The metrics VCs use to assess a recurring revenue model go well beyond MRR. Net revenue retention and cohort shape tell the real story, and founders who understand them before they fundraise close faster and on better terms.
Most founders building a recurring revenue model make the same mistake: they optimize for the number that looks best in a pitch deck. MRR climbs, the chart goes up and to the right, and they assume investors are satisfied. Some are. But the ones writing larger checks aren't looking at MRR in isolation. They're pulling apart the quality of that revenue, and the difference between a business that compounds and one that just keeps running to stay still shows up in metrics most founders aren't tracking.
Not all recurring revenue is equal. A SaaS business with $3 million ARR and 115% net revenue retention is worth significantly more than one with $5 million ARR and 85% NRR. The first is growing without adding a single new customer. The second is covering attrition with new sales, and that treadmill shows up clearly in the data. That distinction drives valuation multiples in a subscription business model more than almost anything else, and founders who understand it before they sit down with a Series A investor have a real and measurable advantage.
Net revenue retention tells you how much revenue you keep from your existing customer base over twelve months, after accounting for upgrades, downgrades, and churn. Start January with $100,000 in MRR from your current accounts. End December with $115,000 from those same accounts, without counting a single new logo. Your NRR is 115%. It's one of the few metrics that captures the compounding dynamic of a healthy subscription business model, the point where growth comes from the installed base rather than the sales team alone.
The benchmark matters. Bessemer Venture Partners, whose State of the Cloud report is among the most-cited references in SaaS investing, considers 120% NRR elite for enterprise software. For SMB-focused products, 100 to 105% is more realistic given the higher churn rates in that customer segment. Below 100% means you're losing ground on your existing base and using new customer acquisition to cover it. That works until growth slows, and in a tightening market, it's a fragile model to carry into a fundraise.
Snowflake is the clearest example of what elite NRR looks like in practice. The company reported net revenue retention above 130% consistently through its hypergrowth years because its consumption-based pricing meant customers naturally spent more as their data storage and query needs grew. The product usage curve and the pricing model were aligned. Most founders don't think about that alignment until they're already locked into a pricing structure with flat monthly rates or fixed seat tiers that cap what an existing customer can ever spend.
MRR vs ARR and what the waterfall actually shows
MRR and ARR aren't just different time horizons for the same number. MRR is a monthly snapshot of contracted revenue, normalized. ARR is MRR multiplied by twelve, which works cleanly for pure subscription models with consistent monthly billing. The problem is that most SaaS companies have a mix: annual contracts, monthly plans, usage-based components, professional services, a legacy deal or two. When founders annualize lumpy or one-time revenue to build their ARR figure, they're inflating a number that investors will quietly unwind during diligence.
What VCs actually want is ARR broken into components: new ARR added in the period, expansion from existing customers, contraction from downgrades, and churned ARR. This is the ARR waterfall, and building it requires that your CRM and billing systems are actually communicating. Many early-stage companies discover, when they try to construct this for the first time, that their data is inconsistent, contracts booked differently, upgrades recorded against different fields. Fix that before you're in a fundraise. Not during one, when every inconsistency becomes a negotiating point for a lower valuation.
Cohort analysis: the view investors trust most
If there's one tool that separates founders who genuinely understand their recurring revenue model from those who've learned the vocabulary, it's cohort analysis. A cohort is a group of customers acquired in the same period, tracked across subsequent months to show how their revenue evolves. A well-shaped cohort chart shows early churn that stabilizes quickly, with surviving customers holding steady or spending more over time. A curve that keeps declining for twelve to eighteen months with no flattening tells you that every cohort is eventually worth very little, regardless of what the aggregate MRR chart looks like.
HubSpot's early cohort data, disclosed in its S-1 filing ahead of the company's 2014 IPO, showed exactly the shape that makes a recurring revenue business credible to investors. Customers who stayed past the first year expanded their usage substantially, producing rising revenue per cohort even as some accounts churned out. That pattern, contraction in customer count but expansion in revenue from the survivors, is what a genuine land-and-expand subscription business model produces. Most founders describe their motion that way. Far fewer have the cohort data to demonstrate it.
Building the data before you need it
The practical problem is that cohort analysis and ARR waterfalls require clean historical data, and most early-stage companies don't have it. Stripe for billing, a spreadsheet for customer tracking, a CRM configured inconsistently by someone who left two years ago. You can't retrospectively reconstruct reliable cohorts from that, and attempting it produces numbers that fall apart under light scrutiny. You can only start building clean, consistent records from wherever you are now, and do it well before the pressure of a fundraise creates urgency around the gaps.
Decide on your revenue recognition rules and stick to them. If a customer pays annually upfront, record it as MRR spread across twelve months, not as a lump sum in the month of payment. If a customer upgrades mid-contract, define exactly how you account for the incremental value. These decisions seem minor until you're showing a Series B investor a three-year ARR waterfall that was built on four different internal interpretations of what counts as a contract start date.
Tools like Chargebee, Maxio, or ChartMogul can automate most of this if you're building a SaaS revenue model from scratch, but they're only as good as the data you put in. Companies that arrive at fundraises with internally consistent revenue data tend to close faster and at better terms. Not because investors are impressed by the software, but because diligence moves quickly when the numbers don't contradict each other and nobody is waiting three weeks for a corrected report.
Frankly, investors have seen enough pitches built on inflated ARR and thin cohort stories to be skeptical by default. The founders who earn credibility quickly show the waterfall unprompted, name their NRR, and have the cohort chart ready before anyone asks for it. That preparation doesn't just answer investor questions. It tells them something about how you actually run the business, which is what they're really trying to assess, because a clean recurring revenue model is evidence that you understand what you're building well enough to measure it accurately.
Also read: How to Negotiate SaaS Contracts and Stop Overpaying at Every Renewal • Build a SaaS Metrics Dashboard That Survives Investor Due Diligence • SaaS Pricing Page Best Practices That Turn Browsers Into Buyers
