The Quiet Shrink: Three Years of B2B SaaS Workforce Data

We tracked roughly 200 B2B SaaS companies over 36 months. What we found wasn’t a single dramatic event. It was a slow, steady contraction that nobody talks about because it doesn’t come with a press release.

The story people tell about SaaS right now goes something like this: AI is replacing jobs, companies are doing more with less, and the ones who adopt early win. It is a tidy story. The data we collected tells a messier one.

We pulled monthly headcount, department breakdowns, hiring activity, departure counts, and seniority data from multiple LinkedIn-derived intelligence platforms. The sample covers US-based B2B SaaS companies with fewer than 300 employees, observed from March 2023 through February 2026.

Our hypothesis going in: small SaaS companies are structurally shrinking, not through dramatic layoffs but through attrition that never gets backfilled. And the functions disappearing first are not the ones building the product. They are the ones selling it.

The data mostly confirmed this. But a few things surprised us.

01Hiring Has Collapsed

The first chart shows active job postings indexed to their peak in May 2023. Every bar represents one month.

By December 2025, postings had fallen to about a third of their mid-2023 levels. There was no single quarter where hiring “froze.” It just kept declining, month after month, with a brief seasonal bump each January that never recovered the prior year's losses.

The second chart breaks this into two signals. The dark line shows the share of companies with at least one active job posting. The dashed line shows average postings per company. Both declined in tandem, which rules out the possibility that a few large companies were distorting the aggregate. This was broad-based.

At the May 2023 peak, two-thirds of companies were actively posting roles. By December 2025, that number was closer to one-third. The average number of postings per company dropped from roughly 7 to under 2.

This is not a pause. Pauses recover. What we see here is a structural shift in how many people these companies think they need.

02Sales and Marketing Are Disappearing Fastest

If headcount is going down everywhere, where is it going down the most? We tracked three departments: engineering (labeled “technical” in LinkedIn-based classification systems), sales, and marketing.

Engineering lost about 14% of its headcount over the 36-month window. Sales lost 22%. Marketing lost 21%. For every engineer who disappeared, nearly two salespeople or marketers left with them.

The sequencing is worth paying attention to. Marketing started declining in late 2023, before most of the AI adoption wave. Sales followed about two quarters later. Engineering was still growing through October 2023 and only turned negative in mid-to-late 2024.

What this suggests: sales and marketing are the departments getting the biggest productivity lift from AI tooling. Outbound sequencing, lead scoring, content generation, ad optimization, and pipeline management have all been transformed by AI products in the last 18 months. A three-person marketing team with the right AI stack can produce what a seven-person team produced in 2022. The same is true for sales development. Our research on lean sales orgs found that companies with fewer than 2.5% of headcount in sales generated 2.8x more revenue per employee than traditionally staffed companies. The data here supports that finding from a different angle: companies are learning that leaner go-to-market teams, augmented by AI, can maintain or improve output.

Engineering is declining too, but slower. Writing code has AI assistance now, but the complexity of production systems and infrastructure still requires human judgment in ways that outbound emails and campaign copy do not. The next frontier is not just automating volume but training AI to replicate a specific rep's voice, which further reduces the need for large outbound teams.

03People Leave. Nobody Replaces Them.

We tracked monthly departures (employees whose LinkedIn profiles moved away from a company in a given month) alongside active job postings. The dark bars are departures. The line is postings.

Through 2023 and most of 2024, departures held steady at roughly 1% to 1.8% of total headcount per month across the full sample. Normal attrition. People changing jobs, moving on, getting better offers. The rate bounced around, but it did not trend.

Then it dropped. From August 2025 through February 2026, the monthly departure rate fell to about 0.5% to 0.7%, its lowest point in the entire observation window.

Why did departures drop so much at the end? Two reasons, and neither one is good news.

First, the base is smaller. If your company went from 80 people to 60 over two years, your monthly departure count will be lower even if the rate stays the same. Fewer people means fewer exits.

Second, the job market cooled. Carta's H1 2025 compensation report found that voluntary departures across startups are down 47% from their April 2022 peak. Changing jobs became rarer across the entire startup world, not just in our sample. People are staying put, not because they are happier, but because there are fewer places to go. The external labor market is not offering the pull that it did in 2021 and 2022.

The more important signal in this chart is the gap between the dark bars and the line. Job postings dropped faster than departures for nearly the entire window. Each month, more people walked out than roles opened up. That is the mechanism behind the quiet shrink: no layoff announcements, no restructuring memos. Just a hiring bar that keeps rising while attrition does what it always does.

04The Org Gets Top-Heavy

When headcount declines, who actually leaves? The data shows it is disproportionately the people at the bottom.

We tracked the ratio of individual contributors (senior-level and entry-level employees) to leadership roles (directors, VPs, C-suite, founders, and partners) over the full 36 months. The chart uses a 3-month rolling average to smooth out noise from small-company variability.

Through mid-2024, both ratios were rising. Companies were adding ICs faster than they added leaders. The org was getting flatter. Then the trend reversed. By November 2025, the IC-to-leadership ratio had dropped about 4% from its mid-2024 peak. The IC-to-manager ratio followed the same arc, down about 1.5%.

In absolute terms: IC headcount fell 17%. Leadership fell 15%. Managers fell 16%. Everyone is shrinking, but the bottom shrinks fastest. This pattern is consistent with research on attrition-based downsizing. When companies reduce headcount through natural attrition rather than targeted layoffs, senior employees are insulated. Junior roles are the ones that go unfilled when someone leaves, partly because junior employees are more mobile and partly because their positions are easier to argue against replacing.

The result is an org with more overhead per unit of output. Senior people cost more. They also tend to spend more time in meetings, on strategy, and on coordination. If the IC layer that actually writes code, creates campaigns, and closes deals keeps thinning out, the company's ability to execute shrinks faster than its headcount suggests.

Whether this is a problem depends on context. A company that has replaced IC work with AI tooling can operate with a higher ratio of senior to junior staff. A company that just lost its junior people and did not automate anything has a bloated org chart and no one to do the work.

05What This Means for SaaS in 2026

Four data points, one conclusion: small B2B SaaS companies are in a slow structural contraction.

Hiring is down by two-thirds from its 2023 peak. Sales and marketing are being cut at twice the rate of engineering. People are leaving through normal attrition and not getting replaced. And the org is getting top-heavy as the junior layer thins out.

This is not purely the AI replacement story. AI is making some of this possible (smaller teams that produce the same output), and the revenue data supports it. But the hiring decline also reflects tighter capital, slower revenue growth, and the correction from a 2021-2022 hiring boom that was never sustainable. The funding velocity data tells the other half of the story: fewer seed companies are reaching Series A, and the ones that do are sharply category-specific, with generalist software raising at 7x their current revenue — less runway for the kind of hiring that defined 2021.

The question for 2026 is whether this contraction stabilizes or accelerates. If hiring continues declining at the current pace, these companies will be 30-40% smaller than their 2023 selves within another year. At some point, that crosses a line from “efficient” to “too small to compete.”

We will be tracking this quarterly. The next update in this series will break down which specific roles companies stopped filling, and whether the companies that invested in AI tooling are actually performing differently from those that did not.

06Implications

The data points to concrete takeaways for operators, investors, and anyone building inside a SaaS company right now.