Manufacturing Alpha vs. Engineering Failure: What Medallia Reveals About When GTM Capital Actually Works

On April 22, Thoma Bravo agreed to hand the keys of Medallia to its lenders. The take-private had closed in 2021 at a $6.4 billion valuation. Five years later, $5.1 billion of equity is gone, and Blackstone, KKR, Apollo, and Antares now own a customer-experience software company they never set out to operate. Orlando Bravo's own framing was unusually direct: "We made a mistake, and that caused us to pay too much." Blackstone's Brad Marshall was sharper. Medallia, he said in February, had been underperforming "not because of anything related to AI, but due to what we believe to be execution-driven issues."

I have a small piece of personal context here. I knew Medallia when it was twenty people, in the late 2000s. The husband-and-wife founding team — Borge Hald on the technical side, Amy Pressman as the commercial leader — had come out of Stanford GSB a few years earlier with an idea that was genuinely new: take Net Promoter Score, which had been productized by Bain as a consulting framework, and build it into enterprise software. They were the first to do it at scale. Qualtrics, ForeSee, and the rest of the category came later or pivoted in.

This isn't a story about a failed company. Medallia at twenty people was a real business with a real moat. This is a story about a successful company caught at the wrong moment by the wrong kind of capital.

In the early 2010s, what Medallia sold wasn't software. It was a methodology wrapped in software. Fortune 500 CXOs bought a thesis about closing the loop on customer signal, and the thesis was differentiated because the founders embodied it. Borge built a platform that other engineers respected. Amy carried the commercial intensity that made the methodology stick inside the buyer's organization. The competitive landscape was thin. Qualtrics was still mid-market. Confirmit and ForeSee were instrumentation plays without the methodology layer.

This is the period when embedded operational expertise is the company. You cannot separate GTM from product from culture. They are one thing, and that one thing compounds — every new logo deepens the methodology, every methodology improvement closes the next logo faster, every closed logo trains the next sales hire.

The compounding engine was real. For a while.

A recent analysis by Jacco van der Kooij at Winning by Design makes a useful technical point about Medallia's trajectory. Looking at the S-1 data, ARR climbs steadily through the public-market era — from roughly $374M to $576M post-IPO. The headline number looks fine. But the second derivative — the rate of change of the growth rate — oscillates around zero. Each time the engine builds momentum in one quarter, the next quarter resets it. The company is growing in absolute terms while the acceleration of that growth has been negative or neutral for years.

Van der Kooij's framework is a powerful autopsy tool. It separates companies whose growth compounds from companies whose growth has to be repurchased every quarter — the former through working sales motions and product loops, the latter through marketing spend, headcount expansion, or eventually M&A. Medallia by the late 2010s was clearly in the second category. The methodology had been catching up by competitors. The original sales motion that worked at $50M ARR did not replicate cleanly at $400M ARR. Reps were being hired faster than the playbook scaled.

Here is the part that matters: by the time the math reveals the deceleration, the window for operational repair has already closed. The 2019 IPO masked it. The 2021 take-private papered over it. The 2026 wipeout finally surfaced it. Eleven years from compounding-engine failure to public reckoning, with $5.1 billion vaporized in between.

The 2021 take-private was not reckless. Thoma Bravo runs one of the most respected playbooks in software private equity, and that playbook works: install professional leadership, drive cost discipline, optimize the operational stack, hold for multiple expansion. The sponsor brought a new operating team in early 2025 and began working a turnaround plan. By every measure of process quality, this was a competent execution against a real plan.

The problem is structural, not behavioral. Growth PE playbooks assume the underlying GTM engine is intact and just needs operational efficiency layered on top. They are designed to take a working motion and make it more profitable. They are not designed to rebuild a non-compounding motion from inside a leveraged cap structure.

Medallia's GTM engine had already decompensated by the time the take-private closed. So the playbook had to compensate financially. The loan was 100% PIK in year one — payment-in-kind, meaning interest was added to principal rather than paid in cash — because the company was burning cash at acquisition. The loan balance grew from roughly $1.8 billion to $3 billion through a combination of accumulated PIK and draws to fund acquisitions. Acquired revenue substituted for organic growth that would not arrive on its own. Each financial tool — leverage, PIK, M&A — assumes organic compounding exists underneath it. Layered onto a non-compounding engine, they accelerate failure rather than prevent it.

By the time Bravo's new leadership team arrived in 2025, the cap structure had foreclosed the operational fix. At $400 million in ARR with $3 billion of debt, you cannot rebuild a sales motion. There is no slack. There is no time. There is no air for the kind of patient, embedded GTM work that a real turnaround would require. When Blackstone declined to extend the PIK toggle in late 2025, the binary closed: write another equity check into a sinking position, or hand over the keys.

The lesson here is not that Thoma Bravo is bad at what they do. They are excellent at what they do, when the inputs match the playbook. The lesson is that GTM execution capital has a window, and the window is upstream of where Growth PE operates.

At $1 to $5 million in ARR, the GTM motion is still malleable. The founder is still in the building. The ICP is still being defined. The first ten enterprise reps are about to be hired, or have just been hired and are not yet productive. The methodology — if there is one — is still being codified. At this stage, embedded operators do not repair a broken engine. They help build the engine correctly the first time, before the cap structure, the headcount, and the customer base have all been committed to a motion that will not scale.

The economics are also categorically different. A Series A SPV deploys $6 to $7 million against a $20 million post-money entry, with no leverage, alongside operators who have built these sales motions before, on a 3 to 5 year hold to a Series B re-rating. There is no PIK toggle. No acquisition treadmill. No $3 billion debt stack foreclosing the operational work. The capital and the operational thesis are the same instrument, deployed at the moment when both still have leverage.

The asymmetry is not that we are smarter than Thoma Bravo. The asymmetry is that we are operating in the part of a company's life where execution capital still has room to compound. By the time a company is large enough for Growth PE, that leverage has compressed. What remains is financial engineering, and financial engineering cannot manufacture GTM compounding. It can only paper over its absence.

The 2020/2021 vintage contains many companies in the Medallia pattern — acquired at peak multiples, growth never compounded organically, now staring down recapitalizations or worse. Most of these companies did not fail because of AI disruption or market conditions. They failed because the GTM engine stopped compounding years before anyone reading the ARR line could see it, and capital arrived too late to fix it.

The right time to deploy operator capital is not when the second derivative goes red on a public-market dashboard. It is when the founder is still selling every deal personally and a working sales playbook would change the trajectory. That window is short. It closes earlier than most allocators think.