Investigation · PRODUCT & TIMING
The Same Idea Failed in 2001 and Saved Us in 2020. Nothing About It Got Smarter.
7 min read · June 2026
Thesis
A launch is judged on what it does. It mostly lives or dies on when it showed up.
The Question
Grocery delivery was a punchline in 2001 and a lifeline by 2020. Webvan burned through close to a billion dollars and folded. Two decades later a customer who would not buy lettuce on a screen was doing it twice a week without thinking. The idea did not get smarter in between. The packaging barely changed. So the question that actually interests me is the one the post-mortems skip: how much of a product's fate is decided before anyone judges a single feature, purely by the moment it walked into the room?
The Hypothesis
There is a term in the academic work on market timing, temporal legitimacy. Plain version: a market carries a quiet sense of when a thing is supposed to arrive, and a product that lands outside that window gets marked down even when it is good. My hypothesis was simple and a little uncomfortable. For a lot of the launches we love to dissect, timing explains the result better than the spec sheet does. Not all of it. I am not claiming features do not matter. I am claiming we over-credit them, because the spec sheet is sitting right there and "the market was not ready" is harder to put in a slide.
How I Looked At It
I pulled four public cases where the timing is visible to anyone. The grocery arc, Webvan in 2001 against Instacart and the rest around 2020. Google Glass, consumer edition, 2013. Quibi, April 2020. Apple Vision Pro, February 2024. I was looking for one thing in each: the gap between when the company decided the market wanted this and when the market actually did. I stayed with public record only, because the whole point was to see if the timing story holds without inside access.
What I Found
The grocery arc is the cleanest, because the idea is the control variable. Webvan did roughly what Instacart later did, a generation too early, against a customer who was not ready and an infrastructure that could not carry the cost. Nobody made the concept cleverer by 2020. The moment arrived, and a pandemic kicked the door in.
Glass was not a bad device for what it was. It walked into a year when being filmed by a stranger's face felt new and threatening, and the culture coined "Glasshole" before the hardware got a fair hearing. The social permission did not exist yet. The same head-worn camera now lives quietly on factory floors and in surgery, where nobody feels watched, and the timing question never comes up.
Quietest and cruellest is Quibi. Short premium video for people with ten minutes between things, launched the week the world stopped having ten minutes between things. The commute it was built for vanished. Around 1.75 billion dollars raised, gone in roughly six months. You can argue the content was thin, and it was, but I do not think great content saves a product whose core occasion for use just disappeared.
Vision Pro is where I have to be honest about my own argument. It is the case I trust least, because the 3,499 dollar price is a confound I cannot cleanly separate from timing. What I can see in public is the shape. Announced into a market that had been promised spatial computing for a decade and had learned to be tired of the promise, shipped to soft demand, followed by reports of trimmed forecasts. The device is impressive. It arrived into exhaustion, not appetite.
Three of four held. The fourth is muddied by price, and I am leaving it muddied rather than pretending otherwise.
What It Means For Business
The feature roadmap gets all the meeting time and the timing question gets none, and that is backwards relative to how much each one moves the outcome. Before I argued about the spec, I would force four questions on any team telling me the market is ready. Has anyone outside the building actually asked for this yet, in words, with money. What has to be true about the customer's week for this to fit, and is it true now. If a competitor shipped this exact thing in eighteen months instead of now, would they look early or look late. And what is the cheapest version we could put in front of a real buyer to find out before we commit a year. If those questions land badly, the call is wait, and waiting is the decision I would defend, because being early costs exactly as much as being wrong and looks identical until someone else is late.
The Decision
What I’d Test With More Time
I would want to separate timing from price properly, using cases where the same product was effectively re-launched at a similar price into a different mood. I would also want to know how often "we were early" is a real diagnosis and how often it is a comfortable excuse for a product that was simply not good. I suspect it is both, in a ratio I cannot prove from the outside.