Why Borderline Cases Lose in Promotion Calibration (And How to Not Be One)

Your manager walked out of calibration and said something like: "It was close. The room went back and forth. Ultimately they decided to hold off until next cycle."

That's the borderline experience. You weren't a clear no. Nobody questioned your competence. But you weren't a clear yes either. And in calibration, "close" almost always resolves in one direction.

Down.

The three buckets every calibration room sorts people into

When managers sit down for calibration, the discussion doesn't treat every candidate equally. Within the first few minutes, the room implicitly sorts promotion cases into three categories:

Clear promotes. The manager presents the case. Another manager in the room nods — they've worked with you, or they've seen your name on a cross-team initiative. One clarifying question gets a specific answer. The room moves on in under four minutes. These cases are essentially pre-approved. The calibration meeting just ratifies what the room already sees.

Clear no-promotes. Either the manager doesn't nominate you, or they present a case that quickly reveals the evidence isn't there. Someone asks a question the manager can't answer. The room moves on even faster. Nobody argues.

Borderline cases. Your manager makes a reasonable pitch. Another manager pushes back: "Is that really next-level scope, or is that solid execution at the current level?" Your manager answers, but not crisply. A debate starts. It goes on for ten, sometimes fifteen minutes. People check the clock. The room has twenty more names to get through. And then someone says, "Let's revisit next cycle when there's more evidence."

That last sentence is the death of borderline cases. The room isn't hostile. It's tired, short on time, and built to say no when the answer isn't obvious.

Why the system is rigged against "maybe"

Calibration isn't designed to find everyone who might be ready. It's designed to confirm the people who clearly are. That asymmetry, caring more about false positives than false negatives, is baked into how every major tech company runs the process.

The organizational cost of promoting too early outweighs the cost of promoting too late

Companies have learned what happens when someone gets promoted too early. The newly promoted engineer struggles to deliver at the new level. Their first review is mediocre. Peers who were passed over resent the decision. The manager who championed the case loses credibility for future pitches. In extreme cases, the person gets managed back down, which is painful for everyone.

The cost of making someone wait an extra cycle? They're frustrated, maybe they update LinkedIn. But organizationally, nothing breaks. The Pragmatic Engineer's analysis of how calibrations function at Big Tech found that committees are designed to prevent the first scenario, not the second. Borderline cases absorb the consequences of that design choice.

Budget constraints turn "maybe" into "no"

At most tech companies, the number of people who can be promoted in a given cycle isn't unlimited. It's a percentage. When the budget is tight (and post-2023, budgets at most companies have stayed tight), the clear cases fill the slots first. Borderline cases compete for whatever is left, which may be nothing.

Vince Sacks Chen, an engineering manager who wrote publicly about calibration, described how during lean budgets the exact same case that would pass in a growth year gets tabled. The evidence doesn't change. The budget does.

Time pressure favors the obvious

A typical calibration session has 30 to 50 people to discuss in a two- or three-hour block. Clear promotes take three minutes. Clear no-promotes take one. That leaves the borderline cases fighting for whatever time remains. A room full of managers who've already been arguing for two hours isn't inclined to extend the benefit of the doubt.

Harvard Business Review research found that speaking order and verbal persuasiveness directly affect calibration outcomes. A borderline case that comes up late in the session loses partly to fatigue, not just to weak evidence.

What the room actually looks like when a borderline case comes up

Here's a composite of how borderline discussions play out, based on patterns described by engineering managers who've written about calibration, including Susnata Basak (former Google promotion committee member) and managers interviewed in the Pragmatic Engineer newsletter.

Minute 1: The pitch. Your manager presents: "She led the observability overhaul for the payments service. Reduced alert noise by 50%. I think she's ready for senior."

Minute 2: The first challenge. A peer manager asks: "Who else was involved? Did she drive the technical decisions, or was she executing someone else's design?" Your manager pauses. "She was the primary engineer, but the staff engineer on the team defined the initial architecture."

Minute 3: The reframe. Another manager: "So she executed well on a project that was already scoped. That's solid L5 work. What makes it L6?" Your manager tries to recover: "The cross-team coordination was significant — she worked with three other teams." The room isn't satisfied. Coordination and driving are different words in calibration.

Minute 5-8: The spiral. More questions. Your manager answers each one, but each answer opens another question. The discussion isn't hostile. It's probing. But with each probe, the room's confidence drops. The work wasn't bad. The case just wasn't airtight, and airtight is the standard when slots are limited.

Minute 9: The deferral. A director says: "I think she's tracking well. Let's give it another cycle and look for more cross-team ownership." The room agrees. Everyone moves on.

Your manager leaves the room feeling like they fought for you. And they did. But fighting isn't the same as winning, and the structure of the meeting made winning almost impossible once the first question didn't get a crisp answer.

The five things that make a case borderline

Understanding why cases land in the borderline bucket is the first step to avoiding it. These are the patterns that calibration rooms flag, over and over.

1. The work is good, but the level claim is ambiguous

You shipped something real. The impact is there. But the room can't tell whether it was next-level work or excellent current-level work. This is the most common borderline trigger, and it happens because nobody — not you, not your manager — explicitly mapped the work to the promotion rubric before the meeting.

The room isn't going to do that mapping for you. If your manager can't say, "Here's specifically why this is L6 and not strong L5," the default assumption is that it's the lower one.

2. Only your manager can speak to your work

When your manager makes the pitch and nobody else in the room has any context on who you are, the room is being asked to trust one person's assessment. That's a weaker position than a case where a second manager says, "I worked with her on the cross-team incident response. She drove the resolution."

Research on manager advocacy published in Harvard Business Review found that managers who only start advocating during promotion cycles are far less effective than those who build visibility for their reports year-round. If your name is unknown in the room, your manager is starting from a deficit that a 90-second pitch can't overcome.

3. The evidence is concentrated in one project or one quarter

Promotion panels want sustained evidence of operating at the next level, not a single impressive quarter. If your strongest proof point shipped last month, the room can't assess whether the output is repeatable. Kevin London, an Amazon engineer who documented his promotion experience, described his first attempt being rejected because the panel wanted to see next-level work sustained across multiple review periods.

Two quarters of consistent next-level evidence is the informal standard at most companies. One quarter of exceptional work gets filed under "promising, but let's see more."

4. The manager can't survive the follow-up questions

The pitch lands fine. Then someone asks: "How does this compare to [other engineer] who did similar cross-team work?" or "What was the scope before she took it over?" If your manager hesitates, the case deflates in real time.

This usually isn't your manager's fault. They're managing six to ten people and preparing cases for all of them. The engineers who hand their manager a written case document, with quantified outcomes, rubric mapping, and pre-answered objections, are the ones whose cases survive the follow-up. The engineers who assume their manager tracked everything are the ones who end up borderline.

5. The self-review is activity, not evidence

Your manager's calibration pitch is built from what you gave them. If your self-review said "worked on the migration project" instead of "led the migration of the billing pipeline, reducing failed transactions by 35% and eliminating a manual reconciliation step that cost the finance team four hours per week," your manager has to fill in the gaps. And the gaps get filled with adjectives — "strong contributor," "very reliable" — that carry zero weight in a calibration room.

How to move from borderline to clear

The distance between borderline and clear isn't about doing more work. It's about making the work you've already done legible to a room full of strangers.

Build the second voice before calibration season

If the room only hears your manager's voice, your case is structurally weaker. The fix isn't working harder in your current silo. It's working on something that puts your name in front of another manager. A cross-team project. A technical design review where adjacent teams see your output. Leading an incident response that touches multiple services. The goal is simple: when your manager says your name in calibration, you want at least one other person in the room to nod.

Write the case document your manager won't

Two weeks before review season, hand your manager a half-page with your two strongest projects, quantified outcomes, rubric alignment, and a clear level claim. Don't wait for them to ask. Most managers aren't bad at calibration prep because they don't care. They're stretched across too many reports and too many meetings. You're making their job possible, not doing it for them.

If you haven't built a case document before, how to write a promotion case walks through the exact format.

Pre-answer the objections

Think about what another manager would push back on. If your biggest project involved a staff engineer defining the architecture, be ready to explain what you owned and decided independently. If the impact is hard to quantify, find a proxy metric or a qualitative outcome that demonstrates the before-and-after. If the scope was team-level, not org-level, explain why team-level scope at that complexity was the right level claim.

Write these answers down and give them to your manager. They can't anticipate every objection. But if they walk in with three pre-loaded responses to the most likely challenges, the pitch holds.

Make the level claim explicit

Don't leave it to the room to decide whether your work was current-level or next-level. Tell your manager: "I think this project demonstrates senior-level scope because I drove the cross-team coordination and the technical decisions without guidance." Your manager can agree, adjust, or push back — but you've anchored the conversation on a specific claim instead of leaving it open for the room's most conservative interpretation.

Start the next cycle's evidence now

If you're reading this and your next review cycle is more than two months away, you have time to build the kind of sustained evidence that panels want. The engineers who get promoted cleanly aren't the ones who sprint in the final quarter. They're the ones who have two or three quarters of documented, next-level work that their manager can speak to in 90 seconds without having to guess.

---

Calibration rooms move fast. The engineers who get promoted aren't the ones who work the hardest. They're the ones whose cases land in three minutes because the evidence was already clear. CareerClimb helps you build that evidence all year, so when your manager walks into the room, they have more than memory to work with. Download CareerClimb and start building a case that isn't borderline.