This article is based on my conversation with Philippe Willi, Group CEO of Tourism Software Group, on the RolyPoly podcast.
Most roll-up stories start with a thesis and a cheque. Philippe Willi's starts with a rafting company in Interlaken, a high school teacher who made an introduction, and a booking tool built in 2007 so guests could reserve canyoning trips online. That tool became TrekkSoft, a Swiss Ltd founded in 2010 with three co-founders working for free and one employee on the payroll. TrekkSoft raised venture capital in 2013, scaled to around 100 people, and then in 2017 cut itself in half.
Two acquisitions made in the growth years, paid for with equity because the bank said no, turned out to be the real business. They became Tourism Software Group (TSG): eight tourism software companies, roughly 230 people across Europe, run on what Philippe calls a Constellation-like playbook and funded from its own cash flow.
What I found valuable in this conversation is that Philippe has run both models inside one company. He knows what the venture playbook does to a founder's decision-making because he followed it for four years, and he knows what a vertical software roll-up looks like when you deliberately choose low debt and slow growth over the equity story. He is unusually candid about the mistakes. Here is what I took from it.
Raising VC Because the Cool Kids Did
TrekkSoft bootstrapped for three years. Philippe's day job was still the adventure company, bookings were growing, and by his own account that was “pretty okay.” Then his co-founder Valentin said something that Philippe now tells as a joke: all the cool kids get venture capital.
He is honest that the reasoning was a little more considered than that. SaaS needs money up front before it breaks even, there was a well-documented playbook promising a faster route, and three first-time founders who had never met an investor assumed that playbook was the only one. They pitched at travel conferences, closed a round in 2013, and the deck said what every deck has to say: a huge addressable market and a plan to own online booking for adventure operators worldwide.
The part worth sitting with is what the equity story did to the company. Philippe says he feels “very guilty” about it. The ambition has to be enormous or venture capital makes no sense, so the founders spent their time thinking about the equity story rather than the business. Between 2013 and 2017 TrekkSoft built a content marketing and self-signup machine targeting small adventure operators, and it ran fine. The churn did not. Small customers signed up and left, and the founders found themselves raising more and more capital to feed a machine that was losing customers out the other side. His phrase for it was “an inefficient machine.”
Would he have got here bootstrapped? He doesn't think so. VC pushed them forward and they would not have reached today's scale without it. But he is equally clear that venture capital was the wrong investor type for TrekkSoft. He simply didn't know that private equity, family offices and business angels existed as alternatives with different expectations. That is a theme I keep hearing on this podcast, and Alex Prokofjev made a similar point about choosing a structure around your own objectives rather than the investor's.
Cutting 100 People to 50, and Telling the Board
In 2017 TrekkSoft went from 100 employees to 50 in a few weeks. Philippe describes it as very painful; many of the people let go were good and working hard, and the ex-employee WhatsApp group still exists today.
I asked what triggered it, because the usual trigger is running out of money, and TrekkSoft still had enough left to do deals. His answer had three parts. The founders had lost confidence they could hit their targets without raising an unrealistic amount of capital. Runway mattered, as it always does when you are burning cash. And they knew going in that the decision would destroy equity value, because they were valued as a hypergrowth startup and were about to tell their investors the strategy didn't make sense.
How did the investors take it? “They 100% did disagree with us.” The board meetings were not fun, and Philippe does not recommend the experience. But the investors backed the team anyway, and his read on why is one I'd pass on to anyone raising: an experienced investor believes in the founder team, not the plan, because they already know the plan will change. It speaks well of TrekkSoft's backers that they behaved that way. It also underlines how important the investor selection is in the first place.
Funding Acquisitions “the Most Stupid Way You Can”
Before the cut, still in the growth years, TrekkSoft made two acquisitions. The equity story called for them, as a complement to organic growth, not the main event. Philippe's description of how they were funded is the most self-deprecating line in the episode.
TrekkSoft was a cash-burning startup. He went to UBS, the bank they had used since day one, to talk about acquisition financing. The bank said no: you are burning cash, quite a bit of it, so we don't lend to you. So they paid with equity, which Philippe calls “the most stupid way you can do acquisitions” because it is the most expensive capital there is. Today he would do it very differently. Back then it was the only option. (If you want the lender's side of that conversation, the Tech Credit Partners episode on debt financing for acquisitions covers why a bank thinks that way.)
The structure was cash plus deferred payments, no rollover equity. The reason is a habit Philippe says he learned over the years: when he talks to a seller, he tries to sit in their chair. From the seller's side, shares in TrekkSoft were a terrible offer. The group was loss-making, cash-burning, and valued at three or four times what it was actually paying for the target. Why would a founder want that paper? So he didn't ask them to take it.
Then the part he calls luck. Both targets were founder-run SMEs with EBITDA and real cash flow, not VC companies. Both founders are still with the group today. And those two cash-generating businesses stabilised a group that was otherwise a turnaround case with overhead on top. The acquisitions that were supposed to complement the startup ended up being the business.
A VC Valuation Is Not a Price
I raised something I'd been chewing on. A friend of mine is looking to acquire a VC-funded business whose valuation is enormous relative to its revenue, and the Bending Spoons headlines are full of companies being bought for a fraction of their last round. My take is that a venture valuation prices in the founders' next four years and a best-case future; it was never a statement about what the business is worth to a buyer today.
Philippe agreed and sharpened it. A venture investor is pricing the terminal value in a discounted cash flow, so the next one to three years barely matter relative to the best-case outcome. On top of that, VC-backed companies come with preferred shares and liquidation preferences, which make the cap table hard to read for an acquirer. TSG looked at VC-backed targets and walked away. The two SMEs they did buy could be valued the way the market values a software SME: five, six, seven times EBITDA, pick your multiple. A business with negative EBITDA and a theoretical DCF value needs a completely different playbook.
My conclusion from the exchange is that we need two different words. A valuation is what someone is willing to pay. If no acquirer will pay the number on the term sheet, it isn't the business's value; it's the price of a specific bet by a specific investor.
Dinner With Mark Leonard
Philippe is open about being a Constellation Software fan. His hobby is comparing serial acquirers (Constellation, Everfield, Chapters Group, Bending Spoons and others), he reads every president's letter, and he still thinks Constellation is the best of them. He wanted to meet Mark Leonard so badly that he tried everything, and it worked: when Leonard was in Switzerland for a conference he came to Philippe's home in Zurich for dinner with Philippe and his wife. Two years later Philippe met him again for breakfast in Toronto.
The takeaway is not a tactic. Philippe had gone to a Constellation shareholder meeting expecting the cliché private equity figure, expensive suit, sharp, maybe a bit Wolf of Wall Street. Leonard was the opposite: down to earth and, in Philippe's words, one of the most thoughtful people he has ever met. His wife, who is not in the business, noticed the same thing from the outside: every answer at dinner was considered before it was given. Philippe's read is that this is a cultural element of Constellation that comes straight from its founder, and I suspect he is right. Culture in a decentralised group is mostly a copy of how the person at the top behaves.
What TSG Centralises, and What It Refuses To
The integration question is the one I most wanted to ask a Constellation-style operator, because the answer is so different from what a private equity buy-and-build would give.
TSG started out extremely decentralised, “like Warren Buffett,” in Philippe's phrase: great entrepreneurs run their businesses, and the group leaves them to it. The target founder is someone who wants to keep running their company, not leave it. Philippe quotes a Constellation executive on why: there is a correlation between how often the group visits a business unit and how unhappy its founder is. As the deal count has grown TSG has introduced a structured process, but the instinct is still to centralise as little as possible.
The exception is in the investment thesis. Every TSG company runs booking or ticketing systems that transact on behalf of tourists, so the group processes around a billion in gross merchandise volume. Pool that and you get a better deal from Stripe or whichever provider you use, and there is margin to be made. Payments are fully centralised. R&D, sales, marketing and support are 100% left in the business units. The CFO office does more than it used to, steering things like which bank to use, but Philippe still describes the group as highly decentralised.
What replaces centralisation is best practice: here is how we think R&D should be done, here is how a successful business unit applied AI, now go and do it your way. And the mechanism for spreading those practices surprised me from a group of software companies. Change only happens face to face. Philippe says it is impossible to drive change over an online meeting. So TSG runs a group retreat for managing directors and business unit management (London in mid-October this year) plus department-specific mini retreats for R&D and for support. The feedback that shaped them: keep it specific. A support person sat in a session on the Rule of 40 learns nothing they can use on Monday.
I asked whether they had been burned into drawing the line there. No. It is cultural, and it is structural: TSG keeps a very lean holding company, so there is simply no machine at group level to force integration. If a business unit refuses to change at all they will get more pushy, but that is the exception.
This is where my own work connects. The groups I see at PMI Stack have the same shape, a thin holdco over companies that still run their own systems, and the thing the holdco cannot delegate is the one consolidated view of how every entity is performing. Which brings us to how TSG measures that. (For a contrasting take on where to draw the centralisation line, Sam Turner's hub and spoke model is worth reading alongside this.)
The Rule of 40 as a Board-Level Traffic Light
For anyone outside software: the Rule of 40 came out of venture capital and says EBITDA margin plus organic growth rate should add up to 40 or more. Twenty percent organic growth plus a 20 percent margin equals 40. Philippe adds one condition that matters a lot: TSG uses an unadjusted EBITDA margin, as close to cash flow as possible, not a creative accounting number.
That is how every business unit is benchmarked. At the board meeting each company is green, yellow or red against the Rule of 40. The goal is all green, which is not always the case, and red is tolerated only briefly. He also hears from M&A advisors that AI is producing companies that consistently hit a Rule of 50, and some of TSG's larger peers have moved their target to 50. TSG is staying at 40 for now, but he follows the debate. (This is a specialised version of what I wrote about in group KPI alignment: the metric only works if every entity reports it the same way.)
On AI more broadly, Philippe's board-level view is narrower than the hype. TSG expects a large efficiency gain and more margin points from AI. What it does not yet see is AI driving revenue. Its markets are red oceans: to win a client, someone else has to lose one, and no AI use case has changed that so far. My own experience from the agency world is slightly different. I haven't seen much efficiency, I've seen people build more stuff in the same hours, including features that would never have cleared an ROI bar before. Both can be true, and I'd expect the gap to show up in margin for TSG either way.
Tier Your Customers Before You Touch Pricing
One thing I took away from the Redeye serial acquirer conference is that pricing is often the biggest lever a software group has, because acquired companies frequently haven't raised prices in years. Philippe credits Mark Maurer, COO of Chapters Group, with the phrase he now uses for the exercise, shared over a beer: price normalisation.
The method is simple. When TSG buys a company it tiers the customers one to four by size, then calculates the margin per tier. Combined with a pricing review, this is classic private equity value creation, and the previous owners almost always underappreciated it. Customer tiering is the one thing Philippe would recommend to every acquirer, because in TSG's case it was an eye-opener again and again.
I asked the obvious fear question: have you ever raised prices and lost more than you gained? The answer reframes the fear. Tiers one and two are where churn must be kept as low as possible and where the red ocean makes TSG very careful. The surprise was at the bottom. In tier four, and sometimes tier three, TSG found it was overspending on small clients by a lot on a fully loaded basis, burning cash to serve them. The fix was not a blunt price rise. They made the payments product mandatory for those tiers and stripped out services, remodelling the offer so the small customer was profitable rather than simply more expensive. Pricing after an acquisition is a product question first.
The Slowest Roll-Up in Europe Sleeps Best
On our prep call Philippe had joked that TSG might be the slowest growing roll-up in Europe. I asked whether that was intentional.
His answer: add quality of sleep to the equation, because there is a correlation between speed, debt levels and how well you sleep. He has run his hobby analysis on the leveraged peers, Bending Spoons, TeamSystem and others, and looked at the debt levels and concluded he would not sleep. TSG keeps its debt low on purpose and funds acquisitions from its own cash flow. He describes the flywheel honestly: generate cash, redeploy it into acquisitions, get better at both. “We're okay. We're not super good.” The balance sheet question, how much debt to put on it, is where they are very, very careful.
I pushed on whether the over-leveraged groups are leveraged for the same reason TrekkSoft chased hypergrowth: they promised an equity story to someone and now have to live up to it. Philippe's answer was more generous than mine. Ambition is a legitimate variable. The public-market equity story, the IPO route that Constellation, Chapters, TeamSystem and others have taken, only works if you are vertical agnostic. TSG chose one vertical, tourism and travel, which is large but still a restriction. And risk appetite differs by who is bearing it. A family office puts maybe five percent of its wealth into direct investments. For Philippe, and for most founders, the group is 100 percent of their net worth. Those two people should not run the same balance sheet.
The line he closed on is the real lesson of the episode. During the venture years TrekkSoft worked on an equity story to make somebody else happy. They stopped. Whatever TSG builds now, the team has to believe in it 100 percent, and it cannot be driven by a third party's perspective. My summary back to him was that it ties to the first question: find the investor who fits your thesis rather than reshaping your thesis to fit an investor, because the alternative is the board meetings he described. His reply: you'll sleep much better. It might take longer, but there is only so much you can buy with money.
Who does he want on the podcast next? Mark Leonard. I'll see what I can do.
.png)
