A missed disconnection notice isn’t a UX gap. It’s a customer who finds out their power is off when the lights don’t come on, not before.

I held design authority on this program from inside the team leading overall delivery and data integration, not from a seat on either team actually producing screens. Notifications were already shipping across six channels: an outage map, a desktop in-app banner on the customer portal, a separate mobile in-app banner inside the mobile app, email, SMS, and IVR. The split wasn’t arbitrary. The outage map only rendered on desktop, and email and SMS behaved differently enough by device that they needed separate channels to track. A seventh channel, push notifications, was just coming online with one category live. Two vendors split the work, a dedicated design execution vendor and a platform vendor’s embedded designers who owned the design system. Neither reported to me. I still had to get both building to one evidence standard, and every notification that didn’t land meant a customer picking up the phone instead, which was at five to fifteen dollars a call.

No One Owned Whether It Landed

Everything on this program shipped on time. Notifications fired on schedule across six channels, and every channel had a delivery team standing behind it. What nobody had was a shared way to say whether a notification that fired on schedule meant anything to the person receiving it. The measurement existed. The process for making five vendor teams answer to it did not, and closing that second gap turned out to be the real work.

A Gap No One Had Closed Yet

Getting Five Teams to Row Together

The harder problem wasn’t the measurement. It was getting five vendor teams, each running its own version of agile, to route design and research evidence through a shared process before a feature shipped.

I built the intake as cross-department accountability at the roadmap level, anchored by a shared Definition of Ready that made research findings part of sprint planning instead of an afterthought. It piggybacked on existing OKR readouts, which let every team see its work against the same evidence standard rather than just its own. Getting it live meant working closely with the product owner, process owners, and scrum master to build buy-in, since none of it held up in front of the teams who’d actually run it without their support behind it.

The design teams and their respective vendor organizations had no reason to adopt any of it. Neither reported to me, and both already had their own definition of done. I got them to the table by making the evidence standard useful to them, not by asking them to comply with it. I showed the design execution team which patterns were generating the most retest failures before a feature even shipped. I showed the platform vendor’s designers where their own design system decisions were driving down the efficacy ratio on channels they owned. Neither wanted a mandate from a team they didn’t answer to. Both wanted fewer defects showing up in their own queue, and the evidence standard was the fastest way to get there.

None of that existed when I joined. I wasn’t installing a new tool for teams, I was asking teams already under real delivery pressure, most of whom had no obligation to listen to me at all, to pause just long enough to align on a customer experience bar we hadn’t agreed on yet, without stalling a schedule five other stakeholders were counting on.

The intake held up under real volume. Across the channels already live, I ran the triage that adjudicated 231 open observations against that same shared evidence standard, closing 132 as passing retests. That’s not a proof-of-concept number. It’s what the process produced once five teams actually started running through it.

What the Intake Actually Proved

The Distinction I Look For Now

Neither the metric nor the intake was the hard part. Neither one meant anything until five teams who didn’t report to me agreed to run through it on their own. When I saw a channel underperforming its ratio, my reflex was to redesign the content myself and hand back a fix. What actually moved the ratio was slower work. I had to build enough shared evidence that the teams making channel and content decisions started reaching for the same standard without me in the room.

That’s the distinction I check for now before I propose a process. Am I building something I’ll have to keep enforcing, or something a team will keep using once I’m gone. That question matters most with the teams who were never going to answer to me in the first place.