High NRR is Magic. It Can Also Hide a Lot of Problems.

Having high net revenue / dollar retention is the magic of B2B. If you have 120% NRR, you double in less than 5 years, even with no new customers. Above that, you really get into magical compounding growth.

But here’s the 2026 reality check: high NRR has gotten rarer than it used to be. Median NRR for public B2B companies is now around 108-110%. HubSpot is flat at 103%. Zoom enterprise is at 98%, meaning customers are literally shrinking their spending. The companies still hitting 120%+ NRR are almost all AI-natives or have repositioned hard into AI budget. Classic B2B benchmarks have collapsed by 10-20 points across the board.

So if you still have high NRR in 2026, that’s even more magical than it used to be. But also even more dangerous, because the masking effect is amplified.

1. High NRR can mask mediocre growth in new logos and customers. This is issue #1.

You need to be careful in a high NRR environment that customer count is still growing fast enough. High NRR can hide it. With 150% NRR, you can hide just 20-30% new customer growth and it will be hard to see that slow growth in new customers in the revenue growth numbers.

Look at PagerDuty. At IPO in 2019, they had 139% NRR. Truly elite. In 2022, still at 124-126%. The expansion machine was working beautifully. But underneath, customer count growth was slowing. Then slowing more. Today, PagerDuty has roughly the same number of customers it had three years ago. NRR finally collapsed to 100%. And the stock is down 75%+ from its highs. Worth about $1B on $500M in ARR. Just 2x ARR. The high NRR masked the flat customer count for years. When the NRR magic finally ran out, there was no new logo engine left to fall back on. They had been harvesting, not building. They just couldn’t see it because the NRR number looked great.

The 2026 version of this problem is even more brutal because of seat compression. Workday CEO Carl Eschenbach said it directly: “We are seeing customers committing to lower headcount levels on renewals.” Even great companies are watching their per-customer revenue shrink as AI compresses headcount. If you’re not aggressively replacing the lost seat revenue with new logos, you’re harvesting, not building.

2. High NRR can mask logo churn.

I also see this. In high NRR environments, there isn’t enough talk about logo churn. Losing a few $10k customers seems OK if you upsold $100k. Again, the high NRR can mask the customer losses. But you’re losing your future if you lose more customers than you need to.

3. High NRR can mask a lot of mediocre customer success reps and forward deployed engineers

I know this is a bit controversial, but it’s true. In low NRR environments, everyone can see the customer success folks that are underperforming. Like you can see it in sales. But when the product organically has 120%+ NRR, everyone sort of gets a pass. And in 2026, with so many CS organizations now reporting to the CRO and operating as renewals desks, the underperformance is even harder to spot. The high NRR is doing all the work, not the team.

4. High NRR can hide the fact you’re falling behind in AI.

This is the new one for 2026, and the team can hide a lot here. Existing customers expand for all kinds of reasons that have nothing to do with whether your product is AI-competitive. Price increases. Adding seats. Buying the new edition because it’s bundled into the renewal. Multi-year deals with built-in escalators. All of that drives NRR. None of it tells you whether a buyer evaluating your category fresh in 2026 would still pick you over an AI-native alternative.

The leading indicator that you’re falling behind in AI isn’t NRR. It’s new logo win rates against AI-native competitors. It’s whether your demos still land. It’s whether your product roadmap has actual agentic capabilities or just “AI features.” The team can point to 125% NRR and say everything is fine. Meanwhile, the new logo number is quietly dying because every fresh buyer is choosing the AI-native option. By the time the NRR finally cracks, you’ve already lost the next two years of growth. Watch new logo win rates by competitor. That’s the AI tell.

So just a few thoughts on what to do about it:

**First, strive for new logo growth of at least 50% of your total revenue growth. Palantir is doing this.**If you’re growing 100% a year, strive for at least 50% growth in new customers and revenue from new accounts. Once the ratio starts to tilt past 2:1 of your new bookings from existing customers, you probably aren’t adding enough new customers.**Second, measure logo retention as often as you do NRR.**Align them and track them together. Strive for 90% logo retention in general wherever you have 120%+ NRR. And watch the GRR/NRR gap. If your GRR is 92% and NRR is 95%, the gap tells you exactly how thin the magic actually is.**Third, remember to segment your NRR and logo retention by deal size and motion.**This will expose trends you don’t otherwise see, especially in your smaller customers. And watch direct sales separately from self-serve. MongoDB’s total customer count grew 20% YoY into 2026, but their Direct Sales Customers (the enterprise field sales accounts) went from 7,500 in early 2025 to 7,000 by late 2025. A 7% decline in the core enterprise business hidden by self-serve growth and consumption revenue. The aggregate often hides the rot.**And finally, be honest if net new customer count ever drops below 20% a year.**That’s a warning sign growth will stall in the future. Even if upsells and other high NRR attributes are masking it. I see too many founders not sounding the warning bell when net new customer count falls below +20% a year.

**If you have 120%, even 130%+ NRR in 2026, that’s magic. Real magic. Do whatever it takes to keep it and even drive it up further. **

The bar has gotten higher, and most of your competitors no longer have it.

Just remember as founders, it can also mask a lot of issues. Issues that will lead to slower growth than you’d otherwise have. Own those ones, because otherwise, the team has too much on its hands to do it themselves.

Hidden issues image from here