It’s easy to see why gadget makers are so fixated on touchscreens. Swiping is intuitive. It enables sleek lines for a futuristic aesthetic. It’s the easiest way to banish bezels and maximize screen real estate. So I understood why Fitbit was chuffed when it introduced the inductive button on the Fitbit Charge 3. Begone side button protrusion, and behold the slim profile of a modern fitness tracker!

This is what an unforced error looks like.

With the Charge 3, Fitbit replaced the Charge 2’s physical button with a smooth groove that you could easily mistake as part of the overall design. Put your finger on the groove, and it activates an invisible touch sensor on the inside of the device. When I initially reviewed the Charge 3, it seemed to work: a short press brought me back to a previous screen, a long press summoned the shortcut menu, and it vibrated to let me know it had done my bidding. But something that isn’t a pain during a brief review period can become one over time.

Back then, Fitbits (and other trackers) didn’t have the most responsive touchscreens. Swipes weren’t always detected, or sometimes they didn’t register correctly. That can be mildly annoying, but it’s less of an issue if you know there’s a back button to undo your mistake. But the back buttons on older Fitbit smartwatches, like the Versa 2, often got stuck. (The company never had the sturdiest hardware.) That, plus a slimmer profile, is probably why Fitbit even thought up the inductive button in the first place.

Physical buttons are dependable and predictable. You know what they do, and you know when you’ve pressed them. You can feel them depress, and some even have a satisfying little click. But while Apple has shown us it’s possible to emulate a button on a laptop-size canvas, Fitbit shows how risky it can be to try the same with wearables.

In Fitbit forums, some users were so befuddled by the Charge 3’s button that wasn’t a button that they didn’t know where to begin. If you pop “Fitbit Charge 3 button” into Google, the top two results are variations on “where is the button on the Charge 3?”

It’s telling that users weren’t aware that the Charge 3’s button existed until they needed it, and I can see why. With the Charge 3’s larger touchscreen display and no other visible controls, it’d be easy to assume all you needed were swipes and taps to use the device.

The bigger issue: even those who discovered the button found using it unreliable. The haptic feedback, meant to confirm you pressed the button, ended up being a source of confusion. If it vibrated earlier than you expected, did that mean you botched a long press? If it vibrated later, did you accidentally do a long press when you meant to do a short press? Because I didn’t have any firm confirmation, I often ended up going back a screen when I intended to launch a shortcut or vice versa. That’s annoying even in a chill setting and extremely frustrating if you’re mid-workout.

Other users also reported issues with the button not working after a time. To fix that issue, you had to press the inductive button for 15 seconds while it was on the charger to restart the device. But if the button doesn’t work and you can’t tell you’re actually pressing it — how in the dickens is that supposed to work?

The tragedy is that no one really asked for the inductive button. Fitbit’s physical buttons on older devices would occasionally get stuck, but by and large, all this could’ve been solved by making better buttons. (As other wearable makers did.) Instead, Fitbit overengineered a “solution” that created new problems and then doubled down on it for future trackers and smartwatches. It popped up on the Charge 4 and the Versa 3 and Sense, where the inductive button’s flaws were even more apparent because Fitbit added a double press. Theoretically, that meant you could program another shortcut — provided you were able to figure out the difference between a single press, a double press, and a long press.

Personally, I could not. While I generally had a positive experience with the Sense and Versa 3, the button wasn’t part of it. Simply bending my wrist ended up triggering half a dozen unintentional shortcuts. Bending my wrist while typing? That would start a run. Stretching out my calves before bed with a downward dog? That was also a run. In the course of testing, I ended up starting 15 phantom runs that I then had to go back and delete.

For the most part, other smartwatch makers have settled on a mix of physical buttons and touchscreens. The Apple Watch Ultra, for instance, has a huge honking screen, but it also added a third physical button — the Action Button — to the mix. Even Apple, a company that likes to do away with buttons whenever the opportunity arises, understands how important physical buttons are on its smartwatches. Meanwhile, Garmin continues to update its fitness watch lines with touchscreens, all while reassuring athletes that its five-button navigation system ain’t going anywhere. Scroll and swipe when you want to but with the reassurance that physical buttons will be there when sweaty fingers, gloves, and laggy displays are an issue.

So it’s a relief that Fitbit, which has since been acquired by Google, has somewhat come to its senses. With last year’s Sense 2 and Versa 4, Fitbit went back to physical side buttons. But it may be too little, too late. The Pixel Watch is here, and as I wrote in my Sense 2 review, we’re one Pixel band away from Fitbit completely disappearing into the Google machine. Switching back to the physical button earlier wouldn’t have changed Fitbit’s ultimate fate. But it would’ve made its last few devices a bit more enjoyable to use. For plenty of people, that would’ve been enough.