What the data room cannot show

An investment thesis for an industrial technology company usually rests on a handful of commercial claims. Pilots will turn into fleet deployments. Design partners and letters of intent (LOIs) stand for paying buyers. The champion who ran the evaluation holds the budget. A channel will carry the product to plants the company cannot reach directly. The sales cycle in the model is the one buyers run.

A data room can show that these claims were made. It cannot show that they are true.

The contracts, the pipeline report and the management references all come from inside the company, and the market model sizes a category without saying whether these particular buyers sign.

The return now depends more on those claims than it used to. Bain's 2026 private equity report puts buyout holding periods at around seven years, up from five to six between 2010 and 2021, and estimates that a 2.5x return over five years now needs 10% to 12% annual EBITDA growth where 5% used to be enough [1]. Growth of that kind has to come from buyers who pay, which is the part of the thesis a financial model inherits rather than tests.

This paper sets out which thesis claims to test with buyers, how to turn them into pass/fail criteria before the first call, how to grade the answers, what a short diligence window can and cannot evidence, what to do when the target's customers are off limits before signing, and how the findings carry into confirmatory diligence and the 100-day plan. The work runs alongside a commercial due diligence (CDD) provider. It does not replace one.

Five claims recur in industrial technology deals. Each has a version that is true and a version that is not, and the documents look the same in both cases.

Pilots convert. The target reports a number of paid pilots and a conversion assumption. The pilots are real. What the documents cannot show is whether the budget that paid for a pilot has any authority over the fleet decision that follows it. In industrial buying, the pilot is often funded from an innovation or site budget, and the rollout needs a capital approval owned by someone who never saw the pilot. The mechanism is covered in Too successful to stop, too unproven to scale. For diligence, the point is narrow: a pilot count is evidence about pilots.

Design partners and LOIs are buyers. A design partner who received the product free, or at a discount in exchange for feedback, has shown interest in the problem. An LOI shows intent from the person who signed it. Neither shows that the function that owns the budget will pay the list price. The committee nobody mapped sets out why the people who sign an LOI are often not the people who sign the purchase order.

The champion holds the budget. Management references are usually champions: the engineer or operations lead who brought the product in. They are sincere and they are selected. In most industrial purchases the champion can recommend, and someone else approves.

The channel will carry it. Many industrial technology plans assume that a system integrator, distributor, equipment maker, cloud marketplace or developer community will sell, implement and support the product at plants the company cannot reach. That assumption has three parts: the partner holds the relationship, the partner can implement, and the partner earns enough to prioritize the product. The data room usually evidences the first part with a signed partner agreement and says nothing about the other two. Nor does it show distributor concentration, territory terms, integrator capacity, marketplace co-sell share, or trial-to-paid conversion, which is where channel-led revenue quality is decided. How to test each is in Through the integrator.

The sales cycle is the real one, and win-loss is where it shows. The model's cycle length is usually derived from the deals that closed. Deals that stalled, or were lost after a successful evaluation, are the better evidence of the real cycle, and they are rarely in the pipeline report in any form a buyer could interpret.

None of these claims is exotic. They are the ordinary commercial assumptions of an industrial technology business. They are also the claims that decide whether the growth in the model happens.

The standard diligence inputs each answer a real question. None of them was designed to answer these five.

Management references confirm that satisfied customers exist. They cannot tell you about the buyers who did not become customers, and they are chosen by the people whose forecast is being tested.

Expert-network calls are fast and useful for orientation. Each call gives one person's opinion for an hour. The opinions are not scored on a common rubric, and nobody grades them against criteria set before the calls began, so ten calls can support almost any reading of the thesis.

A market model sizes the category, the growth rate and the share a company of this kind might hold. Whether a particular plant's operations director will approve this product at this price is a different question, and it is the one the forecast depends on.

A CDD provider covers the whole target: market, competition, customers, pricing, and the plan. That breadth is the value of the work. Buyer evidence on the five claims above needs depth in a few places instead of coverage across all of them, and it fits best as a separate workstream with its own questions and its own grading.

The first step happens before any interview. Each thesis claim the return depends on is rewritten as a statement that could fail, with the evidence that would count as a pass stated in advance.

A thesis line such as "strong demand from mid-size discrete manufacturers" cannot fail. It becomes testable when it names the buyer, the behavior and the threshold. For example: "Among operations directors at mid-size discrete manufacturers in the target segment, at least six of ten interviewed confirm an unfunded problem the product addresses, and at least four confirm a budget line that could pay for it within the next planning cycle."

The numbers in that example are illustrative, and ten is a floor, not a target. A Verified grade on a claim the return depends on usually needs more interviews than that, and the criterion should say how many. What matters is that they are written down before the first call, agreed with the deal team, and not redefined after the findings arrive. A criterion set afterward will always fit what was found.

Most theses contain four to six claims that carry the return. The rest are context. Diligence time spent on context is time taken from the claims that decide the outcome, so the list is short on purpose.

Four groups of people can test a commercial thesis, and each tests a different claim.

Current customers of the target, where contact is allowed. The question is not whether they like the product. It is whether the person who holds the budget for expansion would approve it, at what price, and on what timetable.

Buyers in the same segment who are not customers. These are the buyers the growth in the model has to come from. They test urgency, budget and the real buying process without any relationship with the target to protect.

Win-loss: buyers who evaluated and did not buy. Lost and stalled deals show where the sales cycle breaks. They are the most informative interviews in the set and the hardest to arrange.

Channel partners, where the plan depends on them. They test whether the partner holds the plant relationship, has the people to implement, and earns enough from the product to prioritize it.

Inside each account, the roles matter more than the number of logos. The budget holder, the operations or plant leader who would run the system, maintenance, and the IT and operational technology (OT) security owner each hold a different veto. An interview program that reaches only one of them tests only one veto.

Every interview is scored on the same five dimensions: urgency of the problem, its priority against other spending, whether a budget exists, whether this person has authority over it, and openness to this kind of product. A scored interview can be compared with the next one. An unscored conversation cannot, and that is the difference between evidence and a collection of opinions.

Each thesis claim receives one of three grades, and each grade names what it rests on.

Verified means the claim was confirmed in scored interviews with the people who make the decision, at a specific price and for a specific application. Surveys, internal consensus and the target's own documents do not qualify.

Benchmarked means the claim is supported by benchmarks, analyst material, interviews with specialists, or a smaller or less senior set of buyer interviews than the criterion required. It is useful context and it is not enough to justify the capital on its own.

Assumption means the claim is assumed rather than evidenced. It is stated plainly, with the risk it carries and a way to test it.

The grades are not a verdict on the company. A thesis can proceed with several Benchmarked claims if the price and structure reflect them. What the grades prevent is the common failure in which a Benchmarked claim is written up in the investment memo with the confidence of a Verified one. The evidence standard behind the grades is set out on the method page.

Diligence windows are short, and they are set by the deal process, not by the evidence. That is a constraint to plan around, not a reason to skip the work. The practical answer is to state, before work starts, which claims the window can grade and to what level.

The pattern below is an example, not a commitment. It assumes one buyer segment, warm access to buyers, and interviewees who can meet within days of first contact.

Thesis claim	Two weeks	Four weeks	About eight weeks
Problem urgency	Benchmarked	Verified	Verified
Decision authority	Benchmarked	Verified	Verified
Willingness to pay at a named price	Assumption	Benchmarked	Verified
Displacement of the current solution	Assumption	Benchmarked	Benchmarked
Channel viability	Assumption	Assumption	Verified

Two weeks gives a Benchmarked read on whether the problem is urgent and who decides. Confirming both with the budget holders takes about four. Price takes longer, because it depends on interviews with people further from the champion, and on a second round once the first round has shown who those people are. Channel viability goes straight from Assumption to Verified because it needs a second group of interviews, with partners, once the buyer interviews have shown which partners matter; until those happen there is no partner evidence to grade at all. Displacement often stays Benchmarked even with more time, because a buyer's stated intent to switch is weaker evidence than a switch.

Sample size explains part of the pattern. In a well-known study of needs research, Griffin and Hauser interviewed 30 customers of portable food-carrying devices and, from that data, hypothesized that 20 to 30 one-on-one interviews are needed to capture 90% to 95% of customer needs [2]. A short diligence window rarely allows that many interviews per segment. At ten interviews, a single strong opinion can move the result, which is why the grades in a short window stop at Benchmarked for anything beyond urgency and authority.

Warm access changes the pattern more than time does. When an investor or management team introduces ten qualified buyers, each grade usually arrives about one window sooner. Cold access, several segments or regions, and restrictions on contacting customers move it later.

Before signing, the target may not allow its customers to be contacted, or may allow only the references it selects. That is common and reasonable. It limits what can be graded Verified, and the brief should say so.

Three things still work inside that limit.

Interview the segment instead of the customer list. Non-customers in the same segment test urgency, budget, authority and price as well as customers do, and better on the question of where growth will come from. They cannot test satisfaction with this product, and they do not need to.

Interview selected references on different questions. A reference chosen by management will be positive about the product. The same person can still answer who approved the purchase, what the expansion decision requires, and which functions have to sign. Those answers are rarely rehearsed.

Mark each grade with its evidence type. A Verified grade that rests on non-customers in the segment and a Verified grade that rests on the target's own expansion buyers are different findings. The brief states which kind of evidence each grade rests on, so the investment committee (IC) can weigh them correctly.

Whether the investor's firm is named to the people interviewed is agreed at scoping. Before signing, this work usually runs without naming the firm.

The output is a short brief written for a reader who was not in the interviews. Each thesis claim carries its grade, the interviews behind it, and the risk it leaves open. The format is described on the specimen brief page.

The grades then do three jobs.

They shape the IC discussion. An Assumption on a claim the return depends on is a price, structure or Stop question, and it is better raised before signing than discovered in the first board meeting.

They set the confirmatory diligence agenda. Benchmarked claims are the questions to close once customer access opens up after signing. They are specific: which buyers, which roles, which criterion.

They become the first tests in the 100-day plan. The claims still open at close are the commercial hypotheses the value creation plan inherits. Writing them down as tests, with the same pass/fail criteria, means the board can see within two quarters whether the plan is working on evidence or on momentum. What to do when one of those tests fails is the subject of a companion paper, Fund, fix or stop.

This is commercial due diligence in the narrow sense: evidence from buyers on the claims the return depends on. The buyer-evidence workstream is narrow by design. It takes the few thesis claims a market model does not reach and tests them with the people who would sign, on criteria set in advance.

In practice it runs next to the CDD provider's work, with a separate brief. The CDD provider's market view defines the segments. The buyer interviews then test whether the growth assumed in those segments is supported by buyers who hold budget and authority. Where the two disagree, the disagreement is itself a finding worth taking to the IC.

The same applies to expert-network calls. They are a good way to learn a market quickly. Scored buyer interviews answer a different question: whether this buyer, at this price, will sign.

Two developments raise the stakes for industrial technology deals.

The first is the return requirement described above: longer holds and a higher growth bar mean the thesis has to be right about buyers, not only about the market [1].

The second is the number of industrial AI products whose commercial case has not been tested at scale. In a Deloitte survey of 600 executives at large manufacturers with US headquarters or operations, 92% said smart manufacturing would be the main driver of competitiveness over the next three years, while 29% reported using AI or machine learning at the facility or network level [3]. S&P Global's 451 Research found that the share of companies that discontinued most of their AI initiatives between proof of concept and production rose from 17% in the fourth quarter of 2023 to 42% in the fourth quarter of 2024 [4]. Gartner predicts that more than 40% of agentic AI projects will be canceled by the end of 2027, citing escalating costs, unclear business value or inadequate risk controls [5].

None of these figures says that a particular target will fail. They say that the gap between interest and paid deployment is wide in this market, and that a thesis built on interest carries more risk than its documents show. Validating physical AI before the capital is irreversible covers the additional questions physical AI raises, including liability and safety.

Use these as a starting list. Each question is asked of the role named, and each answer is scored.

Pilots convert

• To the budget holder: which budget paid for the pilot, and which budget would pay for the rollout? Are they the same?
• To the plant or operations leader: what would have to be true at the second and third site for you to approve it there?

Design partners and LOIs are buyers

• To the design partner's budget holder: would you pay the list price today? If not, what would change your answer?
• To the LOI signatory: who else has to approve before a purchase order is issued?

The champion holds the budget

• To the champion: who approves spending of this size, and have they seen the product?
• To that approver, where reachable: what problem would this have to solve for you to fund it this year?

The channel will carry it

• To the partner: how many of your customers have asked for this? Who on your team would implement it, and what do you earn from it compared with what you already sell?
• To the end buyer: who do you expect to call when it breaks?
• To the deal team: which two partners carry the largest share, on what territory and stocking terms?

The sales cycle is the real one

• To a buyer who evaluated and did not buy: where did the decision stop, and who stopped it?
• To a buyer who bought: how long from first meeting to purchase order, and what happened in the longest gap?

Deal-process questions for the deal team

• Which thesis claims does the return depend on, in order?
• Can the target's customers be contacted before signing, and on what terms?
• Can the deal team or management introduce ten qualified buyers in the target segment?
• Which claims will the IC need at Verified, and which can it accept at Benchmarked with a price or structure response?

1. Bain & Company, "Global Private Equity Report 2026," press release, 23 February 2026. https://www.bain.com/about/media-center/press-releases/2026/private-equity-resurgence-gathers-steam-as-new-era-challenges-firms-to-enhance-value-creationbain--company-global-pe-report/
2. Abbie Griffin and John R. Hauser, "The Voice of the Customer," Marketing Science 12, no. 1 (1993): 1 to 27. https://pubsonline.informs.org/doi/10.1287/mksc.12.1.1
3. Deloitte, "2025 Smart Manufacturing and Operations Survey," 600 executives at manufacturers with US headquarters or operations (revenue of $500 million or more, 1,000 or more employees) surveyed August to September 2024, press release via PR Newswire, 1 May 2025. https://www.prnewswire.com/news-releases/deloitte-survey-reveals-smart-manufacturing-is-driving-advantage-but-needs-focused-investment-and-implementation-302443462.html
4. 451 Research, S&P Global Market Intelligence, "Voice of the Enterprise: AI & Machine Learning, Use Cases 2025," online survey of 1,006 respondents in North America and Europe fielded 21 October to 25 November 2024. Figures read in the 451 Research Discovery Report "Best practices for delivering AI at scale," May 2025, https://www.verizon.com/business/resources/T51/reports/ai-at-scale-best-practices.pdf, and as reported by CIO Dive, 14 March 2025, https://www.ciodive.com/news/AI-project-fail-data-SPGlobal/742590/.
5. Gartner, "Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027," press release, 25 June 2025. https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027

What is not yet sourced

• The S&P Global figure (reference 4) was read in S&P's own May 2025 Discovery Report. The full Voice of the Enterprise study is paywalled and was not reviewed.
• The Griffin and Hauser finding (reference 2) comes from needs research in one consumer product category (portable food-carrying devices), and the authors framed it as a hypothesis from their data. Applying it to buyer interviews in industrial diligence is our reading, not their result.
• The two, four and eight week pattern is an indicative example from Thing Company's method, not a measured result.
• The five recurring thesis claims, the observation that pilots are often funded from a different budget than the rollout, and the value of lost-deal interviews rest on Thing Company's diligence experience, not a published study.

Thing Company is an independent market validation practice for industrial technology. We test whether a buyer exists at a price that works. Each engagement is led by a senior practitioner from Thing Company or its expert network.

For investors, the Diligence Sprint applies this standard inside a deal timeline. Working on a deal? Start with a Sprint Assessment.

Not ready to talk? Run the free thesis claims check self-check from the toolkit.