A career ladder is a product. Ship it like one.

If your design team doesn’t have a career progression framework, your senior designers are improvising it for each report.

I shipped a framework at Panorama in late 2023. Not because anyone asked for one. Because I watched three people leave in eighteen months and read the exit-survey themes: “didn’t know if I was on track,” “didn’t know what good looked like,” “didn’t feel like growing here meant anything specific.” That’s a system of design debt, and I was the operator.

The model I built has three properties:

• Clear levels. Six rungs, named, with concrete expectations at each. No surprises. No “I’ll know it when I see it.” Senior ICs at level 4 should be able to read their ladder and predict whether they’ll be promoted on a normal cadence.
• Equitable criteria. The criteria at each level aren’t biased toward the work the leadership team happens to value. The thing the leadership team values will be obvious from twelve months of their calendar; if the criteria comment-line up with that, you’ve accidentally rebuilt the old system.
• Growth dimensions. A career ladder without growth dimensions tells people what to do. Growth dimensions tell them how to grow — and how to be evaluated when the work isn’t a clean fit for the framework. Ours had four: craft, leadership, strategy, and influence. Designers grow along different vectors at different times.

The thing that surprised me: shipping it took less time than I expected, and adoption was almost immediate. The reason is that the team had been waiting for it. People don’t leave because the work is hard. They leave because the work is opaque.

The thing that didn’t surprise me: a clear framework doesn’t remove hard conversations. It changes who has them, and when. “I’d like to be promoted” becomes “Here’s the level I think I’m working at; here’s the evidence; here’s the gap.” It puts the conversation on the right side of the year.

Ship the ladder like a product. Research it (interview your team about what good looks like). Prototype it (v0 with one level fully fleshed out). Test it (put it in front of two people you trust most, then two people you trust least). Iterate. The version you ship is always wrong; ship a wrong version is better than ship no version.