The Custodian · An Investor's Story

The Setup

James is a category-focused investor. Third fund. Known for backing founders early — before the category has a name, before the market has confirmed the thesis. He has a reputation for seeing structural positions that pattern-matching investors miss.

He has also watched a portfolio company that looked right at every stage fail to become what he backed it to be. The founder was exceptional. The product was real. The revenue was real. And eighteen months in, the category dissolved back into the one next to it. He has been trying to understand why ever since.

He agreed to a session because a colleague described it as the instrument he'd been looking for without knowing it existed. He's skeptical. He is also the kind of investor who takes skepticism seriously enough to test it.

Step One — The Recall

Custodian Before I show you anything — I want to ask you something first. In the last three years, is there a company you backed that felt completely right when you made the investment — you had good structural reasons, you'd done the work, you committed with conviction — and then somewhere down the line it cost you in a way you didn't see coming?

James Yes. A company in the enterprise workflow automation space. We led the Series A. The thesis was that the existing category — robotic process automation — was built around a structural constraint that had become removable. The founder understood this with unusual precision. The product was genuinely differentiated. Revenue tripled in the first year. We had enterprise logos. Every metric confirmed the thesis. Eighteen months in, the category hadn't separated. The revenue was still growing but it was growing inside the old category's logic — enterprises were buying it as a better RPA tool, not as a fundamentally different way of organising work. The category never formed.

Custodian Did you have good reasons when you made the investment?

James Yes. The structural analysis was sound. The founder was exceptional. The product did what it claimed. The early deployments produced genuine outcomes.

Custodian So you weren't careless.

James No. We'd done the work. The diligence was thorough.

Custodian And you weren't short on experience reading category creation attempts.

James No. This is our third fund. We've seen this pattern many times. We thought we knew what a genuine category transition looked like.

Custodian Then here is the question I actually want you to sit with. What if you read the metrics accurately — and read the wrong layer?

James goes still. He has been framing this as a thesis error or a timing error. That framing just shifted.

Step Two — Causality

Custodian The early deployments that tripled revenue — describe them. Who were the customers? What conditions were present?

James Early adopters. Organisations with dedicated innovation budgets. In most cases the founder or a senior team member was directly involved in the implementation. Pricing was below what the product would eventually need to sustain. The executive sponsors were personally committed to the new approach.

Custodian And when the company scaled — when they moved beyond those conditions?

James Normal procurement timelines. Unselected buyers. Standard IT integration requirements. Pricing pressure from every direction. The founder couldn't be present in every deployment. The outcomes that had been produced in the early conditions didn't reproduce at the same level.

Custodian So the outcomes in the early deployments were genuine. The conditions that produced them were not representative of what the full market would require.

James Yes. That's precisely it. The demonstrations were real. The reproducibility wasn't yet established.

Custodian You didn't back a bad company. You backed a genuine demonstration that hadn't yet been tested against the conditions that determine whether it scales. The metrics were accurate. They measured outcomes. They didn't measure the conditions under which those outcomes were produced. Those are two different layers. And at the moment of investment — when the narrative is most compelling, when the social confirmation is strongest, when the pressure to commit is highest — those two layers look identical from the outside.

James So it's not that my structural analysis was wrong.

Custodian Your structural analysis of what the category could be was sound. What you couldn't see was whether the demonstrated superiority would hold when the protection was removed. Not because you were insufficiently rigorous. Because the diagnostic question you were asking — how strong are the results — doesn't surface that information. The question that surfaces it is different.

James Under what conditions were the results produced.

Custodian That's the question. And it's answerable at the time of investment — before the outcomes you needed to evaluate them had time to accumulate.

"The metrics were accurate. They measured the wrong layer."

Step Three — What You Would Have Done Differently

Custodian If you had seen that clearly before the Series A — not as a vague due diligence risk, but as a specific structural condition: outcomes produced inside protected deployments, not yet tested against the full market's inertia — what would you have done?

James I still might have invested. The structural thesis was real. But I would have structured the investment differently. Milestones tied to reproducibility outside protected conditions, not just revenue growth. Earlier visibility into whether expansion was happening with unselected customers. And I would have been having different conversations with the founder — not about growth rate but about the conditions of the deployments that were driving it.

Custodian So you wouldn't necessarily have passed.

James No. But I would have invested in a different understanding of what I was backing. I would have known I was backing a demonstration that needed to become a transition — and I would have built that distinction into how I was monitoring the position.

Custodian Notice what just happened. The investment wasn't wrong as a structural bet. The layer at which you were evaluating it was the problem. You paid a significant cost after the commitment — a cost you could have accounted for before it, if the conditions of the demonstration had been visible while they were still actionable.

Step Four — A Live Commitment

Custodian You have a term sheet on your desk right now. What's the investment you're closest to committing to?

James A pre-seed company in the supply chain intelligence space. The founder's thesis is that real-time supplier data has made the existing category — manual vendor risk management — structurally obsolete. The product surfaces vendor risk in a form operations and finance leadership can act on without requiring a procurement specialist to interpret it. We've seen the demo. The outcomes are impressive. Two reference customers, both enthusiastic. The founder is exceptional. The structural thesis is compelling. I'm inclined to lead the round.

Custodian Tell me about the two reference customers. How were they selected? What conditions were present in their deployments?

James Both were introduced through the founder's network. Both have leadership teams that were already aware of the vendor risk problem and motivated to address it. The founder was deeply involved in the onboarding of both. Pricing was at a significant discount to where the company intends to be.

Custodian So the outcomes you're evaluating were produced inside organisations that were selected for receptivity, with the founder present, at pricing that doesn't reflect the full commercial reality. That is what the conditions of demonstration look like before they've been tested against the full market.

James goes quiet. He has seen this before. He has just described it on his own terms.

Conditions examined — before commitment

Both reference customers were identified through the founder's existing network and were already motivated to solve the problem the product addresses. That is selection for receptivity, not sampling from the full market population. The outcomes they produced demonstrate the product's capability inside favourable conditions — not whether those outcomes will reproduce for an unselected buyer who encounters the product through normal commercial channels.
The founder's direct involvement in both deployments means the outcomes reflect a combination of product capability and founder judgment, relationships, and presence. What happens when the product is deployed by a customer success manager, without the founder in the room, inside an organisation whose procurement team has standard requirements and no prior relationship with the founding team — is a different and currently unanswered question.
The pricing discount means both customers made their adoption decision in conditions that will not reflect the full commercial environment. A buyer who adopts at a significant discount is not demonstrating that the product creates enough value to command its intended price in normal procurement. The pricing conditions that made adoption easy may have been as important as the product itself in producing the outcomes you're evaluating.

Custodian Were any of those conditions visible to you before now?

James All three were visible. I noted the founder involvement and the network selection in diligence. I didn't treat them as conditions that determined the layer I was reading. I treated them as normal features of early-stage deployment. Which they are. The question I wasn't asking was whether the outcomes would reproduce outside those features.

Custodian Does it change what you're about to do?

James It changes what I investigate before I commit. I want to understand what happens with a customer the founder didn't introduce — someone who came to the product through normal channels, without a prior relationship, at closer to the intended price. I want to understand what the deployment looks like without the founder in the room. Not because the current customers invalidate the thesis. Because they don't yet answer the question I actually need to answer.

Custodian You just moved the evaluation from how strong are the results to under what conditions were the results produced. That's the diagnostic shift.

"The question isn't how strong are the results. It's under what conditions were the results produced."

What Just Happened

James didn't receive a checklist. He wasn't told to be more careful or to add another diligence step. The Custodian didn't challenge his structural analysis or his read of the founder.

It did something structurally different: it moved the evaluation from the outcome layer to the conditions layer. The outcomes were always going to look the same — whether the company was a genuine transition or a protected demonstration. The Custodian changed which layer James was reading.

He's not making fewer bets. He's making bets from a different diagnostic position. And the difference between those two positions is not visible in the pitch deck, the revenue chart, or the reference calls — which is precisely why it matters.

In category creation investing, the most expensive mistake is not backing a bad company. It's backing a good demonstration and discovering — eighteen months later, after the follow-on decisions are already made — that the conditions which produced the demonstration were not the conditions the full market presents. Seeing that distinction before committing is not a luxury. It's the structural condition for investing in transitions rather than demonstrations.

You Read the Metrics Accurately.