#246 Mark Leonard’s Shareholder Letters
What this covers
David Senra works through Mark Leonard's Constellation Software shareholder letters, pairing passages with outside precedent. The main precedent is Illinois Tool Works, which grew from $369 million to $4.2 billion in revenue in about 14 years under John Nichols. Nichols split larger acquisitions into smaller units until ITW had 365 by 1996. Leonard's letters supply the reasoning. As businesses grow, rules multiply and coordination effort rises faster than headcount, which pulls talented people away from customers and toward internal control. Constellation's units average 44 employees, and a regression Leonard ran found size and performance almost unrelated.
The discussion then moves into finance. Vertical market software can nearly always grow organically without new capital, so its cash funds more acquisitions. A business's net purchase price can even fall below zero, which pushes return on invested capital toward infinity. Leonard explains why bank debt is a poor match for permanent assets, and why he dropped his aversion to it in 2008, when depressed prices and thinner competition opened a brief window. Senra also covers how conglomerates move cash to better uses without paying tax.
The most contentious material is epistemological. Leonard borrows from Archie Cochrane to argue that executives who rely only on experience deserve the same distrust as doctors who ignore clinical trials, since science is organized skepticism and business is too often disorganized optimism. Dedicated initiative champions made new proposals plummet, which Leonard read as proof of earlier waste. Smaller stories add texture: Paul Orfalea sharing Kinko's practices without mandating them, Sam Zemurray seizing unclaimed assets he couldn't afford, and the one other conglomerate that tried this model and eventually centralized.
Constellation Software's compounding comes from a deliberately decentralized model of many small, perpetually owned vertical market software businesses, managed autonomously and improved through small-scale experiments, rather than from centralization or scale economies.
- Small business units are kept small and autonomous because complexity and coordination effort grow faster than headcount
- Initiatives are treated as hypothesis-driven experiments with one accountable owner, which cut the number of new initiatives and wasteful spending
- With 199 business units, the company can test a process in a few units and spread what works, a learning advantage that a copycat with only a checkbook and a phone cannot match
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Illinois Tool Works grew from $369 million to $4.2 billion in revenue in about 14 years under CEO John Nichols, who aggressively acquired businesses and split larger acquisitions into smaller units, reaching 365 operating units by 1996.
“He was CEO of the company as it went from 369 million in revenue to 4.2 billion. And he did that in about 14 years.”
Using bank debt to finance permanent acquired assets creates a fundamental mismatch, since short-term debt can vanish when it is needed most, so Constellation sought long-term capital while still borrowing.
“We continued to seek long-term capital to diffuse the fundamental mismatch inherent in buying permanent assets with short-term debt.”
Conglomerates can move cash from businesses with limited reinvestment opportunities to better uses without incurring taxes or much cost, and they can buy common stocks, whose gains can later fund large acquisitions.
“We can, without incurring taxes or much in the way of other costs, move huge sums of money from businesses that have limited opportunities for incremental investment to other sectors with greater promise.”
As businesses grow, policies, procedures, and rules multiply to handle complexity, and coordination effort increases faster than headcount, which frustrates talented people and shifts focus from customers to internal control.
“A huge body of academic research confirms that complexity and coordination effort increases at a much faster rate than headcount in a growing organization.”
Each initiative should begin with a stated assumption and a test that can show whether it is true or false, so that teams do not keep spending past the point where the original rationale has been forgotten.
“what is the assumption of this new initiative that you're doing, and how can you tell, what is the test you're going to do to know if that assumption's true or false.”
Although the barrier to starting a copycat conglomerate is 'a checkbook and a telephone', Constellation's real moat is its 199 separately tracked business units and open analytical culture, which let it test hypotheses quickly and spread what works.
“While there are terrific moats around our individual business units, the barrier to starting a conglomerate of VMS is pretty much a checkbook and a telephone.”
Running small tests across 199 business units lets Constellation quickly learn whether a process works, and once a best practice works, wide benchmarking access spreads it through emulation, so the advantage grows as the number of units increases.
“Once a new best practice starts working within CSI, wide access to benchmarking information tends to rapidly breed emulation.”
Archie Cochrane's critique of ineffective medical practice maps onto business: many management practices add overhead rather than value, so Constellation runs non-randomized observational studies to identify practices that actually add value.
“The asymmetric effectiveness of most medical treatments rarely influence positive outcomes, while frequently contributing to negative ones made me think critically about what I and most other managers do.”
When a software business's current earnings exceed its original purchase price, and organic growth has reduced working capital, the net purchase price can fall below zero, pushing return on invested capital toward infinity and defying conventional financial statement measures.
“If these businesses have achieved that growth organically, they have also likely reduced working capital significantly, perhaps driving the net purchase price below zero, meaning we got all the money back.”
Because Constellation's software businesses can nearly always grow revenues organically without incremental capital, their cash flows fund more acquisitions, which is how the company compounds over long periods.
“In our businesses, we can nearly always grow revenues organically without incremental capital.”
Constellation does not compete on economies of scale, centralized management, or world-class talent in large markets; instead it seeks motivated small teams in tiny vertical markets, where autonomy and an enduring human-scale team produce superior results.
“We seek out vertical market software businesses, where motivated small teams, composed of good people can produce superior results in tiny markets. That is our game.”
Executives who rely solely on their experience should not be trusted, just as doctors who ignore clinical trials should not be trusted, because science is organized skepticism while too much of business is disorganized optimism.
“We should no more trust executives who rely solely on their experience than we should trust doctors who ignore clinical trials. Science is organized skepticism. Too much of business is disorganized optimism.”
The good ideas found through experimentation are likely counterintuitive, so people who do not run experiments will not arrive at the same conclusions, which makes experimentation a source of durable advantage.
“The good ideas that we find through experimentation, they're most likely counterintuitive, so other people, unless they're running experiments, aren't going to arrive at the same conclusion.”
Organizations tend to over-complicate by adding features and markets without knowing whether they will work, because software cultures reward regular launches of visionary new products, so initiative spending grows unless it is measured and bounded.
“The ethos of software companies requires the regular launching of visionary new products by steely-eyed tenacious developers, product managers or founders.”
Creating dedicated initiative champion positions, so one person was ultimately responsible, caused the number of new initiatives proposed to plummet, which Leonard read as evidence the earlier spending had been wasteful.
“The most surprising adaptation was that the number of new initiatives proposed plummeted.”
After Constellation created dedicated initiative champion positions, the number of new initiatives proposed plummeted, because a part-time leader who does not feel ultimately responsible lets an initiative drift on a low but perpetual burn rate.
“We created dedicated initiative champion positions, so an initiative was less likely to drag on with a low but perpetual burn rate.”
Constellation's business units rarely get large, with an average of 44 employees, and a linear regression showed that unit size and performance are almost totally unrelated, so the company believes small units are better because agility outweighs economies of scale.
“The biggest is 307 employees. And the average business we own currently only has 44 employees.”
Running small experiments across hundreds of business units produces best practices that are likely counterintuitive, which other companies that do not run such experiments will not arrive at, and spreading them through benchmarking drives emulation.
“Once a new best practice starts working within CSI, wide access to benchmarking information tends to rapidly breed emulation.”
Paul Orfalea shared best practices across Kinko's stores without mandating them, reasoning that forcing adoption would lock in the best version at that moment and stop further improvement.
“Because if I make them do it, that was the best it will ever be.”
Opportunities to acquire unclaimed assets that will never reappear feel like a drop in barometric pressure to a good businessman, and a great businessman acts on them even when he cannot afford to, as Sam Zemurray did.
“A great businessman is dumb enough to act on them even when he cannot afford to.”
An ITW senior manager advised that splitting off a new business unit with a clean sheet of paper removes bureaucracy, and that as a company gets big it loses its entrepreneurial spirit, while sharing overhead functions like sales, R&D, and HR causes misallocated costs.
“When you get big, you lose entrepreneurship.”
In 2008 Leonard changed a long-standing aversion to debt because the crisis depressed software asset prices and reduced competition from buyers, so he judged the window for buying attractive vertical market software businesses at low prices to be brief.
“The downturn in the market and in the overall economy has kind of eliminated my competition. This is going to be for a very short amount of time. I should act.”
The most lucrative acquisitions have been distressed software assets that large corporations bought on the periphery of their industry, where anticipated synergies rarely materialize and cultural clashes arise, so the parent later sells them, often during recessions.
“Sometimes large corporations convince themself that software businesses on the periphery of their industry will be good acquisitions, rarely do the anticipated synergies accrue, and frequently the cultural clashes are fierce.”
Only one other high-performance conglomerate followed a strategy of buying hundreds of small businesses and managing them autonomously, and it eventually caved to centralization, which Leonard argues requires an unusually trusting culture and a long-term horizon to avoid.
“They eventually caved to centralization. My hunch is that it takes an unusually trusting culture and a long-term investment horizon to support a multitude of small businesses and their entrepreneurial leaders.”
Leonard eliminated his salary, incentive pay, and expense charges so that his compensation is tied solely to share ownership, which he argues sidesteps the agent-principal problem and makes him a partner rather than an employee.
“One of the reasons of this compensation change is that I get to sidestep the agent principal problem. My compensation for being president is now tied solely to my current ownership of CSI shares.”
When a large business stops generating the organic growth it should, its manager should be asked whether customers and employees would be better served by splitting it, with the preferred outcome being that the original manager runs the core while a protégé runs a spun-off unit.
“Our favorite outcome in this sort of situation is that the original business manager runs a large piece of the original business and spins off a new business unit run by one of his or her proteges.”
Constellation changed its managers' mental models over six years without central edicts by having initiative participants generate data, measuring outcomes, and adjusting based on the results.
“We did not get to this point with central edicts or grand plans. We just had a hunch that our internal ventures could be managed better, and we started measuring them.”
Business managers often fail to recognize that their unit has grown too large, because they lack humility, courage, or time for reflection, and so they do not see that sacrifices are becoming too great.
“No one wants to admit that they've hit their limit. Some business managers lack the humility, some lack the courage, and most lack the time for reflection.”
A perpetual owner's job is to ensure energetic, intelligent, and ethical managers run each business and are incentivized to enhance shareholder value over the very long term, which is why Constellation takes minority stakes in public software companies when value is attractive.
“Constellation's objective is to be a perpetual owner of inherently attractive software businesses.”
Henry Singleton's Teledyne grew by dividing businesses into smaller parts wherever possible, which Singleton claimed keeps managers creative and not wasteful.
“Teledyne, which is Singleton's company, Grows bigger by dividing businesses into smaller parts wherever possible.”
As the idea that vertical market software businesses are attractive spreads, Constellation faces more and more competition for acquisitions, so downturns temporarily reduce competitors and create a window to act.
“Now, as that idea gets spread, I have more and more competition.”
Leonard prefers to keep businesses small and decision-making at the business unit level rather than centralizing functions like R&D or sales and marketing for economies of scale, drawing on his venture capital experience with small high-performance teams.
“My personal preference is to instead focus on keeping our businesses small and the majority of the decision-making down at the business level.”
By 2005, Constellation's initiatives, which are long-term investments in new products or markets that take five to 10 years to reach cash flow breakeven, grew to over half of combined research, development, and sales and marketing spending, which coincided with peak spending.
“Initiatives grew to account for over half of our combined expenditures by 2005, which not coincidentally was the peak of our spending.”
A 4% customer attrition rate in 2009 implies the average Constellation customer stays about 26 years, and these durable relationships exist because the software costs rarely exceed 1% of a customer's revenues while helping them run their business.
“Attrition due to the loss of customers in 2009 was 4%. This suggests that our average customer will stay with Constellation for 26 years.”
The longer Constellation owns a small software business, the larger and better it becomes, so a serial acquisition strategy across new verticals compounds in value over time.
“The longer we have owned a small software business, the larger and better it has become.”
Founders tend to over-prioritize growth at the expense of profit, while private equity owners of software firms squeeze assets for cash, and Constellation seeks the point where incremental investment still produces good long-term returns.
“At Constellation, we try to find an optimum position where incremental investment still generates good incremental long-term returns.”
Leonard's commitment to decentralization, autonomy, and control comes from personal temperament, since he dislikes being told what to do, and he expects highly talented people to avoid centralized command-and-control businesses.
“It starts with personal preference. I don't like anyone telling me what to do.”
Leonard promotes from within nearly always, because mutual trust and loyalty take years to build, while newly hired smart or manipulative mercenaries can take years to identify and root out.
“mutual trust and loyalty take years to build. And conversely, newly hired smart and or manipulative mercenaries can take years to identify and root out.”
Leonard decided to work less intensively in the future, dropping weekends, all-nighters, and 60-plus-hour weeks, because he built an organization of seasoned managers that can operate without his constant involvement.
“I'm just not going to do the weekends, all-nighters, and a constant grind of 60-hour plus weeks that have characterized my earlier career.”
Software businesses have significant economies of scale, and Constellation grows its acquired businesses largely by frequently providing additional products for them to sell into their existing installed base.
“we generally grow our acquired businesses by frequently providing additional products for them to sell into their install base”
Constellation's favorite acquisitions are businesses bought from founders, because a founder who spent a lifetime building a business creates a long-term orientation in employee selection, customer relationships, and product evolution that fits Constellation's model.
“Our favorite and most frequent acquisitions are the businesses that we buy from founders.”
Constellation makes money in two ways: by selling additional products into each acquired business's installed customer base, and by reducing an acquired business to a profitable core, which leaves a smaller but more profitable company.
“History suggests that we generally grow our acquired businesses by frequently providing additional products for them to sell into their install base.”
If small units are better, Constellation should divide its largest business units into smaller, more focused units unless there is an overwhelming reason not to, operate most acquisitions as separate units rather than merging them, and drive down head-office and operating group costs.
“We should, number one, divide our largest business units into smaller, more focused business units unless there's an overwhelmingly obvious reason to keep them whole.”
Constellation has grown by largely abdicating management to the managers of each vertical business, keeping a very thin infrastructure overlay at headquarters and relying on general managers steeped in their verticals, mirroring Buffett's model.
“We've handled our growth to date by largely abdicating management to the managers of each of our vertical businesses.”
Constellation's acquisitions are typically small, in the $2 to $4 million range, but over 500 of them, combined with financial and operational discipline, produced a company with a market cap of about $40 billion.
“Typically, Constellation's acquisitions are small, in the two to $4 million range, but add them all up”
Many of the software businesses available for acquisition were founded with the advent of microcomputers, and their baby boomer founders are now thinking about retirement, which creates a demographic supply of sellers.
“There's a demographic element to the supply of these acquisitions. Most of these businesses came into being with the advent of many and microcomputers and many of their founders are baby boomers who are now thinking about retirement.”
Constellation's preference, borrowed from Buffett, is to acquire businesses in their entirety and own them forever, buying minority stakes only when a full purchase is not possible, with the prospect of eventually acquiring the rest.
“Our preference is to acquire businesses in their entirety and to own them forever.”
Over the last decade, an equally weighted portfolio of the high-performance conglomerates Constellation studies would have more than doubled the performance of the S&P 500, a benchmark Leonard uses to set a high bar.
“Over the last decade, if you had an equally weighted portfolio of the shares of these high-performance conglomerates, you would've more than doubled the performance of the S&P 500.”
Like Warren Buffett and Charlie Munger, Leonard says to work only with people you like, admire, and trust, since many people are rat poison and need to be vetted carefully.
“the need to only work with people you like, admire, and trust, how a lot of people out there are rat poison”
Leonard recommends Daniel Kahneman's Thinking, Fast and Slow for understanding the efficiency, speed, and conceit of intuitive judgment and its infrequent but severe failures, and argues behavioral economics findings provide insights into investing and management.
“He's helped me appreciate the efficiency, speed, and inherent conceit of intuitive judgment, and its infrequent but often abject failures.”
At human-scale companies where everyone knows each other, team members who are not trusted get weeded out, quirky talent is tolerated if it serves the business, and focus stays on growing the pie rather than dividing it.
“Trust and communication are high, and focus tends to be on how to increase the size of the pie, not how the pie gets divided.”
Mark Leonard's Constellation Software grew from an initial $25 million investment in 1995 to a roughly $40 billion market cap over about 27 years by acquiring over 500 small vertical market software businesses.
“With an initial $25 million investment in 1995, Mark Leonard has built Constellation into a world-leading consolidator of vertical market software companies.”
Senra's method is to extract business ideas from people who have spent decades studying business, such as Buffett and Leonard, because the operators' accumulated observations across many businesses offer knowledge few others possess.
“There's probably no one else on the planet that has spent more time analyzing a wide variety of businesses.”
The host reads shareholder letters from a founder's perspective rather than an investor's, because owning many businesses lets the owner observe performance across a wide range and extract best practices to share.
“I don't consider myself an investor. I read shareholder letters and I think about it from a founder's perspective.”
If advising his 35 or 40-year-old self, Leonard would tell him to stay put and become a master craftsman in managing a vertical market software business, since it is the most satisfying job and generates enough wealth to live comfortably.
“I would tell him to stay put, become a master craftsman in the art of managing your VMS business.”
Leonard uses his shareholder letters to write about the business rather than the stock, and despite a 2010 shareholder revolt, the stock rose more than 40 times in the decade-plus after that letter.
“As a rule, I use these letters to write about our business, not our stock.”
Leonard's approach of breaking large businesses into manageable pieces mirrors Rick Rubin's idea of production by reduction, removing everything that is not the essence of something.
“It's production by reduction. I enhance things by removing everything that's not its essence.”
Abnormal business results require non-normal choices, so a company that wants exceptional outcomes must do things differently from typical peers.
“you can't be normal and expect abnormal results.”
Leonard is openly receptive to new information and changes his mind when evidence changes, as shown by his shift from avoiding debt to using it when the environment changed.
“he's receptive to new information, he's going to change his mind.”
Leonard avoids sharing sensitive information with competitors, which the host relates to the entrepreneurial maxim that bad boys move in silence.
“We don't like sharing sensitive information with competitors.”
Leonard's dislike of authority came from his youth, when he challenged teachers, left home early, and did things not accepted by others, and he prizes the ability to choose what he thinks is right.
“I challenged teachers. I left home early.”
Leonard flirted with a fundamental change by considering large UK public-sector vertical software conglomerates, whose whisper prices he could just about stomach if financed standalone like private equity firms, believing Constellation would be better long-term owners.
“My sense is that we would be better owners of these assets and would generate better long-term performance for them than the PE”
Mark Leonard keeps an extremely low public profile, with only one photograph online, so the most reliable public information about him comes from his shareholder letters and investor calls.
“There's only one photograph that you can find online of what he looks like.”
Jeff Bezos's shareholder letters were the host's first, and the host reads letters like them to get practical playbooks from people who mastered a business before playbooks were written.
“To read Jeff's shareholder letters is to get a crash course in running a high growth internet business from someone who mastered it before any of the playbooks were written.”
Leonard's shareholder letters are written in a dense, hard-to-parse way that would reach more readers if simplified, unlike Buffett's letters which speak plainly enough to serve as a customer acquisition channel.
“I've had to reread certain sections a bunch of times.”
Within a year, Constellation had about $485 million of debt outstanding in bank borrowings, used because cash flow, though growing fast, was insufficient to finance acquisitions.
“We had $485 million outstanding on our debt.”