Introducing the POC Valley
Part 1: the brutal gap between "it works" and "we're deploying"
This is Part 1 of a six-part series on selling technology with a physical footprint into legacy industries: manufacturing, industrial, and operationally complex enterprises.
A quick note on who this is for. I work mostly with early-stage founders selling AI hardware into manufacturing. But the dynamics here apply more broadly: Industrial automation, robotics, AI, and agentic tools that need to run on a factory floor, not just in a browser. If you’re selling something that touches operations in a legacy industry, you’ll probably recognize what follows.
These companies face a specific problem. The technology works. The proof of concept impresses. And the deal still dies.
This series explores why.
The POC Valley
Large enterprises run an innovation funnel. The funnel runs from lab demo through POC engagement and validation, into offline pilot, then site and floor acceptance, and finally multi-plant scale.
Most startups die somewhere between POC and Deployment.
The graph below maps three things across this funnel. The blue line shows the survival rate of how many deals survive at the beginning of each stage. The yellow line shows startup investment. The orange line shows corporate investment.
The blue line tells the story. Survival drops drastically from POC to Deployment. Among the 75% deals that get a POC, only 5% make it to deployment.
"We've done a lot of POCs, but it's actually been hard to land a deployment. After a POC, they often ask for more POCs. We can never get to deployment in the factory. How do we do that?" — Founder, GenAI startup selling into automotive manufacturing
The question isn’t whether you can demonstrate value. If you couldn’t, you’d have been disqualified at the Demo stage. The question is what happens next.
Enterprise sales cycles for industrial technology average 12–18 months. Many deals stall indefinitely somewhere in between. That timeline points to something structural. Your buyer isn’t failing to see value. They’re failing to decide.
So what’s happening inside the enterprise during those long, silent months between “we love it” and “we’re deploying it”?
The standard advice, “qualify harder,” “find the champion,” “map the stakeholders”, is correct but incomplete. It treats the problem as a sales execution gap. This series explores the other half: the organizational dynamics that kill deals even when the technology works.
Four Reasons PoC Don’t Convert
1. No Fire: Business urgency is missing
From the outside, the buying signals look right. The technical team is engaged. They’re asking great technical questions. They’ve even scheduled follow-up meetings to review the results.
It feels like momentum.
Then three months pass. You still don’t have the right business case to work on. And you start to realize something you probably already knew: technical excitement and business urgency are different animals.
The engineer evaluating your system genuinely believes the technology is better than the thing they’re duct-taping together. They’re not misleading you. But they can’t convince their bosses in production that it moves the needle on KPIs. In fact, they can’t make anything happen beyond the commitment of a lab unit, which is all their R&D budget allows this year.
AI has become a boardroom talking point:
“Every boardroom in the world is thinking about how AI can enhance efficiency.” — CFO, Series B industrial technology company
That pressure creates immense POC demand. But boardroom interest and operational urgency aren’t the same thing. The board wants to see innovation happening. Operations wants problems solved. Sometimes these overlap. Most of the time, they don’t.
“For traditional manufacturing, it’s difficult to make the tradeoff. Unless it is some moonshot, these AI solutions can feel incremental: how do you justify that against revenue-generating initiatives like territory expansion?” — VP, Corporate Development, leading insurance company
So you end up with two buckets of opportunities. From outside the organization, they look identical. One bucket holds innovation theatre: an activity that satisfies a board mandate but was never going to convert. The other bucket holds real deployment potential: funded initiatives tied to operational pain.
The problem? Both buckets send the same early signals. Same engaged engineers. Same technical validation. Same “let’s schedule a follow-up.” The difference only reveals itself when you ask the uncomfortable questions: Timeline. Budget authority. What happens if this doesn’t get solved by Q3.
A real example.
Five years ago, I was leading a POC for a startup with a Tier 1 seat maker. The project was fabric defect inspection: could our tool catch what their current process was missing? We were comparing our solution against their in-house hard-coded CNN. We captured the key features with high accuracy. The conversations were almost always about accuracy.
Then we asked about timeline and budget authority. We never heard back.
Five years later, a client of mine mentioned they’d just wrapped a POC with the same Tier 1 supplier. Different startup, same inspection challenge. And the same questions about cycle time, robotic movement, how inspection fits into the line were still outside the scope.
The reason the project never moved forward is simple: the current solution was working fine. At least, not badly enough. The urgency wasn’t there five years ago. It still isn’t there now. And the right team was never in the room, which leads to the next gap.
2. No Alignment: Not everyone wins when you show up
Even if your tech validation is tied to a real business problem, and there’s a loosely defined path from here to deployment, you can still hit walls. Along the deployment path, you’ll meet stakeholders whose jobs get harder when you show up.
Every deployment touches multiple functions. Each one has a different question that needs answering before they’ll sign off:
Operations asks: Does this disrupt my line? What’s the downtime during install?
IT/Security asks: What data leaves our network? Who owns it?
Legal asks: What happens if your system fails and we ship defective product?
Finance asks: What’s the ROI timeline? How does this compare to other capital requests?
Plant Manager asks: Who’s accountable if this doesn’t work?
Miss one of these, and the deal stalls, even if everyone else is aligned. And often, the person blocking you isn't even in the room during your POC. They show up later, with questions you never anticipated, and priorities that have nothing to do with your technology.
3. No System: Built for suppliers, not startups
“Big automotive firms are used to humongous sourcing agreements, about 20, 30 pages with clauses like ‘if your part fails, you pay us $20 million.’ As a startup, you cannot sign that. You have no company afterwards.”— Former Corporate Development, Nissan
Large manufacturers have procurement systems built for large suppliers. Not for you.
To work with them, you need to get on an approved vendor list. You need to survive liability clauses written for billion-dollar companies. You need to negotiate contracts designed for suppliers shipping thousands of SKUs a year. Nobody has adapted these templates for a $20,000 pilot from a three-year-old startup.
When that Master Service Agreement lands in your inbox, you’re not close to done.
These systems exist to manage risk. But for you, it means adoption isn’t just a technical decision. And a political question: who owns the blame if something goes wrong?
Some of these terms are negotiable, if you know to ask. Liability caps can sometimes be scoped to the pilot. IP clauses can be narrowed. Payment terms can be adjusted for early-stage vendors. But you have to ask before the MSA hits legal review. Once it does, you're negotiating against a process, not a person.
“Unless you have a champion who can get things done in the company, it’s not going to move ahead. You need a person who can remove all the bottlenecks inside: who can work with legal, who can work on data agreements.” — Former Corporate Development, Nissan
You can’t tell from LinkedIn whether someone is a real champion or just enthusiastic. That only becomes clear in conversation and sometimes only after months of wasted effort. And when your champion leaves? Your project usually dies with them. The replacement has different priorities, different relationships, and no ownership of what came before.
“I was originally not interested in passing this information to the right team, especially because I have no interest in AI. But since you explained the concept so clearly, in an easy-to-understand and highly relevant way, I felt motivated to help you with the introduction to the right team.” — Former R&D Director, LG
There's no formula for finding a champion. But you can narrow the search. Look for someone director-level or above. An operator, not a strategist. Long tenure or at least someone who knows where the bodies are buried. And genuinely curious about innovation, not just checking a box. That won't guarantee anything, but it improves your odds.
4. No Path: Innovation is not connected to deployment or scale
How a company structures innovation determines whether your deal has a path forward. And that structure varies wildly.
“In Nissan, we had a venture capital fund at the corporate level across Nissan, Renault, and Mitsubishi. But manufacturing innovation? That fell under a completely different team. The corporate innovation teams were not talking with the manufacturing innovation team.” — Former Corporate Development, Nissan
Some companies prefer to own innovation and R&D in-house. And even within the same company, the structure shifts. Some centralize manufacturing innovation across all plants. Others push it down to the plant level.
These structures shift over time, and companies reorganize frequently.
“At Cargill, the philosophy used to be that central manufacturing drove innovation. Then we had a reorg, and now every business figures out its own stuff. You need to ask who is actually in the driver’s seat.” — VP Corp Development, Cargill
Here’s where founders often optimize for the wrong thing. Central innovation teams are easy to find. They’re on LinkedIn. They go to conferences. They run the CVC that takes your intro call. So that’s where most startups start.
But central teams are often the most disconnected from deployment. You can get validated at corporate and still have no path to a single plant floor. Deals die slowly in these structures. You’re selling to people who can say yes to a POC but can’t make anyone else adopt it.
The harder path might actually be faster: start at the plant. With distributed deployment, there are fewer gatekeepers. The plant manager who says yes can actually make it happen. The catch? No leverage. You win one site, and you’re starting from zero at the next. Scaling becomes a series of one-off sales, not a rollout.
So the strategic question is this: do you want easy entry with a long, uncertain path? Or hard entry with a shorter path but manual scaling?
There’s no universal answer. But you should know which game you’re playing before you’re three months into a POC with a team that was never going to deploy anything.
Conclusion & What’s Next
McKinsey introduced “pilot purgatory“ in 2018. The numbers have barely moved since. The POC valley isn’t a technology problem. It’s four organizational gaps: missing urgency, misaligned stakeholders, incompatible systems, and disconnected innovation structures. You can’t fix these from outside the organization. But you can learn to diagnose them earlier.
Part 2 of this series will dig into the first gap: how to qualify for business urgency before you’ve burned six months and $200K on a pilot that was never going to convert.
P.S. Some researchers dispute the prevalence of pilot purgatory; see LNS Research‘s counterargument. The dynamic I describe here is specific to startups selling into enterprises, not enterprise internal initiatives.
About this series
Hi! I'm Trista, a former founder, early GTM at UnitX, now a GTM strategist for technical founders deploying into legacy industries.
This series grew out of conversations with founders, operators, and enterprise leaders working at the intersection of hardware, automation, and manufacturing over the past 18 months. This series wouldn’t exist without those who shared their experiences often candidly, sometimes painfully. Special thanks to Pedro, Piyush, Etienne, Alex, and Fred, who allowed me to quote their insights that shaped this analysis.
If you’re building in this space, or buying, or stuck somewhere in between, I’d love to hear from you.
Glossary
POC (Proof of Concept)
A limited-scope trial where a vendor deploys technology in a buyer’s environment to demonstrate it works under real conditions. In manufacturing, POCs typically run on a single line or process. Success here proves technical viability, not commercial readiness.
Pilot
A broader deployment than a POC, often involving multiple lines, shifts, or use cases. Pilots test operational fit: Can this run in production without breaking things? Some companies use “POC” and “pilot” interchangeably; others distinguish them sharply. Clarify definitions early.
Tier 1 / Tier 2 / Tier 3 Suppliers
The supply chain hierarchy in manufacturing. Tier 1 suppliers sell directly to OEMs (e.g., Bosch, Denso, Magna in automotive). Tier 2 suppliers sell to Tier 1s. Tier 3 suppliers sell to Tier 2s. Each tier has different procurement processes, risk tolerance, and decision-making speed.
OEM (Original Equipment Manufacturer)
The company whose brand appears on the final product. In automotive: Ford, Toyota, BMW. In medical devices: Medtronic, J&J. OEMs sit at the top of the supply chain and often dictate terms downstream.
MSA (Master Service Agreement)
The legal contract governing your relationship with an enterprise customer. Covers liability, IP ownership, indemnification, payment terms, and what happens when things go wrong. MSAs are written for large suppliers, expect to negotiate if you’re a startup.
Champion
An internal advocate who actively pushes your deal forward. Not just someone who likes your product, someone with the organizational capital to remove blockers, navigate procurement, and keep the project alive when priorities shift.
Integrators (System Integrators)
Companies that design and install production lines for manufacturers. They assemble equipment from multiple vendors into a working system. In some deals, the integrator is your real customer; the OEM just signs off.
Cobot (Collaborative Robot)
A robot designed to work alongside humans without safety enclosures. Smaller, slower, and more flexible than traditional industrial robots. Common in assembly, inspection, and packaging applications.
Site Acceptance Test (SAT)
A milestone where the installed system is verified to meet technical specifications, typically in a controlled or semi-controlled environment.
Floor Acceptance Test (FAT)
The final milestone: the system works under actual production conditions, and the customer formally signs off. This is where deals close or die.
Innovation Theatre
Activity that satisfies a corporate mandate to “do innovation” without any intent or pathway to deployment. Common in large enterprises under board pressure to show AI progress. Looks identical to real opportunities from the outside.
Pilot Purgatory
A term coined by McKinsey (2018) describing the state where companies run endless pilots that never scale. The POC works, but organizational friction prevents deployment. This series is about escaping it.
Awesome stuff!
You nail so many traps that are so easy to fall into as a startup selling to corporates, Trista!