There’s one metric that tells you more about the health of a B2B company than revenue growth, NRR, or margins. And many founders hide from it when things decelerate.
The #1 KPI of all in B2B is Net New Customer Growth.
Not revenue. Not ARR. Not net retention. The raw count of new logos you’re adding every quarter.
Revenue can grow 25% while your business is quietly dying. NRR can mask a broken sales engine for years. But net new customer growth doesn’t lie. It tells you whether the market still wants what you’re selling.
Almost every public B2B leader is decelerating here. As are many unicorns. The few that are accelerating? They stand out immediately. And they almost all have one thing in common, which we’ll get to. The ones hiding problems behind headline numbers? The customer count data exposes them every time.
The Companies That Report It. And the Ones That Don’t
First, an observation: only about half of major public B2B companies even disclose quarterly customer counts. Salesforce doesn’t. Workday says “11,500 customers” but doesn’t give you a quarterly progression. CrowdStrike stopped reporting after January 2023.
The ones that do report it? That data is a goldmine.
We tracked ten companies: HubSpot, Atlassian, Datadog, Monday.com, Samsara, MongoDB, Cloudflare, Palantir, ServiceNow, Twilio, and Snowflake. Every one tells a different story — and net new customer growth is what reveals it.
The Decelerators: Most of Application SaaS
Here’s what the trendlines look like for the application SaaS layer:
- HubSpot: Customer count growth went from 23% YoY to 16% over two years. The last five quarters: 21 → 19 → 18 → 17 → 16. A steady grind. Still adding ~10,000 net new paying customers per quarter at 288,706 total and $3.1B in revenue. But the growth rate is grinding down. The question is whether 16% keeps declining into single digits — because at that point, you’re relying almost entirely on NRR to grow.
- Atlassian: The >$10K Cloud ARR cohort over the last seven quarters: 17 → 15 → 14 → 13 → 13 → 12 → 10. A clean grind down to 10% YoY. Net new adds in the most recent quarter were only ~544, versus 2,352 the prior quarter. That’s a 77% drop in net new logo velocity in a single quarter. The high end is doing better — Atlassian has 600+ customers >$1M ARR, up ~40% YoY — but new customer acquisition is clearly broken.
- Monday.com (10+ user accounts): From 18% YoY down to 7-8%. If you stopped here, you’d think the business was in trouble.
- Datadog: Total customer growth steady at 9-10% YoY. Not accelerating, not really decelerating. But 9% customer growth driving 28% revenue growth tells you the expansion engine is doing most of the work.
So four different flavors of deceleration in the application SaaS layer. What separates healthy from concerning?
The Answer Is Always One Layer Deeper
The total customer number is the starting point. But the real signal is in the customer cohort data — the $100K+, $500K+, and $1M+ tiers.
- Monday.com’s $100K+ ARR customers are growing 45-48% YoY. Their $500K+ customers are up 74% YoY. CRM crossed $100M ARR in under two years. They’re deliberately trading volume for deal size. The total customer deceleration is the plan, not a problem. But without tracking both metrics, you’d completely misread what’s happening.
- Datadog’s $100K+ ARR customer cohort is re-accelerating, from a 12% trough in Q3 2024 back up to 19% YoY in Q4 2025. They now have 4,310 customers spending over $100K, generating 90% of total ARR. The overall customer growth is modest, but the quality of customer growth is improving.
- HubSpot doesn’t break out a $100K+ cohort the same way, which makes it harder to read. Average subscription revenue per customer is flat to slightly up (+3% in Q4 2025). They’re adding lots of customers, but they’re not obviously moving upmarket the way Monday.com and Datadog are. At 16% and declining, the question mark is real.
- Atlassian has 600+ customers >$1M ARR, up ~40% YoY. So the high end is firing while logo growth craters. That’s the textbook “harvest the installed base” pattern. It works for a few quarters. It’s not a long-term strategy.
The Accelerators: Welcome to the AI Infrastructure Layer
Now for the companies that break the pattern. There are four — and they have something fundamental in common.
Cloudflare: This is one of the most impressive customer growth trajectories in B2B right now. Total paying customers went from ~190,000 at the end of 2023 to ~332,000 at the end of 2025. The YoY growth rate didn’t just hold — it accelerated. From ~22% in early 2024 to 27% at mid-year to 33% in Q3 to 40% YoY in Q4 2025. They added 37,000 net new paying customers in Q4 alone — a record.
The high-value tiers are keeping pace. $100K+ customers grew 23% to 4,298. $1M+ customers grew 55% YoY to 269 — they added 96 million-dollar customers in a single year. Revenue is accelerating too (27% → 34% YoY). NRR ticked up to 120%.
When total customer count accelerates, high-value customer count accelerates, AND revenue growth accelerates — all at the same time — that’s a company gaining market share across every segment.
Palantir: Customer count growth has been steadily climbing: 35% YoY at end of 2023, to 39-41% through mid-2024, to 43% at end of 2024, to 45% in Q3 2025. US commercial customer count is growing even faster — 83% YoY in mid-2024, still running at 49%+ in late 2025.
Revenue growth followed: 20% → 30% → 36% → 39% → 48% → 63%. That’s not a normal trajectory for a company at $4B+ in revenue. US commercial revenue hit 121% YoY growth in Q3 2025.
Palantir’s AIP (AI Platform) bootcamp model is a genuinely differentiated go-to-market motion that’s accelerating new customer acquisition, not just expanding existing accounts.
Twilio: This is the wildest reacceleration story in B2B right now. Customer count growth went from +7% YoY in Q1 2025 to ~+42% YoY in Q1 2026. That’s a 35-point acceleration in a single year at a 19-year-old company doing $4B+ in revenue. They added 43,000 net new accounts in Q1 2026 alone — more than the company’s full-year net adds in 2023.
A fair caveat: Twilio is killing the Active Customer Accounts metric this quarter, explicitly because they think it’s “less informative.” Many of the new accounts are AI startups spending the minimum and don’t materially move revenue yet. So the +42% is partly inflated by long-tail signups. But every Bland.ai-style company, every voice AI startup, every Cursor-style agent app needs a Twilio account. The flood of small accounts today is the pipeline for material accounts in 2027. DBNER is back to 109%.
Snowflake: Quarterly net-new customer adds have grown 40% YoY in Q4 FY26, up from 21% in Q2. NRR running at 125%. The data warehouse is the substrate for every enterprise AI initiative — every RAG pipeline, every fine-tune, every agent that needs structured data lives there.
The Bifurcation Thesis, In One Table
Look at what happens when you sort the cohort by customer count growth direction:
The companies selling to AI builders — communications APIs, edge networks, data warehouses, AI platforms — are picking up tens of thousands of new customers because every AI-native startup needs their infrastructure. The companies selling per-seat application software to humans are seeing logo growth slow even as they extract more from existing accounts.
That’s not a one-quarter accident. It’s a structural divergence that’s been building for 2+ years and is getting starker.
Revenue acceleration looks the same on a chart whether it comes from new customers or installed-base extraction. Customer count growth tells you which is which. The answer is uncomfortable for the application SaaS side of the cohort.
Samsara: How to Manage the Transition
If you want a model for how to manage the transition from hyper-growth to durable growth without losing the customer engine, look at Samsara.
FY2026 (ending January 2026): $100K+ ARR customers hit 3,194, up 37% YoY. They added 204 new $100K+ customers in Q4 alone. $1M+ ARR customers grew 56% YoY with a record 131 deals at that tier.
Revenue growth is decelerating (43% two years ago, 28-30% now), but high-value new customer acquisition isn’t. $100K+ customers now drive 61% of total ARR, up from 56% two years ago. GAAP profitable. $1.9B in ARR.
When high-value customer count decelerates slower than revenue — or stays flat while revenue decelerates — the engine is still working. The business is getting more efficient, not running out of market.
MongoDB: Strong Growth, But a Yello Flag Hiding in Plain Sight
Now for the cautionary tale. This is why net new customer growth matters more than any other metric.
MongoDB’s total customer count looks fine: 65,200+ as of January 2026, growing 20% YoY. Revenue up 27%. $100K+ customers up to 2,799. All seems well on the surface.
But look at their Direct Sales Customers — the enterprise accounts sold through their field sales team:
- Apr 2024: 7,100
- Jul 2024: 7,300
- Oct 2024: 7,400
- Jan 2025: 7,500
- Apr 2025: 7,500 ← Flatline
- Jul 2025: 7,300 ← Declining
- Oct 2025: 7,000 ← Declining faster
The enterprise sales engine is contracting. They’ve lost 500 direct sales customers — a 7% decline — while total customers grew 20%. Self-serve Atlas consumption is masking a real structural problem in the field sales motion.
The headline numbers say growth. The customer count data says the core enterprise business is eroding. NRR and consumption revenue are papering over it. For now.
ServiceNow: What “At Scale” Looks Like
ServiceNow doesn’t disclose total customer counts in a useful way, but they report customers with $1M+ in ACV. That number grew from 1,922 to 2,109 over calendar 2024 — a steady 12% YoY.
Not flashy. But the average ACV in that cohort expanded from $4.6M to $5M+. At $13B+ in subscription revenue, growth comes from getting bigger within the enterprise, not from adding net new logos.
At ServiceNow’s scale, that’s the right model. But for anyone under $5B in revenue, you can’t coast on expansion alone. You’re not ServiceNow yet.
The Framework: How to Read the Signals
After pulling all this data, the patterns sort into five buckets:
Accelerating New New Customer Count (AI infrastructure tailwind): Total customer count growth AND high-value cohort growth are both increasing. Revenue growth follows. Cloudflare, Palantir, Twilio, and Snowflake are here. Three of the four are infrastructure plays that benefit from AI-native startup formation. This isn’t an accident — it’s the dominant tailwind in B2B right now.
Healthy deceleration: Total customer count growth is slowing, but high-value customer cohorts ($100K+, $500K+) are growing faster than total revenue. The business is moving upmarket and getting more efficient. Samsara is doing this well.
Neutral deceleration: Total customer count growth is flat or slightly declining, but the expansion engine (NRR) is strong enough to keep revenue growing at 20%+. Datadog and HubSpot are here. Not alarming, but worth watching.
Application SaaS deceleration: Customer count is grinding down even as the high-end cohort grows. Revenue is being driven by bigger deals, pricing changes, and existing-customer expansion — not net-new logos. Atlassian is the cleanest example. Works for a few quarters; not a strategy.
Concerning deceleration: Total customer count growth looks fine on the surface, but the enterprise/high-value customer count is flat or declining. MongoDB’s Direct Sales Customer erosion from 7,500 to 7,000 is the textbook case. This is the one that eventually catches up with you.
The Ratios That Matter
The 2:1 Rule: Revenue growth should be no more than ~2x your customer count growth rate. If you’re at 50% revenue growth with 25% customer growth, healthy. If you’re at 50% revenue growth with 5% customer growth, you’re harvesting.
Look at how the companies stack up:
- Cloudflare: 34% revenue / 40% customer growth = 0.85x. Customer growth outpacing revenue. Incredible.
- Twilio: ~13% revenue / 42% customer growth = 0.3x. Customer growth far outpacing revenue. Pipeline forming.
- Palantir: 63% revenue / 45% customer growth = 1.4x. Both firing.
- Samsara: 30% revenue / 37% $100K+ customer growth = 0.8x. Healthy.
- HubSpot: 19% revenue / 16% customer growth = 1.2x. Fine, but both decelerating.
- Atlassian: ~22% revenue / 10% customer growth = 2.2x. Approaching the harvest line.
- Datadog: 28% revenue / 9% customer growth = 3.1x. Expansion-heavy. Watch the ratio.
- MongoDB: 27% revenue / -7% enterprise customer growth = broken ratio. Warning sign.
The companies on the right side of the AI bifurcation (Cloudflare, Twilio) have ratios below 1.0 — meaning customer growth is running ahead of revenue growth. That’s the math of a company building inventory for the next two years of expansion. Twilio’s small-account flood doesn’t move revenue today, but every one of those accounts is a Bland.ai or Cursor or some AI-native company that may be a $100K+ customer in 18 months.
The companies on the wrong side (Atlassian, MongoDB enterprise) have ratios going the other direction. Revenue is being squeezed out of a flat or declining customer base. Math eventually catches up.
What To Do If Your Net New Customer Growth Is Slowing
If you’re a founder reading this and recognizing your own company:
Separate the signal from the noise. Are you losing new customer velocity because you’re deliberately moving upmarket (Monday.com), because the market is moving away from you (MongoDB’s enterprise erosion), or because you’re on the wrong side of the AI infrastructure bifurcation (Atlassian, HubSpot)?
Track multiple customer cohorts. Total customers, $50K+, $100K+, $500K+. If the high-value tiers are growing while the total decelerates, that can be fine. If they’re all declining, that’s a five-alarm fire.
Don’t let NRR be your security blanket. Expansion revenue from existing customers is great, but it’s not a strategy. It will eventually slow. The question is whether you’ve built the new customer engine to sustain growth before it does.
Figure out where you sit on the AI bifurcation. Are you selling infrastructure to AI builders, or per-seat software to humans whose seats AI is eliminating? If the latter, you need a deliberate strategy to either (a) build agentic products that get paid per outcome, or (b) own a category so deeply that human seats keep expanding anyway. There’s no third option that doesn’t end in slow erosion.
Invest in new customer acquisition even when it’s expensive. The temptation at scale is always to double down on existing accounts because the ROI looks better. It does look better — in the short term. But every customer you don’t acquire today is a customer your competitor acquires.
Study the accelerators (of course) and find your AI tailwinds. Cloudflare, Palantir, Twilio, and Snowflake are all accelerating customer growth at $2B+ in revenue. That’s not normal. Cloudflare is doing it with a usage-based developer platform. Palantir is doing it with AIP bootcamps that collapse the sales cycle. Twilio is doing it by being the default communications API for the AI-native cohort. Snowflake is doing it by being the default data substrate for enterprise AI. All four found go-to-market motions that compound. If your customer growth is decelerating, the question isn’t whether to accept it — it’s whether there’s a motion you haven’t tried yet.
Net new customer growth is the single most honest metric in B2B. Almost everyone is decelerating right now. The ones who know why — and have a plan — will be fine. The ones who are hiding behind NRR and consumption revenue are running out of time to figure it out.
And the ones on the right side of the AI bifurcation? They’re not just fine. They’re compounding faster than they ever have.
