The Room Without a Door

I should have clocked it on day one. The moment Pieter Schneider said, “We’ll know success when we see it,” I should have closed my notebook, smiled politely, and walked directly into the sea.

Instead, I wrote it down. Success = TBD. Underlined it twice. Drew a small question mark in the margin that, over the following six weeks, would grow into something resembling a cry for help.

Delos Interactive made immersive training platforms - the kind where you’d put a headset on and learn to operate a forklift without actually flattening anyone. Their VP of Product Engineering, Pieter, had reached out through our website form. Inbound. Qualified. Budget mentioned in the first email. I remember thinking: This is the clean deal.

I am, as you may have gathered, an optimist of the most dangerous variety.

The discovery calls felt like they went well, which I’ve since learned is not the same thing. Pieter was articulate. He described pain points with the fluency of someone who’d been living with them for years: deployment bottlenecks, content authoring lag, analytics that were basically decorative. I filled three pages of my notebook. Even used two different coloured pens.

Our AE, Madison Carter, was thrilled. “They’re ready. Let’s get a PoV scoped.”

The scoping document I put together was thorough. Success criteria. Timeline. Integration checkpoints. Exit conditions. I’d even included what wouldn’t be in scope, because I’d been burned before by the infinite elastic PoV that stretches until everyone involved has aged visibly.

Pieter read it. Said it was “very comprehensive.” Then asked if we could “keep things flexible.”

Flexible =???

There’s a specific dread that lives in that word. It sounds reasonable. It sounds collabourative. It is, in practice, a trap door in the floor of every plan you’ve ever made. But I said, “Of course, we can iterate as we go,” which is the thing you say when you’re too cowardly to ask, “Flexible in what way, specifically, and at whose expense?”

Weeks one and two passed in the usual fog of false confidence. Sandbox deployed. Pieter’s engineers - Henrik and Sofia - ran configurations. Sofia filed bug reports, two of which were feature requests. Standard stuff. Status: green. Dashboard: all checkboxes ticked.

Then at the week two check-in, Pieter said, “We’re making progress, but we’re not quite there yet.”

I asked where “there” was.

“We’ll know when we get there.”

I wrote Kafkaesque in my notebook and drew a small door with no handle.

By week three, the goalposts hadn’t moved so much as dissolved. I started keeping a secondary list: things Pieter had mentioned wanting to evaluate, in no particular order and with no particular commitment. Authoring speed. Deployment automation. Analytics granularity. “The feel of the platform.” Something called “cognitive load optimisation” that I’m still not sure is real. Eleven items. None prioritised. None with a success threshold. Less an evaluation framework than a wish list addressed to a Santa Claus who doesn’t exist.

And I didn’t want to admit: I knew. Not in week three - I knew during that first scoping call, when Pieter said “keep things flexible” and I heard the creak of a door that wasn’t properly hung. I chose momentum over alignment. Because momentum feels like progress. Because the deal was inbound with budget in the first email, and I didn’t want to be the person who slowed it down with uncomfortable questions.

I chose comfort over clarity. And comfort, in a PoV, is a debt that compounds weekly.

Week four. The consequences arrived all at once, wearing a pleasant smile.

Cassie Beaumont, Head of Learning Design, appeared on the call like a in a novel you were already struggling to follow. She had requirements from what seemed to be an entirely different evaluation happening in a parallel dimension - accessibility compliance across nine languages, offline mode, integration with a bespoke LMS built by a developer who’d since left the company and possibly the country.

“We’re still defining what success looks like,” Pieter said.

I drew a hamster on a wheel. Unprofessional but accurate.

Meanwhile, Sofia was doing excellent work - genuinely impressed with the authoring tools, camera on, leaning forward. Henrik had gone silent, his last message a solitary “k” in response to a configuration guide I’d spent two hours writing. When I asked Sofia about him, she hesitated. “Henrik doesn’t really know what he’s supposed to be doing.”

Nobody did. We were evaluating everything and therefore nothing.

Madison called. Not Slacked. Called.

“My manager’s asking about Delos. Is it tracking?”

“Sofia loves us. Henrik’s a ghost. Pieter’s still defining success. And there’s a new stakeholder who wants us to integrate with an LMS that may have been coded in a fever dream.”

“What do we do?”

“I need to have a conversation with Pieter that I should have had a month ago. The kind where I say things that are true instead of things that are comfortable.”

But that week, something worse happened: the PoV started to feel like it was going well. Sofia demoed prototypes. Cassie was impressed. Pieter sent an email saying the team was “energised.” Henrik surfaced long enough to ask about API rate limits. Madison updated her forecast. I updated my dashboard. Green everywhere.

Except none of it connected to anything. Sofia’s work was untethered to any criterion. Cassie’s enthusiasm was based on capabilities we hadn’t agreed to deliver. Henrik’s question revealed he’d been building a rogue integration entirely outside sandbox parametres.

Activity everywhere. Direction nowhere. I was still confusing movement with progress, believing that if enough good things happened near each other, they’d somehow coalesce into a decision.

Week five. Cassie emailed at 7:14am on Monday. Subject: “Quick thought.” Nobody who writes “quick thought” has a quick thought. She wanted to know if the platform could support real-time biometric feedback - heart rate, eye tracking, stress indicators - piped into the analytics dashboard.

I stared at it for a long time. Then I did something I should have done on day one.

I called Pieter directly. No agenda. No deck. Just a question.

“Pieter, what does Delos actually need this platform to do? Not everything it could do. Not everything that would be nice. What’s the one thing that, if we solved it, would make this worth buying?”

Silence. Long enough that I checked my phone hadn’t disconnected.

“Content authoring,” he said finally. “We lose three weeks every time we build a new module. If you can cut that in half, nothing else matters.”

One sentence. Six weeks of fog, burned away in one sentence.

“Then why didn’t we start there?”

He laughed - not unkindly. “Because I didn’t know that’s what I thought until you asked me that way.”

I sat with that for a moment. Then I said, “I need with you. We’ve spent five weeks evaluating eleven things and committing to none of them. Sofia’s built something brilliant, but it’s not pointed at this problem. Henrik’s disengaged. Cassie’s evaluating a different product than the one we’re offering. If we keep going like this, you’ll end the PoV with a lot of impressive demos and no basis for a decision.”

Another silence. Then: “So what do you suggest?”

“One week. We take Sofia’s prototype, strip it back to the authoring workflow, and benchmark it against your current process. Real modules, real content, real numbers. If we cut your authoring time, you’ll have something concrete to take to procurement. If we don’t, we shake hands and I stop drawing hamsters in my notebook.”

He paused. “You’ve been drawing hamsters?”

“It’s been a long six weeks, Pieter.”

Week six was the shortest and the best. Sofia rebuilt her prototype around three actual training modules Delos was scheduled to produce that quarter. We benchmarked everything. Authoring time dropped from nineteen days to seven. Sofia’s exact words were, “This is embarrassingly better.”

Henrik reappeared. Turned out he’d been quietly miserable, not disengaged - he’d wanted structure and had been too proud to ask for it. Given a specific task, he was meticulous. He documented the entire integration pathway in two days.

Cassie’s accessibility requirements went into a Phase 2 roadmap. She was fine with it. More than fine - relieved, I think, to stop pretending everything had to happen at once.

Madison closed the deal three weeks later. Not the biggest contract of the quarter, but the cleanest. Pieter told her, during the signing call, that the turning point was “when your SE stopped being polite and started being useful.”

I wrote that down too. Underlined it twice. No question mark this time.

I keep the notebook. Page twelve still has the door with no handle. Page thirty-one has the hamster. Page forty has a single line, added after the close, that I look at before every new engagement:

Comfort is not clarity. Ask the question.

It’s not complicated. It never was. The room without a door had a door the whole time - I just had to stop being afraid to ask where it was.