The Committee You Cannot See

This changes the sales question. Instead of asking which persona to sequence next, ask what the current relationship can legitimately teach you about the committee: who owns the technical risk, which evidence is missing, what has changed since the last conversation, and who else needs to be able to explain the decision internally.

A promising technical conversation from 18 months ago is not simply an old lead. It may contain a real design constraint, a qualification result, an integration concern, a program milestone, or an introduction to someone who was not yet involved. But after a long silence, the facts around the conversation have moved. The contact may have changed remit. The project may have changed phase. The original specification may no longer be the constraint. The relationship may be warmer, colder, or simply different.

Levin, Walter, and Murnighan found that people could obtain useful work-related knowledge from dormant contacts, with outcomes comparing favorably with current ties; reconnecting previously strong ties combined benefits associated with both strong and weak ties.6 The claim is not that an old technical contact will convert. The claim is that prior context can contain value that a current-status field cannot recover on its own.

Rondi, Levin, and De Massis add the boundary condition. Reconnection can fail when people do not refresh the tie and do not know whether they now see the relationship in the same way. Their research identifies remembering the shared episode, catching up, and perceiving the tie similarly as key elements.7 In industrial selling, that sequence becomes a practical memory protocol: retrieve the technical episode, learn what changed, then decide whether the relationship is ready for a next move.

The operating implication is modest but important: do not let the CRM become a graveyard of names that once mattered. Preserve the minimum context that lets a future seller understand why the conversation existed, what was learned, what remains unverified, and what permission the relationship seems to carry. A record that can be reopened without pretending the past is present is a better asset than a larger list with no memory.

Doney and Cannon's study of more than 200 purchasing managers developed and tested a model of how industrial buyers form trust in a supplier and its salesperson. Trust influenced anticipated future interaction, but after controlling for previous experience and supplier performance, it did not independently determine the current supplier selection decision.3 That distinction matters. Trust may open the door to continued evaluation; it does not replace performance evidence, technical fit, or internal consensus.

Zhang, Viswanathan, and Henke found in automotive and food-industry samples that purchasing agents' professional knowledge and ability to reach compromises shaped trust in the agent, while strategic communication affected trust in the buying firm.4 For a deep-tech seller, engineer-to-engineer credibility therefore has a practical shape: understand the problem precisely, communicate the implications clearly, and show where the evidence stops.

A current industrial buyer pulse points in the same direction, while requiring a clear sample boundary. In a Q1 2026 survey of 303 North American manufacturing decision-makers, independent third-party validation or certification was the most trusted early-stage proof type at 29%, followed by customer references from the industry at 20% and supplier technical documentation at 15%.5 This is a directional survey, not a universal market law. Its useful message is that supplier-authored material is one part of a proof system, not the whole system.

The practical test for every important claim is four questions: what exactly is being claimed; what artifact makes it checkable; which stakeholder can verify it; and under what conditions would the claim fail? The last question is often the most credible one. Engineers and operators do not need a perfect story. They need a story that survives contact with the system they have to own.

The most useful dormant contact is not necessarily the person who once showed the most enthusiasm. It is the person whose context makes the current problem more legible. They may know why the earlier evaluation stopped, which constraint was decisive, what the program is called internally, or which other functions entered the conversation later.

Reopening that relationship should aim first at orientation, not at a pitch. Ask what changed. Ask which part of the earlier problem is still real. Ask who would need to be comfortable before the work could proceed. If the contact cannot answer, that is useful information too. It tells the seller that the relationship is not yet a reliable bridge into the buying system.

The message is deliberately smaller than a sales sequence. It gives the recipient a way to correct the seller's memory, explain what changed, or decline. The system earns the next step by making the first step easy to evaluate.

Deep-tech networks are built in places where the right technical people are already concentrated: conferences, trade shows, technical workshops, standards meetings, and customer visits. The advantage is not simply access to more contacts. It is the density of shared context: a specific talk, test method, design constraint, failure mode, production problem, or peer connection gives the first follow-up a reason to exist.

Research on B2B trade fairs supports this relational reading. A study of visitor-exhibitor interactions at an international trade fair found that socialization episodes, including information and social exchange, help generate relationship quality and development; information exchange was linked to trust, while social exchange was linked to satisfaction and commitment.8 Earlier industrial trade-show research, based on data from 27 shows, treated lead-generation efficiency as a performance measure and found that booth-personnel quality, relative spend, and show focus mattered.9

A 2021 study of 362 Portuguese SME exhibitors likewise found that sales-force proactivity influenced networking capability and sales performance, and that continuous participation contributed to performance.10 These studies do not prove that every event contact will become a sale. They do support a narrower conclusion: event presence, staff behavior, and follow-up are part of the relationship system, not separate from it.

The event should therefore end with a relationship record, not a lead count. Capture the technical episode while it is still vivid. Separate what was heard from what was inferred. Note the evidence that was requested and the people who might need to see it. Then decide whether to reopen, learn, map, or wait.

A workable system does not need to turn every contact into a research project. It needs to preserve the few facts that change the next judgment. The loop is simple: retrieve context, verify what changed, map the relevant stakeholders, choose a permissioned next move, and record the new state.

For a founder-led or small sales team, the discipline is to keep the record compact. One line on the shared episode. One line on the change. One line on the proof or stakeholder gap. One line on the next permissible move. Anything more can be added when a live evaluation justifies it.

Technical committees do not need the same proof in the same format. Engineering may need a test method and operating envelope. Quality may need traceability and a standard. Operations may need maintainability, training, and failure recovery. Procurement may need continuity, terms, and supply evidence. Executives may need a decision summary that does not distort the technical caveats.

The seller's job is to make the evidence portable. That does not mean flattening it into a slogan. It means preserving the claim, the artifact, the relevant peer, and the boundary condition so that another stakeholder can inspect the same underlying logic. The relationship record should therefore hold proof alongside memory. A warm contact without shareable evidence is still a fragile bridge.

This is where engineer-to-engineer trust becomes operational. A technical peer can correct a claim, explain a trade-off, or say that the stated use case is not comparable. That correction is valuable. It narrows the story, but it also makes the remaining story more credible. A seller who records the caveat is building a better bridge than a seller who records only the positive result.

The boundary is as important as the reference. Doney and Cannon's findings remind us that trust can shape future interaction without replacing prior experience or supplier performance in the current choice.3 A reference package should therefore accelerate inspection and consensus-building; it should not be used to skip technical validation.