The housings live in a drawer now. Ana Ferreira took them out for the photograph and laid them across the table in the order they were made, twenty-odd of them in grey and terracotta resin, ending with one that looks close enough to a finished product to be sad.
“Everyone assumes I keep them as a warning,” she says. “I keep them because they’re good. That’s the whole point. They’re good and it didn’t matter.”
The product was called Tell — a rig-tension sensor the size of a pocket watch that clamped to a shroud and reported load to a phone. Tom Aldous had built the first prototype before Ferreira joined Halyard, and it was, by every account including hers, a real piece of engineering. Ferreira came to it from four years in Nexa’s hardware group, putting sensors on refrigerated trucks for people who very much wanted them measured. She spent fourteen months making Tell manufacturable: the housings, the tolerances, the salt-water survival that took six of those months on its own.
By October 2020 it was accurate to within 2%. Eleven boatyards had agreed to trial it. A sailing club had asked about fitting forty. The trial units were nine weeks out.
Three weeks at the yards
What ended it was not a spreadsheet. It was Ferreira deciding, late and reluctantly, to go and watch. She spent three weeks at the yards doing nothing but observing what riggers actually did with a tension number. Mostly nothing. The number confirmed a judgement they had already reached with their hands, in about a second, for free. The people who wanted Tell were owners, not riggers — and an owner buys one, once, and never again. Meanwhile every yard she walked into complained about the identical thing, unprompted: sail servicing was six weeks out and nobody would give them a date.
“The sensor answered a question no one was actually asking. The servicing backlog was a question everyone was shouting.”
Eleven minutes, standing up
She brought it to Aldous on a Tuesday in the middle of October, in the loft, standing up, because there were no chairs. He asked one question — whether the yards would pay for servicing on a date certain — and when she said seven had already asked what it would cost, he said fine. The conversation took eleven minutes. They did not sleep on it.
Ferreira thinks that mattered more than the decision itself. “A weekend would have let us find a reason. There’s always a reason. You can always build one more version.” Aldous took it badly for roughly four days and has since called it the best decision the company ever made, which Ferreira notes is an easy thing to say six years and six million dollars later.
The timing is the part she defends hardest. Killing it nine weeks before the trial, rather than nine weeks after, meant eleven yards did not have units on their boats and Halyard did not owe them a decade of support for a product it no longer believed in. “After the trial, it stops being a decision,” she says. “It becomes an obligation with a warranty attached.”
What fourteen months bought
- Roughly $40,000 of tooling and test equipment still in daily use, including a load rig no loft their size has.
- One patent application, abandoned in 2021 rather than renewed, at a cost of $1,100 to stop.
- Two hires made for Tell, both still at Halyard, neither doing anything resembling what they were hired for.
- A habit of instrumenting things, which is now how the loft measures and cuts.
She resists the tidy version of this — the one where the detour was secretly the plan. It wasn’t. Six of the fourteen months were spent after the first customer conversation that should have ended it. What she does now, twice a year, is call people who like her and ask them to buy something, and count only the ones who reach for a card. “Enthusiasm is free,” she says, putting the housings back in the drawer. “We had a mountain of it. You cannot cut a sail out of it.”