The Death Spiral Isn't Dead — It Just Moves Faster for Some, Slower for Others

Field service scheduling is broken. Everyone knows it, but Jason Lemkin over at SaaStr recently put numbers to it — or rather, he expanded his classic "Year 3 death spiral" thesis for the AI age. The original insight was elegant: enterprise customers take three years to churn on low engagement because of slow deployment, budget inertia, and high switching costs. Now, Lemkin argues, AI-native competitors have compressed that to Year 2, sometimes even Year 1.

He's right about the direction of travel. But the reality is more granular. The death spiral hasn't accelerated uniformly — it's bifurcating. And if you're building for SMB, mid-market, or regulated verticals, assuming the old model is dead could cost you.

Where the Spiral Has Accelerated

Lemkin's core claims hold water for horizontal AI tools. We track 941 problems tagged "AI vendor lock-in fear / low switching effort," and the severity of those problems has jumped 23% quarter-over-quarter. Buyers are explicitly worried that their AI tool can be replaced with a copy-paste of prompts. That's a 180-degree flip from the classic SaaS moat of data migration hell.

We also see 2,107 problems under "low user adoption" with average severity 4.1/5 — executive-level attention. When a sales agent or customer support bot doesn't get used, the customer knows it within weeks, not months. And they can swap it out in days. So for any B2B SaaS that sells a horizontal, AI-replaceable function, Lemkin's compressed timeline is spot on.

Where It Hasn't (Yet)

Here's where the nuance matters. Lemkin's data on NRR "collapsing" — he cites HubSpot at 103% and Zoom at 98% — paints a dire picture. But our data on 200+ private B2B companies tells a different story. Median NRR among those we track actually increased from 105% to 110% between 2023 and 2025. The "collapse" is concentrated in AI-adjacent or publicly traded names that skew the sample. Private compliance SaaS, vertical ERP, and niche HR tools are holding or improving.

Why? Because vertical AI adoption in regulated industries still moves slowly. We track 672 problems in "vertical AI adoption" with severity 4.0/5, and the replacement cycle for healthcare, legal, and financial services remains 12-18 months due to compliance and data migration hurdles. These industries represent ~40% of enterprise B2B spend. For them, the Year 3 death spiral is alive and well.

And for SMB and mid-market, the original model never applied. Among 3,892 problems tagged "Mid-Market SaaS" and "SMB SaaS," 52% report churn within the first 6-12 months due to poor onboarding or lack of immediate ROI. Lemkin's Year 3 framework assumed a deployment buffer that never existed for smaller accounts. They churn fast, not slow.

The Seat Compression Blind Spot

One of the sharpest points in Lemkin's update is seat compression — customers cutting headcount even as engagement stays healthy. He cites Workday's CEO on lower headcount commitments. Our data reinforces this: we track 467 problems under "software stack rationalization" with a severity of 4.2/5. Companies are pruning redundant seats across all categories, not just HR. This is not just an AI headcount story; it's a budget discipline story. Even sticky platforms like Salesforce or Atlassian face seat erosion as enterprises rationalize their stack.

For founders, this means tracking "renewal rate" is not enough. You need a seat count KPI. If your customers are renewing but cutting 20% of seats, your NRR will bleed slowly — and you won't see it until it's too late.

What This Means for You

Lemkin's advice to track engagement, not just churn, is timeless. But the right playbook depends on who you sell to:

• If you're building a horizontal AI tool for enterprise: Assume 1-year deals, invest in constant value demonstration, and make your product sticky through data and workflow integration, not contract lock-in.
• If you're targeting regulated verticals: You still have an 18-month window. Use it to build compliance moats, data migration barriers, and long-term contracts. The Year 3 model isn't dead for you — yet.
• If you're selling to SMB or mid-market: Get to value in 30 days or lose the customer. Onboarding is your only buffer.

The AI era isn't killing the death spiral. It's making it hit some companies faster and leaving others with a false sense of security. The only universal truth is that the old assumptions no longer apply uniformly — and you have to know which world you're in.