Investing in Overlooked Businesses
What this covers
Tracy Graham and host Patrick work through how a buyer of lower-middle market companies can treat data as a free option layered on a cash-producing base. Much of the talk is practical. Tracy lays out how to test a proprietary dataset: whether access can be exclusive, whether the data is perishable, and whether models built on it produce insights others would pay for. He insists that data is not the same as insights. His main example is the Graham Allen oncology database, which he says captures about 80% of one region's chemotherapy and could be sold to drugmakers such as Pfizer or Eli Lilly for near-real-time views of physician drug choices.
The range is wider than data valuation. Tracy describes sorting customers into profit drivers, drains, and drags through transactional analysis, notes that companies use only 10% to 20% of the software they pay for, and explains why insurance brokerages suit technology that augments skilled labor. He calls clinical healthcare AI fool's gold, since 90% accuracy is useless for care decisions even though it beats the competition in administrative work. He also takes a contested position: secondary-market customers may prefer a delivered service to another tool that needs data scientists. Distribution through managed service providers comes up as a possible place to invest.
The last stretch is personal. Tracy ties his intuition and situational awareness to growing up in Englewood, Chicago. He describes borrowing Lou Holtz's habit of keeping principles simple, and argues that character shows in hard decisions about conflicts and governance. He also argues that exposing a young person to different environments can change a family's trajectory.
Tracy Graham argues that lower-middle-market businesses in secondary and tertiary US markets are undervalued because of their geography, and that applying data, AI, and analytics to them creates the margin expansion and growth that drive returns.
- Secondary-market owners lack data talent and have run their businesses the same way for decades, so they underinvest in technology and data
- Valuable proprietary datasets are often unvalued by sellers, creating a mispricing opportunity
- Transactional-level, customer-by-customer profitability analysis reveals margin drains that aggregate reporting hides
Undervalued secondary-market lower-middle-market firms gain most from data and AI modernization.
- Lower-middle-market companies in secondary and tertiary markets are generally undervalued in many cases simply because of their location or geography, since people assign them less value.
“we believe that lower-middle market companies and specifically those in secondary and tertiary markets are generally undervalued in many cases simply because of their location or their geography.”
- Helping lower-middle-market companies in secondary markets adopt technology and data produces the most margin expansion initially, because those firms lack talent and are run by owners who have operated the same way for about 30 years, making the transition hard but the payoff large.
“you get the most lift initially in those companies. You get the most margin expansion”
- The modernization of traditional businesses with data and AI will play out over the next 10 to 20 years, and Tracy believes this is the opportunity his firm should focus on.
“I think what you just outlined is exactly what we're going to see over the next 10 to 20 years.”
- Small secondary-market companies face a threat as larger competitors use AI and technology to reach into secondary markets and compete with them, while slow-to-adapt firms fall behind.
“their larger competitors find ways to use AI and technology that allows them to reach into the secondary markets and compete better”
- Lower-middle-market sellers generally place little value on their data because they do not use it, and most lack data scientists, which creates a mispricing opportunity for buyers, though Tracy expects the value of data to be understood within about five years.
“there's very little value by the seller that's put on data because they don't use it.”
- The typical Midwest lower-middle-market manufacturer makes one or two core products in one or two plants, with roughly $50 to $100 million in revenue and $5 to $15 million in EBITDA, and has not adopted a scale mindset.
“They are two plants maximum, one to two plants. They're going to be somewhere between $50 and $100 million in revenue, somewhere between probably $5 and $15 million in EBITDA.”
- The typical go-to-market model for enterprise software is a direct sales force for the largest customers and a channel partner program that sells to every other customer through managed service providers, who bill separately for implementation.
“for our largest customers we are going to have a direct sales force. And then for every other customer, we're going to build a channel partner program”
- Customers can be sorted into profit drivers, which to spend the most time with, profit drains, and profit drags, which are customers dragging down net income, and client success should be structured around this sorting.
“You have your profit drivers, which are the ones you want to spend the most of your time with. You have your profit drains and then your profit drags.”
- Graham Allen focused on secondary-market lower-middle-market companies because few private equity firms compete there, which leaves the segment off the beaten path and undervalued.
“this market that I think is off the beaten path, doesn't have a lot of competition from other private equity firms.”
- Graham Allen believes its ability to use data and AI to lift the margins and growth of its portfolio companies is its alpha generation engine.
“Through that, we believe that's the alpha generation engine, which is our ability to use data and AI to help companies do that.”
Proprietary data wrapped in insight, not the model, creates sellable forecasting value.
- When evaluating a dataset, the first test is whether the company can get exclusive access to it, whether it is proprietary, and whether it is perishable, and then whether models built on it yield insights that are valuable to others.
“can we get exclusive access to it? Is it proprietary in any way? And is it perishable?”
- Through clients and ownership of the oncology groups in the region, Graham Allen's database captures about 80% of the region's chemotherapy, including patient stage, drug use, outcomes, cost, insurer, and schedules, giving exclusive forecasting value.
“So we have probably about I would say 80% of the region's chemotherapy is running through our database.”
- Graham Allen can sell its oncology data to pharmaceutical companies such as Pfizer or Eli Lilly for regional sales forecasts and near-real-time views of which drugs physicians choose, because the database tracks upcoming chemotherapy for about the next 8 weeks.
“We know every patient that's going to receive chemotherapy for the next let's call it 8 weeks or so.”
- Graham Allen plans to use AI internally to improve its own margins and to monetize its proprietary data through external sales.
“The first is to use AI internally to improve the margins of the business. The second thing is to figure out how we monetize some of the proprietary data”
- Graham Allen's team debates whether a company's real value lies in its model or its proprietary data, with one side arguing exclusive regional data enables a better model and the other arguing competitors can recreate the data, so the model matters more.
“What's the real value? Is it the model or the data?”
- Owning a lot of valuable-seeming data is not enough: without creativity and a business model to wrap around it, and without finding interesting insights, the data is difficult to monetize, because data is very different from insights.
“Very different between having data and insights. As you know, real insights are hard to come by.”
- Large manufacturers such as Whirlpool collect field-usage data from sensors on products and can use it to improve engineering, but they often struggle to turn that data into value.
“putting sensors on products and things of that nature, collecting data about their product as it's out in the field.”
- AI allows a business to analyze each customer individually, which is virtually impossible with human analysts, and building that capability into the data platform is more efficient and effective than manual customer analysis.
“allowing you to look at each customer individually and that's virtually impossible if you're going to have a human doing all the analysis.”
- Today's marketing AI is largely everyone doing the same thing, targeting customers better, and the real gains for Graham Allen's portfolio came from clean data and a better understanding of customers rather than from novel AI.
“it's basically everybody's doing the same thing. It's trying to help you target customers better.”
- Most small service businesses evaluate customers at the gross margin level using aggregated data, so customers with identical services can differ hugely in cost to serve, and only transactional-level data reveals which customers are profit drivers, drains, or drags.
“Most people operate at a gross when they're evaluating. These companies operate at a gross margin level or gross profit level and not a net income level by customer.”
Data and technology should improve lower-middle-market economic models, not just add tools.
- Technology applied to traditional businesses should improve the economic model itself, not merely add a tool, meaning the customer and product experience and the value chain are where data creates value.
“I think the customer experience, the use of their product is a big part of how you use data to improve in a manufacturing space.”
- The primary data use cases in manufacturing are improving product quality and customer experience or service, since many manufacturers run call centers that generate large amounts of data, though no one yet has a definitive answer.
“where we think the opportunities are for manufacturing specifically is improving the quality of the product, improving the customer experience, or customer service.”
- Tracy's vision is that a plant will become a big data center with machines carrying sensors, and that most of the work people now do in the plant will eventually be automated.
“when I see a plant, what I see is a big data center with a bunch of machines in it.”
- Pharmaceutical customers want to know, almost in real time, which drugs physicians chose when they did not choose the manufacturer's drug, which lets drugmakers educate doctors on efficacy.
“they want to know when they don't choose their drug, what drug they did choose.”
- Tracy has never walked into a lower-middle-market company and found a data scientist to talk to, which suggests that these firms have not explored how to use their data.
“I've never walked into a lower-middle market company and said, 'Can we talk to your data scientist?'”
- Valuing a proprietary dataset is hard, with no secret formula; the real question is how much margin expansion access to the data can produce, which requires complex modeling.
“What you're trying to figure out is how much margin expansion you think you can get as a result of having access to the data.”
Labor-intensive back offices are inefficient, so technology yields large margin gains there.
- Graham Allen acquired a back-office operation for an oncology group, modernized its workflow with data and technology, and substantially raised margins and growth, showing that labor-intensive administrative processes still run the way they did 30 years ago.
“We have, in effect, modernized that entire process, margins up significantly, growth up significantly.”
- The back office of healthcare providers is highly labor-intensive and inefficient, and its workflow still operates as it did 30 years ago despite more regulation and compliance.
“They haven't changed the model, so the workflow still operates the same.”
- Clinical healthcare AI has been fool's gold for Graham Allen because it requires much higher accuracy than administrative tools: 90% accuracy beats the 60% competition on back-office costs but is worthless for clinical care decisions.
“if we're at 90%, we're competing with 60%. The clinical side of it, if you're at 90%, it's worthless.”
- Plant maintenance and servicing companies that handle outsourced machine upkeep are an attractive target because today their service is labor-costly and inefficient, and connected machines could let them predict which machines need oil or service, though Graham Allen has not yet done a deal there.
“the maintenance companies are out there doing that. That's one we're looking at.”
- Insurance brokerage businesses are attractive because their cost of revenue is highly labor-intensive and they sell intelligence and ideas, so technology can augment highly skilled people much as it does at a hedge fund.
“What they're selling is a lot of intelligence and ideas. We think it is those types of businesses that are going to be ripe for using better technology, not unlike a hedge fund.”
Shared values and character revealed under hard decisions guide doing business right.
- Lou Holtz's effectiveness came from a simple message delivered with passion and never changed, and simplification made the message digestible for players from very different backgrounds.
“the message never changed. In fact, he once said to our team.”
- Principles need to be simple and digestible, not profound, so that people with varied backgrounds can rally around a shared set of core values, which is the Lou Holtz approach Tracy applies to his company.
“Don't complicate it. It doesn't have to be something that sounds profound. It has to be digestible by a lot of people who all have various backgrounds”
- Character is revealed in hard decisions, not easy ones; people reveal who they are when decisions get tough, so the real test of doing business the right way is how one handles conflicts, governance, and conflicts of interest.
“But the business lesson for me is you really know what people are all about when decisions get tough.”
- Tracy measures the quality of her business conduct by whether the people who did business with her 20 years ago still do business with her today, citing Notre Dame as a customer in her first company and now an investor.
“nothing's more important to me than to say the people who did business with me 20 years ago when I started, generally still do business with me today.”
- The foundational business principle is to do right, do the best you can, and treat people the way you want to be treated, because character shows when decisions get tough rather than when things go well.
“the business lesson for me is you really know what people are all about when decisions get tough.”
- Lou Holtz's central lesson was getting diverse young players from very different economic backgrounds to focus on something more important than themselves, a model Tracy says applies to leaders of businesses.
“get all of those people to be focused on something that's more important than themselves. Foundationally, that's how I think about businesses.”
Small Midwest manufacturers need data capability first, with deals priced near 8x EBITDA.
- Large Midwest manufacturers in the recreational vehicle space operate about 100 plants and make a wide range of products, from levelers and jacks to bike racks, which contrasts with the typical small single-product firm.
“In the RV, recreational vehicle space, which we are literally 20 minutes away from. So in that scenario, there are in many cases 100 plants”
- A typical target is a family-owned, 15-year-old single-product manufacturer with about $35 million in revenue, roughly 25% EBITDA margins, and customers including U-Haul, now run by the second generation, and wants to grow but needs help scaling.
“They do about 25% EBITDA margins, one plant in this case about $35 million a year in revenue.”
- Smaller Midwest manufacturers collect very little data and run highly manually, often with a receptionist handling billing, invoicing, and pricing negotiation, so the first step to using data is building the capability to collect and integrate it.
“It'd be scary to you if you saw how it actually works today, very little data, highly manual”
- Graham Allen generally pays around 8 times LTM EBITDA, rarely more, and about 6.5 to 7 times forward EBITDA on a pro forma basis, with each deal judged independently.
“we pay somewhere in the 8 times LTM and then it's going to be 6.5 times forward.”
Software wins by delivering services through intermediaries instead of in-house talent.
- The key question is whether secondary-market customers want another software tool, which requires scarce in-house talent, or a fully delivered service, and Tracy believes winning products may be ones that sell the service rather than a tool that needs data scientists to operate.
“The question really becomes, is tooling what customers really want, or is it the service that they want?”
- Software sold into secondary markets is typically distributed through managed service providers rather than direct sales, because vendors such as Amazon or Salesforce rarely sell directly there, so a middleman controls distribution and is a potentially attractive investment.
“I will tell you, in our region, we have never seen anyone from Amazon come to sell technology, Salesforce, or any of those companies. There's always a middle person”
- Apple's decision to open its own stores and Tesla's decision to avoid dealers show that companies sometimes control the customer experience end to end by bypassing intermediaries, which Tracy sees as a possible model for software distribution.
“Apple decided, 'Look, we're going to build our own Apple stores. We're going to sell it direct.'”
- Companies typically use only 10% to 20% of the capabilities of the software they buy while paying for 100%, so software is one of the most under-utilized assets on the balance sheet.
“companies use 10%, 20% of the capabilities of the software, but they pay for 100% of it.”
Mentors Holtz, Murphy, and Jordan changed Tracy's career trajectory through exposure.
- Lou Holtz told Tracy's mother that a Notre Dame degree would be the best four-year investment he makes over the next 40 years, even if he never played football there.
“he'll get a degree and it'll be the best 4-year investment that he makes for the next 40 years of his life.”
- Tracy's first company completed about 23 acquisitions before exiting in 1999, and the three-year process, with mentors Jay Jordan and Chris Murphy, gave him a graduate-level education in lower-middle-market integration.
“We went on to do 23 or so acquisitions, build up the company, exited in '99.”
- When Lou Holtz refused to invest in Tracy's company, he introduced Tracy to Jay Jordan, which proved more important, and Jordan later invested and bought 60% of the business for a few million dollars.
“I think one of the most important things he did was he introduced me to Jay Jordan.”
- Chris Murphy, chairman of 1st Source Bank, cancelled a meeting, flew Tracy to Naples, Florida, and hosted him for seven days with his family, giving Tracy lessons and a relationship that has lasted nearly 25 years.
“We got on his plane and we flew to Naples, Florida. I spent 7 days with him and his family.”
- Giving a young person access to different environments, such as the Hamptons or Naples, can change how they see the world and can alter a family's trajectory for generations.
“I think it's a metaphor for just giving a kid access and letting him see the world differently than they've ever seen the world”
- The kindest acts Tracy received were from Chris Murphy and Jay Jordan, who gave her access and exposure to a world she had never seen, and that changed the trajectory of her thinking, career, family, and generations of her family.
“Both of those two have done things that are so kind to me and really changed the trajectory of my thinking”
- Lou Holtz and the University of Notre Dame have stayed in regular contact with players for about 25 years after graduation, and that continuing bond shaped Tracy's career by connecting him to key mentors.
“I think the University of Notre Dame and specifically Coach Holtz continues to reach out to players”
Early ventures and personal origins taught lessons about capital, pivots, and intuition.
- Tracy ran out of money quickly after receiving $80,000 from Jerome Bettis because he thought that sum was enough to do everything, showing that a first-time founder misjudges how far capital goes.
“I thought $80,000 was the world at that time. I thought I had enough money to do everything.”
- Tracy named Graham Allen Partners after childhood friend Kenneth Allen, who was unable to attend St. Rita High School because his mother would not pay the $20 entrance exam fee and who was killed in April 1991.
“I wanted to really never forget him. There is no living Allen.”
- Tracy's first venture, an online phone-book directory started in 1996, failed to find users because few people had internet access and most who did used 28.8 modems, so the team pivoted to providing internet access.
“people didn't have internet access yet so there was no way they were going to do any searching on the internet.”
- Growing up in Englewood, Chicago, where walking to school required awareness, built innate skills such as intuition, creativity, and situational awareness that are valuable in evaluating companies and structuring win-win deals.
“Those are things that I got a lot of reps on and, in a sense, have become innate.”
- Tracy's early work involved hundreds of conversations with business owners who were also customers, giving her a feel for how secondary-market owners think and why they did or did not invest in technology.
“having literally hundreds of conversations with owners of businesses because they were also our customers.”
- A scholarship to a private high school in Chicago changed the trajectory of Tracy's life by exposing him to a much wider world beyond his neighborhood.
“I was fortunate enough to have an opportunity to attend a private high school on a scholarship.”