A clean benchmark is irresistible. It gives a founder a number to put on a slide, a target to take into planning and a simple answer when an investor asks how the company compares.
That is probably why a claim about European SaaS founders “hitting 28% margins” travels so well. It is also not what the underlying data says.
GP Bullhound’s 2025 European SaaS report separates a survey of more than 100 private European SaaS companies from its public-market index. Around 57% of the private-company respondents were EBITDA-positive. Separately, average EBITDA margins among the listed European SaaS companies had moved into the high-20s, with most index companies above 20%.
Those are different populations. A listed company and a founder-led private business differ in scale, maturity, reporting and capital structure. Treating the public index average as a private-founder benchmark creates precision where none exists.
The correction matters, but the more interesting question comes next: what can a smaller SaaS company actually learn from the numbers?
The data rewards coherence, not one magic ratio#
GP Bullhound’s report shows several encouraging patterns. Profitability among private respondents is becoming more common. Median gross margins across ARR cohorts remained in the mid-70s to low-80s. Companies between €5 million and €50 million in ARR improved revenue per employee. Net revenue retention remained above 100% overall, although enterprise-focused vendors improved while SME-focused and mixed models softened.
These measures describe different parts of the business, but they become useful when read together.
High gross margin creates room to invest. Retention protects the recurring revenue already won. Revenue per employee indicates how much commercial output the organisation supports. Spending discipline determines whether that leverage reaches earnings and cash.
A company can make any one of those figures look better for a while. Freeze hiring and ARR per employee rises, even as support deteriorates. Cut product investment and EBITDA improves, even as future retention weakens. Discount heavily and growth accelerates, while the economics underneath it get worse.
The stronger European SaaS companies in the dataset do not reveal a single trick. They show what happens when several parts of the operating model reinforce one another.
Gross margin leaks through ordinary work#
Gross margin rarely collapses because of one dramatic infrastructure bill. It leaks through decisions that looked harmless in isolation.
A customer receives bespoke onboarding that was never priced. A low-value feature produces a high support burden. A “self-service” product depends on someone manually correcting data every week. An AI feature makes too many calls, carries excessive context or retries work that has little customer value.
The right question is not whether your gross margin matches the median of a broad SaaS cohort. Product categories have different delivery costs. A payments platform, data product and collaboration tool should not have identical economics.
The useful question is where direct delivery cost is growing faster than the value customers receive. Look at infrastructure and model cost by product or segment, support effort by plan, implementation hours and exceptions hidden inside standard pricing. This is where an abstract margin target becomes operating evidence.
McKinsey’s 2026 global survey found that AI operating costs constrained use at about one in five responding organisations. That does not make AI features inherently uneconomic. It means their cost needs to be visible. A cheap pilot can become a material service cost when usage scales.
Retention tells you whether the model works for your customer#
GP Bullhound found improving net revenue retention among enterprise-focused vendors and softer results for SME-focused and mixed models. The lazy conclusion would be “move upmarket.”
That ignores the cost of doing so. Enterprise customers can bring longer sales cycles, procurement, security obligations, implementation work and concentration risk. SME customers demand a different discipline: quick time to value, low service cost and clear self-service boundaries.
The lesson is to build an operating model for the segment you actually serve.
For an SME product, exceptions can quietly destroy the economics. One customer asks for a custom workflow, another needs manual migration help, and a third requires support outside the standard promise. None seems decisive. Together they turn a repeatable product into a services business that still charges software prices.
For an enterprise product, the risk often appears in stakeholder coverage and renewal visibility. The product may be delivering value while the commercial relationship depends on one champion.
Net revenue retention is valuable because it compresses those forces into one signal, but founders should still inspect expansion, contraction, churn and logo retention separately. One large expansion can hide a worrying number of smaller departures.
Revenue per employee is a clue, not a quota#
GP Bullhound reported further productivity improvement in 2025 for European SaaS companies between €5 million and €50 million ARR. Its 2024 report had already found ARR per full-time employee approaching €150,000 for growth-stage companies.
Founders naturally compare that figure with their own. The comparison can be useful if it starts an investigation. It becomes dangerous when it ends one.
A rising ratio may reflect a better product, clearer processes and genuine operating leverage. It may also reflect an understaffed support team, deferred engineering or work that cannot continue at the current pace.
Read it beside recurring revenue growth, gross margin, retention, service quality and the age of important queues. Then look at the work itself. Where are people repeatedly gathering context, reconciling records, preparing the same report or waiting for a founder’s approval?
That is where leverage is usually found: not in demanding more output from each employee, but in removing coordination that should not consume their time.
AI helps only when the process improves#
There is no credible evidence that buying AI automatically improves SaaS margins. McKinsey’s 2026 survey found that 80% of respondents saw individual productivity gains, while 37% attributed any positive EBIT impact to AI. Its small group of AI high performers was more likely to redesign workflows, measure value and show sustained leadership commitment.
For a SaaS team, the order matters. Remove work that no longer supports a customer or decision. Standardise the inputs. Automate stable rules. Use AI where synthesis or bounded judgement improves the result. Put a human decision before consequential action.
Consider a weekly pipeline review. If the team spends hours assembling data from systems that already hold it, retrieval and preparation are good candidates for automation. If the real delay is six days waiting for a commercial decision, generating the report faster will not fix the process.
This is the part Winglo is designed to support: recurring coordination across marketing, SEO, creative, operations and sales, using shared context, visible workflows and approval points. It does not replace the accounting system or create profitable economics by itself. Its role is to reduce the work around the work.
A better monthly conversation#
A founder does not need an elaborate efficiency scorecard. A useful monthly review can begin with four questions:
- Where did gross margin move, and what changed in delivery cost?
- What do expansion, contraction and churn reveal about the customers we serve?
- Is revenue growing faster than the organisation’s workload?
- Which important work is slow because production, review or decision-making is blocked?
Those questions keep the measures separate enough to remain honest while connecting them to operating choices.
The lesson in Europe’s more profitable SaaS companies is not to chase 28%. It is to make growth, retention, delivery cost and team capacity work as one system. That is less memorable than a benchmark. It is also much closer to how durable efficiency is built.
See the AI stack a ten-person European business actually needs for where operational AI belongs among systems of record and automation. If recurring go-to-market work is consuming more coordination than it should, request access to Winglo.
Last reviewed: 9 September 2026.