending ARR, rolled forward month by month across a full year. This bridge is the artifact that most directly answers the question every Series B investor is asking: where exactly does growth come from, and how much of it is durable.
Investors read this waterfall closely for the balance between its components. A business generating most of its growth from new logos, with weak expansion and rising churn, reads very differently from one where expansion and retention are doing real work alongside new business, even if the headline growth rate looks similar in both.
The mistakes that undermine trust in this section are specific and recognizable. Smoothing seasonality into an evenly distributed monthly number, rather than showing the real pattern, makes the business look steadier than it actually is. Using blended churn rates instead of cohort-level detail hides exactly the deterioration investors are trying to find. Hiding small customer churn behind strong net revenue retention numbers, when a handful of large accounts are propping up an otherwise weak retention story, is one of the fastest ways to lose credibility once a diligence team pulls the underlying data apart.