Theranos extreme KPI drift
Hey all,
As someone who was very deeply read on the Theranos scandal, it's no surprise that I went to watch the box office hit “You can see everything” last week.The movie covers the 34 days in Elizabeth Holmes' life before she begins her sentence in federal prison.
As a viewer it tries to push you into a binary choice: whether she was an unwitting instrument in the fraudulent downfall of the company or a highly devious architect of it. You can judge for yourself but my conclusion is leaning more and more towards the fact that both may be true.
Back to the scandal itself: it is easy to understand how the belief in and desire for a company to succeed can lead to a couple of white lies which snowball and turn into an increasingly large gap between reality and the story that’s being told.
Theranos was basically an extreme version of the KPI drift that we regularly see in portfolio company reporting.
I encourage all those interested in these things who haven’t been in this rabbit hole to take a look - suggest starting with the book ‘Bad Blood’ rather than the controversial (but more popular) interpretive retelling ‘The Dropout’.
If you *have* been down the rabbit hole - let me know I’d always welcome a conversation about it.
Enjoy the rest of the memo!
Three Things I Learned This Week
The banker said he couldn’t show the books to anyone
This week’s PE Data Guy guest, Sal Tirabassi, spent fifteen years as a growth and venture investor, nearly a decade of it as a partner at M/C Partners. Then he crossed the table. His firm, CFO Pro+Analytics, puts fractional CFOs into founder-led and family-owned companies from $3 million to $100 million in revenue. It also takes prep-for-sale work from bankers who need a finance function rebuilt before a process.
Watch this week’s episode of PE Data Guy, with Sal Tirabassi
I asked him about a founder-owned business he helped get to a sale of about $20 million. He is careful to say he was not the broker. A banker was, and the banker had looked at the books and concluded, in Sal’s retelling, “I can’t even take this to market because I can’t show this to anyone.”
The founder took exception. You can see we make money on the tax returns, he said. Why isn’t that good enough?
Sal’s answer is that a buyer will not run the business the way the founder did. They install professional investors, and professional investors need projections they can rebuild.
Here is what he found. The company was two entities that had never been consolidated. One built heavy equipment and sold it to outside buyers. It also sold to its sister company, which rented the same equipment out. Think of a backhoe. A big contractor owns one, and a small builder rents one.
Transfer pricing between the two had never been reflected properly, so the bankers thought the rental business was far more profitable than it was. One of the ERP systems had a glitch that double counted revenue and expenses. Sal’s team cleaned that out and built a consolidated view the company could produce again every month, by entity and combined.
The story that came out the other side was a better one. The equipment was differentiated and few people made it. Every unit the rental company bought from its sister kept the margin inside the family of companies. A buyer can price a vertical integration story like that. Nobody could tell it while the numbers sat in two unreconciled entities.
He found money on the way through too. The CPA had never looked at R&D tax credits for an equipment company. Sal had a short conversation with the build team, who tested every new version and every build-to-suit before it shipped. That is R&D. The company went back three years and claimed.
The line from the episode I keep coming back to is Sal’s. “The building blocks are what matters more than the actual financial statements. The financial statements are just the end presentation.” A good finance function can see gross margin drop and trace it to a SKU or a client, and say why.
I gave him the version I see most. A P&L shows a 10% margin while every business unit reports 15% or better, because overhead is allocated differently in each one. Every dashboard is green except the consolidated P&L. Sal’s view is that finding the why is the easy part. The hard part is pushing the data until everyone else’s dashboard turns yellow too.
That matters at exit because of what a buyer does with a gap. As I said to Sal, a PE fund will not tell you to go away for three months, fix it and come back. They estimate what it costs to fix and they cut the price.
One thing to take into your next board pack. Add up the business unit margins and set them against the consolidated margin. If every unit beats the total, find out where the cost went before a buyer asks.
Watch the full conversation here.
Nearly half of US PE-backed companies are five to nine years in
J.P. Morgan Asset Management’s September note on private equity inventory, built on PitchBook data, puts the share of US PE-backed companies held for five to nine years at 46% in 2026. In 2016 it was 30%.
Managers bought heavily when rates were low, often at high multiples. When rates rose they held on rather than sell at prices that would lock in weaker returns.
The length of the hold tracks the exit price. Companies sold inside five years have historically exited above the GP’s own valuation 87% of the time. Nearly a third of companies held for more than ten years exit below it.
The exits that are happening sit at the top of the market. Global exit value in the first half was $620 billion, while the number of exits was about the same as a year earlier. A handful of outsized deals did most of the work.
The authors are measured about it. They do not expect a broad valuation reckoning, and they note that value creation was already moving from multiple expansion to revenue growth and margin through the 2010s. Their warning is narrower. The longer an asset stays unsold, the more it matters whether the returns come from the business or from deferred exit assumptions.
The five-to-nine cohort is the company I meet most. The value creation plan is green and the hold has run long. The mark is the GP’s number. The exit price is the buyer’s. What closes the gap is what Sal’s banker needed, books a stranger can rebuild and believe.
Read J.P. Morgan’s note here.
Nobody has shown a buyer paying more for AI yet
John Sviokla went looking this month for evidence that buyers pay more at exit for AI-driven gains. “I found no dataset showing buyers pay more at exit for AI-driven gains,” he wrote in Forbes. “What exists is adviser anecdote.”
The anecdotes cut both ways. EisnerAmper describes a distribution company whose AI demand forecasting lifted inventory turnover by 15% and moved its valuation from about 7x EBITDA to about 9x. It also describes a marketing agency whose buyers pushed for a lower multiple, because AI tools could do much of its work.
The adoption numbers are thin too. FTI Consulting surveyed 200 fund and operating leaders. 95% said their AI initiatives met or beat the original business case. Only 36% of portfolio companies have AI in production across use cases, and 7% run it at enterprise scale. FTI itself notes the business cases were often conservatively scoped.
Accordion put the question to technology operating partners in the second quarter. 86% expect buyers to pay a premium for AI-enabled finance within two years. 9% have seen one in a completed deal. 44% say buyers ask about it in diligence and it has not reached the price. Only 38% measure AI’s revenue or margin impact.
Sviokla’s question for exit is whether you can show a buyer measured results from AI the company owns and runs.
That question has a prerequisite most companies skip. You can only show a measured result if the before number was measured too, with the same definition, from the same source. Sal’s founder pointed at his tax returns and the banker said no. A company that points at its AI pilot deck will get the same answer.
If AI is in your 2027 plan, write down this quarter’s baseline for the metric it is meant to move, and agree how it is calculated, before the tool goes live.
Read Sviokla’s piece here. Accordion’s benchmark is here.
Two News Stories From This Week in Mid-Market PE and Data
Ares paid KKR about $58,000 a railcar
Sources. Business Wire, Oct 19 | GATX filing with the SEC, Jan 5
What happened
KKR agreed to sell Infinity Transportation to Ares funds for an enterprise value of about $2.9 billion. Infinity leases about 50,000 railcars to more than 350 customers across the US, Canada and Mexico. KKR’s insurance arm, Global Atlantic, first invested in 2016. Since 2021 Infinity has built a standalone operating platform and grown its fleet, and KKR says book value is up 150% in that time. The deal is expected to close in the first quarter of 2027.
For comparison, GATX and Brookfield closed their purchase of Wells Fargo’s operating lease fleet on January 1, about 101,000 railcars for about $4.2 billion.
Why should you care
Divide and you get roughly $58,000 a car for Infinity against roughly $42,000 for the Wells Fargo fleet. It is a rough sum. Infinity’s price also covers intermodal equipment and a running company, and the two fleets differ in age and mix. Even so, about $16,000 a car sits between them. Wells Fargo sold a bank’s portfolio of assets. KKR sold a platform with its own management team, which Ares says it will back. The price above the asset count is what a buyer pays for the operating business, and it is the part of the price you have to evidence.
More money, fewer deals
Source. KPMG Pulse of Private Equity, Oct 22
What happened
Global PE investment reached $1.5 trillion across 14,650 deals in the first nine months of 2026. By value the year is on pace to be the third best on record, after 2021 and 2025. By count it is on pace to be the slowest since 2020. On a rolling twelve-month basis deal count fell to 20,362, the lowest since the first quarter of 2021. Energy and natural resources drew $243.3 billion and is on track for a record.
Why you should care
Capital is going into fewer and larger deals, and KPMG’s word for the year is selectivity. For a mid-market company heading to a process, that means fewer bidders looking harder at each asset. A buyer with plenty of options spends its conviction on the company whose numbers hold up the first time they are tested.
Read KPMG’s summary here.
Free Tool of the Week - The Buyer’s Private Scorecard
This is the rubric a buyer’s diligence team uses to grade your data. Six categories, three ratings each, and a note on what the buyer is thinking as they score each one.
Sal’s founder found out what an outsider thought of his books when the banker told him. Grading yourself first is cheaper.
Sign-off
If any of this lands and there is something you think we can help with, just reply. We read everything that comes through.
As always, forward this on to your favorite PE-backed friend.
Cheers,
Graeme
Graeme Crawford | Crawford McMillan