You Don't Get Promoted in the Review. You Get Promoted Two Quarters Earlier.

The promotion review is a ratification ceremony, not a courtroom. By the time you're in the room making your case, the verdict was written weeks earlier, in a calibration meeting you weren't invited to, on evidence you accumulated over the prior two quarters. If you want to know how to get promoted, stop preparing for the review. Start managing the cycle that ends before it begins.

That's the part nobody tells you. You think the review is where you argue your case. It's where someone else reads out a case that was already settled. The people who get promoted aren't better arguers. They started building the packet a quarter before anyone asked them to.

Where does the promotion decision actually get made?

Not in your one-on-one. Not in your self-review. It gets made in a calibration meeting, and you're not in it.

At Google, calibration committees evaluate candidates at the level-group level: all the L5s going for L6 are compared against each other at once. The committee doesn't read your raw peer feedback. They read a packet your manager synthesized: a summary of your impact, framed in the language of the next level. And the promotion process is deliberately disconnected from your performance rating. They're not deciding if you did well this cycle. They're deciding if you've been operating at the next level long enough to be undeniable.

So the thing being judged is a document. Not you, not your work directly. A manager-written summary of your work. Which means the real question isn't "was I good this year." It's "did my manager have the material to write a packet that survives a room full of skeptics."

How early do I actually need to start?

Earlier than feels reasonable. At Google, strong candidates have typically been operating above the bar for at least two review cycles before the promotion conversation starts, and the climb from L5 to L6 tends to run 24 to 48 months. The packet is a retrospective, not a proposal. It documents work you already did at the higher level.

Here's the timing trap. Plenty of large companies run half-year performance cycles, with calibration landing a month or two after the feedback window closes. So an engineer who starts their promo push at self-review time is often already too late. The manager is consolidating packets from the previous half's work. Your sprint of visible effort in the final weeks doesn't make the packet. It makes the next packet, if you're still here.

Whenever you ask "when should I start," the honest answer is "last quarter." The second-best answer is today. The promotion clock runs ahead of the calendar you can see.

What's the difference between a weak and strong promotion play?

The gap is almost never talent. It's whether you treated the next level as something to demonstrate or something to be granted.

Weak: Nadia ships solid features all year. She assumes the work speaks for itself and raises promotion at her annual review. Her manager says, "let's revisit next cycle." The calibration meeting already happened two weeks earlier. Her output was real. The packet was thin, because nobody was building it.

Strong: Marcus tells his manager in January he's targeting the H2 cycle. He asks what the committee will need to see, then keeps a running log of cross-team impact for six months. When calibration comes, his manager isn't reconstructing from memory. He's quoting from a thesis with evidence attached.

Same year, same effort, different infrastructure. This is consistency over intensity in its purest form. Nadia ran one heroic sprint at review time. Marcus did the boring repeatable thing of writing down what he shipped, every week, for two quarters. The boring thing won, the same way the quiet defaults you set early compound for years.

Why doesn't doing great work just get noticed?

Because the people deciding can't see your work. They see a summary, and the summary depends on a human memory that has a hard limit.

Julia Evans put it plainly in her piece on brag documents: managers cannot retain comprehensive details of everything you accomplished, and they need that detail to explain to other people why you should be promoted. Now stack the structural reality on top. The committee reads a manager-synthesized packet and never sees your raw work or peer feedback. So the chain is: your work, then your manager's memory of your work, then a written summary of that memory, then a committee's read of the summary. Three lossy steps between what you did and what gets judged.

Two failure modes follow directly. Your manager forgets the migration you led in Q1. Or your manager changed in Q3 and never saw it at all. Either way the packet comes in thinner than your output warrants, and a thinner packet loses to an equal one with better documentation.

Your job is to make forgetting impossible. A weekly running log your manager can quote from directly closes all three gaps at once.

Is a mentor enough, or do I need a sponsor?

A mentor talks to you. A sponsor talks about you, in the room, when you're not there. Only one of those gets you promoted, which is exactly why a sponsor beats a mentor for advancement.

The data is stark. Male managers with a sponsor are 23% more likely to advance to the next level; female managers, 19% more likely. And it pays the sponsor too: senior executives who sponsor rising talent are 53% more likely to be promoted themselves.

But here's the catch most people miss. Only 27% of self-identified sponsors actually advocate for their protégé's promotion, which is the one thing sponsorship is supposed to do. So even a "sponsor" might just be a well-wisher with a title.

Weak: Priya has three senior engineers who give her career advice and tell her she's doing great. None of them are in the calibration room. Strong: James has one VP who saw his work on a cross-org migration, believes in him, and names him in calibration as "operating at senior level." James gets promoted. Priya gets feedback to "be more visible," the kind of vague verdict you should read as information about exactly where you stand rather than an insult.

The test for whether someone is a sponsor: can they advocate for you in a room you're not in, or do they just encourage you in rooms you are in? If you can't answer, you probably have a mentor.

How do I get scope big enough to justify a promotion?

You stop waiting for it to be assigned. At the senior levels, the scope that earns a promotion is usually invented, not handed out.

Weak: a senior engineer waits to be assigned a project large enough to justify a Staff promotion. Two years pass. The big project never arrives, because nobody hands you scope at that level. Strong: a peer notices nobody owns the company's observability strategy, writes a proposal, gets three team leads to co-sign it, and is running a 40-engineer working group by the next cycle. The scope didn't exist. They created it and claimed it.

This is agency over fatalism applied to your own career ladder. The circumstance is real: companies rarely gift-wrap next-level scope. The choice is also real: you can find the unowned problem and put your name on it. The committee can't promote you for scope you don't have. It can promote you for scope you built.

One guardrail. Inventing scope works when you've got backing. Grabbing scope loudly without a sponsor can read as overreaching, and that hurts you in calibration more than staying quiet does. Build the proposal, get the co-signs, then go.

What does this look like as a system?

Here's the weak approach against the strong one, end to end.

Move	Weak player	Strong player
Timing	Raises promotion at the review	Tells manager the target cycle two quarters out
Evidence	"My work speaks for itself"	Weekly running log the manager can quote
Advocacy	Has mentors who give advice	Has a sponsor who names them in calibration
Scope	Waits to be assigned a big project	Identifies an unowned problem and claims it
Framing	Lists tasks completed	Frames impact in next-level language

None of these requires more talent. All of them require treating your promotion as something you build infrastructure for, the same way you'd build infrastructure for anything you actually wanted to ship.

What's the honest cost of all this?

Naming the trade-off, because the category usually hides it.

The system rewards documentation and visibility, which means it under-rewards quiet, essential work. If you do the on-call, the glue work, the unglamorous reliability that keeps systems running, the standard packet format works against you. Flashy project outcomes write up cleanly. "Nothing broke for six months because of me" does not. That's a real disadvantage, and pretending otherwise would be dishonest.

Sponsorship has the same problem at a deeper level. It requires proximity to power, and that proximity is not handed out evenly. Only 31% of entry-level women have a sponsor, against 45% of men. 20% of White employees have sponsors; only 5% of Black employees do. The "broken rung" shows up in the outcomes: for every 100 men promoted to manager, only 93 women are, and just 60 Black women. And four in ten entry-level women have had no promotion, stretch assignment, or leadership training in the past two years, against three in ten men. So "go find a sponsor" can quietly blame people for a structural access gap they didn't create.

Two more honest limits. The calibration-committee machinery is mostly a big-tech pattern. Early-stage startups often promote on a manager's gut and a single conversation, no committee at all, so the playbook flexes. And the two-quarter lead time assumes a stable team and manager. In a reorg or a high-attrition shop, track records and sponsor relationships reset, and you start the clock over.

Hold both. The system has real inequities. There's also almost nothing consistent documentation and a deliberate sponsor relationship can't eventually move. You manage the part you control without pretending the rest isn't there.

What should I do this quarter?

Three moves, doable this week, even if your cycle is two quarters out.

1. Write the promo doc now. Open a running log today. Every week, add one line: what you shipped, the scope, the impact, the next-level signal. By cycle time your manager has a packet to quote, not a memory to excavate.
2. Audit your sponsors against your mentors. List everyone who advises you. Cross out anyone who can't advocate for you in a room you're not in. If the page is empty, that's your real priority, ahead of any skill you were planning to sharpen.
3. Claim one visible initiative. Find a problem nobody owns that someone two levels up cares about. Write a one-page proposal. Get one or two co-signs. That's the scope your next packet is built on.

The promotion you want is decided two quarters before the review. The only question is whether you spend those two quarters building the case or hoping someone else builds it for you.

Want to pressure-test whether your "sponsor" can actually advocate for you, or build the running brag doc your manager can quote at calibration? Talk to Praxy on WhatsApp. I'll help you turn this quarter's work into next cycle's packet.