Mechatronics engineer turned product manager.

For years our moat was a drivetrain that designed out the most common failure point on a machine of this type. Competitors took the conventional route, which works until it doesn't, and when it doesn't you have an immobile robot in a live sewer and an operator wondering whether he did the maintenance right.

The previous generation proved it out: effectively maintenance-free, no failures on that path in the field. It's what sales led with and it's what customers paid for.

Then we launched a successor. Same architecture, tighter package. An assumption that had held comfortably in the older, roomier design stopped holding once everything moved closer together. Units started coming back with the drivetrain seized, sometimes taking more with it.

The tricky part was it was impossible to reproduce in house. We ran accelerated cycles under the conditions we thought were responsible and never saw the failure. We spent about two months trying to save the architecture: shields, geometry changes, reorienting components. Every iteration helped some units. None of them closed the failure mode.

Then I looked at the rate. It was in the low single digits, on a small production batch. Not normally a number you act on, especially when acting means destroying your own moat.

And here's the thing: the money wasn't the issue. Repairs were cheap and the total sat in the low thousands. Nothing on the P&L was forcing my hand.

But the pipe inspection community is well connected. Everyone knows everyone and everyone talks. A handful of unhappy contractors cost you more than a handful of sales.

And mind you, these guys are direct. They are either five-star or one-star reviewers. Soon enough we were hearing about it from customers who hadn't bought yet.

I moved the platform onto the conventional design. Which meant telling a sales team that had spent years selling the thing we were about to remove that our newest product would ship without it.

It wasn't well received. But we aligned that reliability is what ultimately mattered, as long as we were upfront about the trade-offs either way. The drivetrain was modular, so customers could still pick and choose.

• A maintenance routine published immediately, so customers already working could keep working.
• Free swaps for affected customers.
• A service notice to every owner of the platform, including the ones who hadn't failed, so they could inspect their own units.

Of course we don't want to advertise our own oversights. But customers actually appreciated the transparency, especially the ones who hadn't had issues yet.