The Tech Deal series - 3. Interview: a PE investor on software and AI acquisitions
Where the series has been: Post #1 set out why a technology transaction behaves differently from an ordinary M&A deal. Post #2 took the LOI apart and discussed what to lock down before the SPA in tech M&A. This post looks at the same transaction from the buy side.
Julia Beij is an Investment Manager at Smile Sail, an evergreen private equity (PE) fund that supports ambitious entrepreneurs and management teams in growing into Software & AI Leaders (‘Sail’) out of Amsterdam and Leuven. It invests in companies with EBITDA of EUR 2 million to EUR 10 million (or break even in case of high growth) at entry, with a concentration in financial, healthcare, industrial software. She came into investing from M&A advisory work at ABN AMRO and Lincoln International, by way of Capital A.
This is the first of three interviews in the series, and the only one from the buy side of a private equity deal. The conversation was recorded in Dutch on 11 September 2026 and is published here in English. I am grateful to her for the time she gave it. The questions and the framing around them are mine; the answers are hers.
The question running through all of it is how a private equity investor arrives at a number for a software company, and what moves it afterwards.
The evergreen structure, and why it matters to a seller
An investor’s fund structure, holding period and investment horizon are especially relevant to sellers looking to rollover after the transaction. As it determines how long a seller who rolls over will be asked to stay, and when that seller will be asked to sell again, sellers are wise to include it in their counterparty analysis at the LOI stage.
Bart Stevens: Smile Sail is an evergreen private equity fund, funded by entrepreneurs, founders, operators and captains of industry rather than by institutional parties as for many other PE investors. What does that do to your exit horizon?
Julia Beij: Above all, it gives us flexibility: both in terms of holding period as well as in deal structure, whether majority, minority or co-investment. Like a closed-end fund, we ultimately work towards an exit, but the evergreen structure means we are never forced to sell. We can time the exit for the moment that is best for the company, which aligns us closely with the entrepreneurs and co-shareholders we work with.
Stevens: Does that appeal more to founders and sellers?
Beij: Yes, the message resonates well with founders and management teams. We are a long-term partner and work towards an exit when the timing is right. When we jointly want to do another acquisition close to the regular exit timeline, we still have the flexibility to invest more capital and properly integrate the business. In addition, proceeds from exits are largely recycled into the fund, so we are not tied to fundraising cycles and have full attention to our investments. When regular funds need to raise their successor, it may take considerable partner time and can result in pressure to show realised returns, which may unwillingly bring an exit forward. That pressure does not apply to us.
Stevens: Smile Sail focuses on software and AI; within that scope, do you break down further into sectors or specialisations?
Beij: B2B software, AI and IT services is already a focused mandate, so in principle we look at everything within it. Based on our network and experience, we do have three verticals we focus on: financial, healthcare and industrial software. For example, in healthcare and life sciences, several people in our wider team have a strong network, which helps us understand a company faster and add more value.
AI defensibility: a new valuation axis
Start where the market moved. Listed software valuations in the United States fell hard in the first quarter of 2026, on the fear that AI agents would break per-seat pricing. The SEG SaaS Index, which tracks 106 mostly US-listed B2B software companies, put the median enterprise value at 4.9x trailing revenue at the close of 2025 and 3.7x three months later. Much of that ground has been recovered since the summer.
Stevens: What impact did that wave have for you, and how you look at targets in the Dutch midmarket?
Beij: AI is developing fast, and that creates uncertainty. The private market in the Benelux was affected too, but to a lesser extent and with some delay. Buyers and sellers have both become more cautious, because nobody knows exactly what agentic AI will mean for different business models.
What has changed is that quality matters even more, and companies are now assessed along a new axis: AI defensibility, meaning how resilient the business model is to AI. The question is what makes a target genuinely unique. If the differentiator is the software code itself, AI agents can replicate it relatively easily. The same applies to horizontal, generic point solutions that do the same job in every sector, such as a CRM or project management tool. A vertical proposition is a different matter: software built for one sector, with the compliance, workflows and regulation of that sector built in, deeply integrated with customers and with high switching costs. For an AI agent the bar to replace that is much higher, which makes it far more interesting to us. That was always true to some extent, but the bar has risen considerably over the past year.
Stevens: So the specialisation of the software determines how resilient it is to replication by AI agents.
Beij: Yes. A company needs a moat against AI which partly lies in vertical specialisation. The premium lies in how well a company puts AI to use. If AI-built modules on top of the product create more value for customers, and the company can charge for that through higher revenue per customer, that earns a premium in the valuation.
Stevens: Given the speed at which AI is developing, you must revisit that analysis far more often than before. How do you deal with that?
Beij: It is a weekly discussion within the team and the investment committee. We look closely at the potential impact of AI and at whether we can help a company get ahead of it. Targets that have clearly not yet considered the impact, or started investing too late, will feel that in the valuation. Entrepreneurs who have a plan, even if they are not there yet, get more leeway.
We also look at the wider competitive field: how well capitalised other players are and how AI will affect them. If well-capitalised customers, suppliers or competitors are already a step ahead with AI, that makes it difficult. If they are not, and the company can gain a competitive advantage through AI itself, that is more interesting. AI is now just as much a standard part of our initial assessment as growth and margin.
Stevens: When does that tip from a lower valuation into simply not interesting at all?
Beij: When we think the business model will not exist in a few years, or will look completely different, it is no longer something you price in. The risk profile is fundamentally different, so we do not invest. In practice that mainly applies to the more generic solutions.
What drives the multiple
AI defensibility is the newest layer of an appraisal that existed before it. Underneath sit the metrics that have always set a software multiple.
Stevens: Take two software companies with the same revenue and different multiples. Why is one worth 8x and the other 4x? Do you have a philosophy on that, or is it case by case?
Beij: A multiple reflects the overall attractiveness of a company. Beyond the level of revenue, we look at its quality: growth, retention, margin and predictability, for example through long-term customer contracts. The quality of the management team and a clear plan for the next phase of growth matter as well.
Then there is the market or niche the company operates in, and its position within it. A clear market leader in an attractive, growing niche commands a higher multiple than one of five comparable players in a niche that is barely growing, or expected to shrink.
Stevens: Which metric tends to be decisive in practice?
Beij: The combination of growth and the margin that can ultimately be achieved. We are comfortable investing in a fast-growing company with limited margin today, as long as we can see a strong earnings model within a few years.
Buying platforms vs. add-ons
Whether a target is a platform or an add-on sets the entry multiple, and it also sets how important it is to the buyer that the management team stays on after closing.
Stevens: What kind of companies, at what stage, do you invest in? And how do you determine whether you see a target as a platform, or more as an add-on to an existing platform?
Beij: We invest solely in B2B software, AI and IT services companies. They need to be profitable, or have clear visibility on profitability in the short term. Typically that means EBITDA of EUR 2 million to EUR 10 million at entry, with equity tickets of EUR 10 million to EUR 50 million. High-growth businesses with recurring SaaS revenue are the exception: they can be interesting from around EUR 4 to 5 million in revenue, provided they grow by more than 30 to 40 per cent a year and sit in an attractive niche.
A platform must be a good investment in its own right: a strong management team, a scalable business model, and a good return even without further add-ons. For an add-on we mainly look at the fit with the platform: synergies, how the integration will work, what it will deliver, and whether the deal can be done at a lower multiple than the platform.
Stevens: Is it mainly the management, or the technical scalability, that decides which of the two a company is?
Beij: Both. Management is important, because we do not usually acquire 100 per cent and run the company ourselves; it is a partnership with the existing management team. The team may need strengthening, and we often add a CFO or support on the technical or commercial side when we come in. That is fine, but there has to be a clear leader with a vision for the market and a plan for the company.
For a platform that is crucial. For an add-on that is fully integrated into the existing business, it matters less if the entrepreneur plans to retire after a proper handover in two years. As for scalability: we are a growth investor, so we need confidence that organic or inorganic growth is possible.
Due diligence
A software diligence usually runs across the accounts, the IP, the data and the customer contracts.
Stevens: Let’s say you have read a compelling information memorandum and you start looking under the hood. Which due diligence findings are the ones where the seller is most surprised that you found them?
Beij: Mostly at smaller companies: they often report on a cash basis, booking revenue when it is invoiced or received rather than spreading it over the service period. Once we restate that on an accrual basis, revenue and EBITDA can look materially different. The same goes for working capital: normalisations for seasonality or bad debts that an experienced CFO would have made upfront are often missing. That surprises entrepreneurs, but it simply reflects a less professionalised finance function, often without an experienced sell-side adviser.
Stevens: And from a legal perspective, are there typical things you come across at software companies?
Beij: At software companies we look specifically at IP ownership, data protection, cybersecurity and tax matters such as R&D credits, because they go straight to the core value of the business. How we deal with a finding depends on whether the risk is known. A quantifiable issue goes into the price; a known but uncertain risk into a specific indemnity. For unknown risks we often use W&I insurance, which never covers what you have already found in diligence. So far none of these areas has been a dealbreaker for me, precisely because they can be covered this way.
Stevens: When is a finding simply a downside you price in, and when does it make a company uninvestable?
Beij: It comes down to whether the issue is measurable and quantifiable. One-off costs, net working capital, normalisations: those we can build into the price. It is different when something undermines trust, because sellers have not been transparent, or when an issue changes the core value of the company. Say customer concentration was not disclosed, and revenue turns out to depend heavily on one customer without a strong contract behind it. That changes the whole investment thesis, and we would rather walk away.
Stevens: And after you have signed an LOI, are there things that come up in diligence you really had not seen coming?
Beij: Occasionally. No company is perfectly organised when we invest, and much of that you address together after closing. But I have seen contract pricing terms emerge that could cause a major commercial shift and have a real impact on the business.
Earn-out or rollover
Both earn-out and rollover are used to bridge a valuation gap. Each comes with different strings attached.
Stevens: If you have a valuation gap to bridge, when do you go for an earn-out and when for a rollover?
Beij: If the gap is measurable, for example because we take a different view on EBITDA or ARR growth, you can document and measure it, and an earn-out can bridge it. If it is more about keeping a seller engaged in the success of the business over the coming years, we prefer a rollover.
Stevens: In my own practice earn-outs are a frequent source of conflict. You get an interim period where two businesses run alongside each other, the books cannot be integrated but operations quietly are, and views on how the collaboration is going may start to diverge over time. How much management freedom do you insist on upfront, and how much do you leave with the seller?
Beij: You never want an earn-out to stand in the way of long-term success. Here a platform differs from an add-on. A platform often does not need to be integrated, so combining the financials matters less. With an add-on, the question is whether an earn-out is workable at all, and what you would measure. If you integrate the add-on from day one for the synergies, so that revenue, staff and profits immediately merge, measuring on EBITDA is no longer realistic. If the add-on stays a separate label within the group for a certain period of time, adding to the overall proposition and enabling cross-selling while operating fairly independently, you can attach the earn-out to specific metrics such as ARR growth. Best practices include limiting the earn-out in time, i.e. one fiscal year, and using multiple metrics, i.e. a combination of topline and bottom-line performance.
The art is to agree clearly upfront what does and does not fall under it. We typically commit not to change pricing, the sales organisation or accounting policies during the earn-out period. At the same time, you want the flexibility to make further acquisitions or integrations. And you want to avoid key people spending their days managing their earn-out rather than building the long-term success of the business.
Stevens: I assume you also agree what falls inside the metric.
Beij: Yes: what falls under EBITDA, and what is and is not in the business plan. If we jointly make investments outside the business plan, they do not count towards the earn-out. And if we use the company’s cash to fund another acquisition, we add that cash back when we calculate net debt for the earn-out or vendor loan adjustment, so the seller’s equity value is not reduced by a decision we took jointly.
Stevens: Those are essentially your normalisations.
Beij: Exactly: items you fix in advance so that decisions like these have no impact. If the earn-out is missed because the forecast was too optimistic, that is a different matter.
Chosen without a track record
The counterparty question returns here from the other side: what a seller chooses when there is no track record to weigh.
Stevens: Which investment or transaction in your career are you proudest of, and why that one?
Beij: RegLab. Despite strong competition in the process, the two founders chose Smile Sail as their partner on the basis of personal fit and expertise. I see that as a great compliment, because the fund hadn’t made another investment yet: we had no portfolio and no joint team track record yet. What helped was our specialisation in software and our track record of building international market leaders, which made us a real sparring partner from the first conversation rather than just a provider of capital. That kind of trust you do not get from a track record, you earn it in the process.
Stevens: How did you earn that trust in the process?
Beij: We came to the table with a lot of relevant knowledge, including KYC software experience from our wider network. That allowed us to show straight away that we understood the market, and to move quickly to growth opportunities and how to take the company further, rather than spending a lot of time on how the business works. You always spend some time on that, but sector knowledge puts you a step ahead.
Next up
The valuation frame here runs straight into the next two posts. Post #4 takes the metric a software company is typically valued on, ARR, and focuses on how its definition is negotiated. Post #5 turns to the equity bridge and what happens to deferred revenue in it.
