Staff Engineer Mentor. For the IC whose "next cycle" already happened three times.

You're the person the team asks when the architecture question gets hard. You wrote the load-test tool half the org runs on. Review comes round, you hear next cycle again, and when you ask what the specific gap is you get keep doing what you're doing. That isn't an answer. That's someone who hasn't done the work of finding out.

In most companies the promotion to Staff isn't decided by your manager at all. It's decided by a committee your manager has to argue in front of, in a room you'll never sit in, from evidence they assemble mostly from memory. If nobody has been collecting that evidence, your case is whatever your manager can recall on the day, which is usually two things, one of them out of date.

So we build the packet for them. Three artefacts from the last twelve months, written the way a committee reads: what changed, who outside your team felt it, what it was worth. Then we rehearse handing it over so it lands as help rather than as a complaint, because that's the part that goes wrong. Then a standing quarterly check, so the cycle after this one isn't decided from memory either.

Sixty days: a named timeline in writing, and a specific list of what ready means at your company as opposed to what it means in a book.

The title's on your LinkedIn. The job description was three bullets somebody pasted out of a Reddit thread. Half the time you're asked to be a very senior IC, the other half you're asked to run a project because you're Staff, right. Last quarter you said yes to too many things, mostly because you had no basis for saying no.

The archetype posts don't fix this, and I say that as someone who wrote one. Knowing you're a Solver rather than a Right Hand tells you nothing about which Thursday meeting to decline.

What fixes it is a written scope your manager has signed. Which archetype you're running here, at this company, this year. The two or three weekly rituals that make it visible, because scope nobody can see quietly gets reassigned. Which meetings you're in and which you're not. And a decline template, so that saying no costs you thirty seconds instead of a week of guilt. By week 6 the document exists and your manager has agreed to it, and the agreement is worth considerably more than the document.

You ran a small team ad hoc for a while and liked roughly half of it. Now your director keeps putting the EM role in front of you. The Staff+ ladder officially exists, and you can't name one person above you who's on it.

The fear underneath is usually not about management at all. It's that declining marks you as unambitious, and that the offer won't come round again.

Sometimes that fear is correct and I'll say so when I think it is. A company with nobody visible above you on the IC ladder is telling you something true about its ceiling, and no amount of career planning rewrites an org chart. Around half the Staff-track people I work with end up changing companies. That isn't a failure, it's the honest read on a path that only exists on a slide.

What we do is get the decision off the vibe and onto paper. Both routes costed out: title, money, and what an ordinary Wednesday looks like at 24 months, here and at a company that genuinely staffs the IC ladder. Then a real trial, four weeks of running the team properly with me in the background, because that teaches you more than any amount of introspection will. And a way to decline that keeps the door open, if declining is where you land.

No target date on this one. Some people know inside two sessions. Some take a quarter, and taking a quarter is not the same thing as avoiding it.

Nobody promoted you. Nobody assigned you the team. But the architecture question comes to you, the RFC routes through you, and when it breaks at 2am you're the one they wake. Then the comp review came round and handed you the same 4% as everybody else.

Telling your manager you'd like it made official rarely moves anything, and it helps to understand why. From where they sit, the problem is already solved. The work is getting done, you're doing it, and making it official costs budget and a headcount conversation to fix a problem only you have.

So we stop asking and start documenting. Your version of the tech lead role, your manager's version, and your team's version, each written separately and then reconciled, which is where it gets interesting because those three rarely match. The rituals you own, by name. On-call adjusted to the responsibility. Then the comp conversation built on the gap between what you carry and what you're paid, rather than on how patient you've been.

By week 8: scope in writing, either formal recognition or a dated plan to get there, and a comp conversation in the calendar rather than in your head.

One warning. Sometimes this surfaces that the org has no intention of formalising anything. You'll find that out in week 3 as opposed to year 3, which feels worse and is a much better position to be in.

You shipped the payments feature in a fortnight. Copilot and Claude did most of the typing. Your review turnaround is the bottleneck now rather than your throughput, your peers are quietly split on whether any of that counts as your work, and your manager has no way to assess what you did this quarter beyond a general feeling that it was a lot.

This part is genuinely new and I'd rather not pretend otherwise. Nobody has ten years of data on it. What I do have is a lot of engineering leaders in one room arguing about it, and a rough consensus forming: the reviewable unit stopped being the line and became the decision.

Which is workable, because decisions are observable. What did you choose to trust. What did you catch. What did you escalate. What did you refuse to ship. None of those is a volume metric, and all of them are things a committee can be shown.

So we write the standard for your team. Which review rituals change, what a promotion committee should look at for an AI-heavy IC, and one number your manager can defend upward without it collapsing under a follow-up question. You send it as an RFC, because a standard you proposed reads very differently from one you were subjected to.

By week 6 the standard exists with your name on it, which is incidentally a Staff-scope artefact in its own right.

Every incident, every architecture call, every hard review. You're the escalation path. Nobody logs it, everybody relies on it, and the last cycle handed you the same 4% as the rest. You already know that raising it gets you told you're on the high-potential list, followed by another initiative.

The reason this never moves is that your work is invisible in the one place it would have to be visible. Glue work leaves no artefact behind. Nobody signs off a raise for a general sense that somebody is load-bearing.

So the first move is unglamorous and it works: three or four weeks of you logging it. Which incidents, which reviews, which unblocks, how many hours. It's tedious, and most people are quietly shocked by their own number. Then we turn that log into the single page your manager needs in order to walk into a comp round and argue for you, because they need ammunition roughly as much as you need the money.

Then we rehearse the ask, including the version where the answer is no and what you do on the Monday after a no. By week 6 there's a quantified artefact, a comp conversation booked, and a fallback you thought through while calm as opposed to deciding it angry.