Bending Spoons just announced most of the Airtable team will be laid off, explaining that small teams ship more and better. He pointed to a new Core team that has shipped 30+ top customer requests in 3 weeks, plus new Performance and Infra teams.
Almost every B2B CEO and board is asking how lean they can get and still perform as well or better.
Three B2B companies made deep cuts in 2026, and all three now have public numbers on what happened afterward:
- Cloudflare cut 20%, mostly back office, during its best quarter ever. Growth then accelerated.
- Atlassian cut 10%, mostly R&D, with its stock down by half. Cloud growth then accelerated, and the stock jumped 35% in a day.
- Bending Spoons cuts most of the staff at every company it buys. It now runs at $2.57M in revenue per employee with a 51% operating margin.
Each one cut different functions, at a different stage, for different reasons. Which functions you cut, and when, determines how lean you can go.
The Scorecard
#1. Cloudflare Cut 1,100 People. Two Quarters Later, Growth Went From 34% to 36%.
On May 7, Cloudflare announced a plan to move to an agentic AI-first operating model, cutting about 20% of its workforce, with $140M to $150M in charges. It had 5,156 full-time employees at the end of 2025.
Cloudflare was doing well when it made the cut. That same quarter it posted its best results ever: $639.8M in revenue, up 34%.
The stock dropped more than 20% after hours, even with the revenue beat.
Q2 revenue came in at $696.1M, up 36%, and the stock rose 17% after hours. Cloudflare raised full-year guidance to $2.86B to $2.87B, ahead of the $2.81B consensus. Net dollar retention was 120%.
Cloudflare cut back office and kept engineers and sellers. Prince said the cuts were mostly back-office roles AI had streamlined, with very few engineers or customer-facing salespeople affected. Quota-carrying sales was spared across every region and department.
Most of the growth came from demand. Management credited Workers and agentic workloads as key drivers of the quarter. Cloudflare is in the middle of the machine-to-machine traffic boom, and a smaller G&A team let it keep up with that demand at lower cost.
Prince also expects Cloudflare to have more employees in 2027 than at any point in 2026. He’s planning to hire again, for different roles.
My rough math: Q1 annualized revenue over 5,156 people works out to about $496K per employee. Q2 annualized over roughly 4,050 people works out to about $690K. That’s a ~40% increase in one quarter, nearly all from the lower headcount.
Three things Cloudflare did well:
- One cut. Cloudflare said smaller, repeated cuts over several quarters drag out uncertainty and slow down building.
- Generous terms. Departing staff got full base pay through the end of 2026, US healthcare through year end, and equity vesting through August 15.
- Agent usage came first. Internal AI tool usage rose more than 600% in three months, with teams across engineering, HR, finance, and marketing running thousands of agent sessions a day. Agents were already doing the work before the cut.
#2. Atlassian Cut 10% in March, Mostly in R&D. Five Months Later, Cloud Growth Hit 31%.
Atlassian cut about 1,600 people in March to put more money into AI and enterprise sales. More than 900 of those roles were in software R&D.
The market wasn’t feeling good about Atlassian at the time. It had lost more than half its market value since the start of 2026, on fears about AI-native competitors. The stock rose about 2% after hours on the news.
The June quarter:
- Revenue of $1.766B, up 28%, with cloud revenue of $1.213B, up 31%.
- Atlassian had been guiding to 23% cloud growth as recently as February.
- RPO up 44% to $4.8B, and net revenue retention above 120%.
- Record enterprise deals at the $1M, $3M, and $5M thresholds, with $5M+ ARR customers growing 70%.
- A 12% GAAP operating margin and GAAP profitability.
- The stock jumped 35% the next day.
Cannon-Brookes also plans to buy up to $250M of Atlassian stock on the open market.
The limits: fiscal 2027 guidance steps total growth down to about 13% and ARR growth to 18%, partly because Data Center revenue is expected to decline about 17% as customers move to cloud. And roughly a point of FY26 growth came from acquisitions.
Atlassian cut 900+ engineers and product people, and the product side of the business sped up. MCP server and Teamwork Graph CLI usage passed one million monthly active users, more than doubling in a single quarter. Atlassian also split the CTO role into two AI-focused positions at the same time, so it changed how engineering works along with how many people do it.
#3. Bending Spoons: $2.57M in Revenue Per Employee and a 51% Operating Margin
Bending Spoons is the extreme case, and since it’s now public, we have its numbers.
It cuts deep at nearly every company it buys. Eventbrite announced major layoffs weeks after the deal closed, and Bending Spoons reportedly cut 75% of WeTransfer’s staff after buying it. Vimeo has been through two rounds since the $1.38B acquisition.
The results:
- Revenue per full-time employee went from $1.12M in 2023 to $1.64M in 2024 to $2.57M in 2025, with AI cited as one of the drivers.
- 2,284 employees at IPO.
- Q1 2026 adjusted operating profit of $308M on $601.3M of revenue, a 51.2% margin, up from 37% a year earlier.
- More than 90% of code pull requests in the quarter were written or co-written by AI.
- The IPO priced above range at $29, a $19.5B market cap, and closed up almost 40% on day one.
$2.57M per employee is roughly 4x Cloudflare’s post-cut number.
Net revenue retention was 95% in 2025. The acquired base shrinks a little each year on average. Bending Spoons grows by buying more companies.
So Bending Spoons shows you can run a mature, slow-growth product with a small fraction of the original team and make it much more profitable. It doesn’t show you can grow a product 30% at that headcount.
Vimeo shows what gets given up. Right before the close, Vimeo had shown an ambitious roadmap, including a rebuilt review product with Premiere Pro integration and natural-language search. The January layoffs came soon after. Bending Spoons runs its products for margin.
At a $20M ARR company growing 60%, the Bending Spoons model would hurt you. At a $15M ARR product growing 5% with 150 people, it’s worth studying closely.
#4. Cloudflare Cut G&A, Atlassian Cut R&D, Bending Spoons Cut Almost Everything
- Cloudflare was growing 34% with strong demand. It cut back office so G&A would stop growing in step with revenue, and kept engineering and sales intact.
- Atlassian had slowing growth and a stock down by half. It cut engineering, reorganized how the product gets built, and put the savings into AI and enterprise sales.
- Bending Spoons buys products it doesn’t plan to grow. It cuts nearly every function to maximize margin.
Each plan fit its company. A 75% cut would have broken a business growing 34% like Cloudflare. Keeping all of Vimeo’s engineers would have cost Bending Spoons a lot of margin.
#5. SaaStr: 3 Humans, ~20 Agents, About 6 That Do Most of the Daily Work
We’ve gone further than any of these three, at a much smaller scale.
SaaStr runs at $15m-$20m in revenue with 3 humans and about 20 AI agents in production. That’s around $6M of revenue per human.
The agent count peaked near 30. We’re consolidating back toward 20. About 6 do most of the daily work.
What the agents run entirely:
- Quote to cash. When a sponsor signs a deal, our agent 10K flips the deal to Closed Won in Salesforce, adds missing contacts, sends the bill.com invoice, and runs collections with a 7-day escalation. It took 3 supervised deals before it ran on its own on the 4th. One incorrect invoice since.
- Commissions. 10K proposed doing them itself, and now it does.
- Paid campaigns. 10K builds LinkedIn and X campaigns end to end, including audiences, creative, 4 A/B variants, and retargeting. A human still hits publish.
- Outbound. Vendor AI SDRs handle about 90% of it.
- Our system of record. We canceled Notion after about 7 years because 10K replaced it.
What humans still do: judgment calls, relationships, and checking the agents’ work. A recent 13-hour day with 10K went to checking, re-ranking, and fixing, with no new features built. We spend far less time doing tasks and much more time reviewing them.
If you cut 50% and expect agents to run unsupervised, you’ll be disappointed. Someone has to own what the agents produce. At Atlassian, that’s the managers and engineers who stayed. At SaaStr, it’s me.
#6. How Much Each Function Can Cut
Back office, G&A, and ops: 50% or more. This was nearly all of Cloudflare’s cut. We run all of it with agents. Start here.
Marketing ops and demand gen execution: 30% to 50%. Agents now handle campaign building, list hygiene, creative variants, and reporting. Humans still own strategy and brand.
Engineering at a large, mature org: 10% to 20%, possibly more. Six months ago I would have told you to protect engineering. Atlassian cut 900+ R&D roles and cloud growth went up. Bending Spoons runs mature products with 90%+ AI-assisted PRs. Engineering orgs that grew fast through 2021 usually have room to cut.
Engineering at a small, fast-growing company: keep it. On a 25-person team at a $20M ARR company growing 60%, each engineer is now 2x to 3x more productive. Keep them, and consider hiring more.
Quota-carrying sales at a growing company: leave it alone. Cloudflare spared it explicitly. Atlassian cut elsewhere to invest more in enterprise sales. If the pipeline is real, the reps carrying a number go last.
Customer success: depends on ACV. At low ACVs, agents can carry most of it. At $100K+ ACVs, customers expect a person, and large customers drive 73% of Cloudflare’s revenue.
#7. Revenue-Per-Employee Benchmarks for Your Next Board Meeting
- ICONIQ, typical B2B (State of Software 2025): $237K ARR per FTE
- Cloudflare before the cut: ~$496K (Q1 annualized)
- ICONIQ Pacesetters at $100M+ ARR (2026 index): $655K median
- Cloudflare after the cut: ~$690K (Q2 annualized)
- Bending Spoons 2025: $2.57M
- SaaStr: ~$3M per human
ICONIQ’s 2025 data put the typical B2B company at $237K per employee. Its 2026 Pacesetter Index puts the median at $655K for its fastest-growing companies above $100M ARR. That index comes from ICONIQ’s top-growth portfolio companies, so treat it as the top of the market. Cloudflare’s cut moved it from about 2x the typical company to just above the Pacesetter median in one quarter. If you’re under $300K in late 2026 and growing under 30%, expect your board to ask about it.
