What a home surface is for, the 2 rules I gave the designer who owns ours, and why workflow software can't run the consumer experiment model.
Once a product gets a home surface that offers useful springboards into deeper product functionality, every squad wants onto it, and that’s the point of the thing. A home surface is where a clinician lands before they’ve decided what to do, so it’s the one place in the product with enough reach to change what they do next. It’s also the one place where a bad call is visible to everyone, every day. The designer steering ours has spent the last few months in that tension, and a recent 1:1 turned into the clearest version of the rules I’d want any designer in that seat to carry.
Rule 1 is about what the surface is for. My framing, which he has sharpened, is that a home surface should springboard. I’m starting to recognize the pattern as a formula of sorts: a datapoint we already have, the moment it should surface, and an action the clinician would approve. “You spent this many hours on scheduling this month, and there’s a feature that fills those slots for you” is a springboard. “Here’s a feature, buy it” is a billboard. The difference is obvious to the person reading it, and clinicians in particular have a low tolerance for being sold to inside a tool they already pay for. He put it better than I did: we have to earn that trust before we start trading on it.
The value existed and people still couldn’t get to it. The hurdle was the structure, and structure is the part a home surface is made of.
Rule 2 is about who gets a slot. The designer owns the framing and the gates. If a team asks for a card because the surface will help customers discover something they’d benefit from, that’s a conversation. If a team asks for a card because a number needs hitting, that’s a no, and the designer has to be able to say it without escalating. I’d rather give someone that authority explicitly than watch a home surface fill up one reasonable exception at a time.
The part that surprised me was how much of the conversation was about testing, because testing is where the gate rule gets enforced or lost. At Facebook you’d run hundreds of experiments across the whole base and let the numbers pick the layout. That model breaks in workflow software. Our customers have routines, and the routines are load-bearing. One practice we spoke with keeps its team list sorted by the date each person joined, because that’s the order the admin team has memorized, and a switch to alphabetical, which any of us would call an improvement, would cost them time every day. You can’t run hundreds of live experiments on people like that. What you can do, and what he and his PM partner have pushed for, is treat an early-access window as the experiment: a small group, structured questions, layout changes made while people still expect change, and the rules settled before the surface goes wide.
A recent launch on another part of our product taught us why that sequencing matters. The value was there, and it was documented, and a meaningful number of customers still couldn’t find their way through the new structure to reach it. The hurdle was the structure, and structure is the part a home surface is made of. Content can be fixed in a card. Structure can only be fixed by moving things, and moving things on a home surface after launch is how you spend the trust you were trying to earn.
If you’re the designer holding a surface like this, the 2 rules travel: know what the surface is for, in a sentence you can say to a squad lead who wants on, and hold the gate yourself. And to the leaders reading, give that designer the authority before the requests arrive, because they will, and every one of them will sound reasonable.