06

The Quarterly 90-Day Growth Contract

Senior-to-Principal Growth OS

Overview

Why use a 90-day contract

A growth contract is a written agreement about one capability, one real opportunity, the support each person owes, and the evidence they will inspect. “Contract” means mutual clarity rather than a legal promise, and 90 days is a bounded learning window—not a promotion deadline.

Quarterly planning prevents career development from becoming an annual retrospective. It is long enough to observe patterns across meaningful work and short enough to revise a poor hypothesis before it becomes a year-long assignment.

The seven fields

A usable contract fits on one page and names both engineer and manager obligations. The following structure is a Growth OS synthesis grounded in the opportunity-centered career guidance from GitLab and the impact language in the cited progression frameworks.

FieldQuestionExample
OutcomeWhat real result matters?Reduce release recovery time across four services
CapabilityWhat behavior will be tested?Create cross-team direction and ownership
OpportunityWhich real assignment provides the test?Reliability operating-model redesign
EvidenceWhat observations would change our view?Adoption, outcome trend, decision feedback, owner independence
AuthorityWhat may the engineer decide or convene?Own recommendation and cross-team working group
SupportWhat will manager and others provide?Sponsor forum, monthly observation, Staff peer coach
BoundariesWhat limits risk and workload?Two pilot services; no platform rewrite; stop conditions named

Add the start date, day-30 check, day-60 check, and day-90 review. Dates create accountability; they do not create entitlement to a promotion decision.

Write an outcome, not an activity list

An outcome describes a meaningful changed state, while an activity describes motion. Growth requires action, but activity alone cannot show whether technical leadership helped.

Activity-shapedOutcome-shaped
Write an architecture strategyThree domain teams adopt shared compatibility decisions and ownership
Lead weekly platform meetingsResolve the two dependencies blocking the release path
Mentor two engineersTwo engineers independently lead later design decisions using the taught method
Present to directorsDirectors make a resourcing decision with explicit technical tradeoffs

Use a sentence with a beneficiary and a change: “For [people/system], change [baseline] toward [desired state] by [mechanism], while protecting [constraint].”

Define evidence before the work

Predefining evidence reduces hindsight bias, the tendency to reinterpret expectations after seeing results. Combine outcome evidence, behavior evidence, and corroboration instead of relying on visibility or a single metric.

For a cross-team reliability assignment:

  • Outcome: recovery time and repeated incident rate.
  • Behavior: decision framing, dissent handling, delegation, and ownership transfer.
  • Corroboration: service owners can explain the decision and act independently.
  • Context: roadmap changes or staffing losses that altered the original plan.

Evidence can disconfirm the plan. A good review may conclude that the problem was not worth solving, the assignment lacked authority, or another capability matters more.

Make manager obligations explicit

Development fails when the engineer receives a stretch goal but no access, authority, or cover. Put manager actions in the contract so opportunity is treated as part of the system rather than a reward for already-visible people.

Manager obligations may include:

  • Explain why the outcome matters and which constraints are real.
  • Introduce stakeholders and state the engineer's mandate.
  • Remove conflicting work so the assignment fits sustainable capacity.
  • Observe at least one relevant meeting or artifact each month.
  • Give prompt feedback grounded in behavior and effect.
  • Sponsor the engineer for appropriate visibility and decisions.
  • Escalate structural blockers the engineer cannot resolve.

Engineer obligations may include:

  • Keep the outcome, risks, decisions, and evidence legible.
  • Seek dissent and relevant stakeholder constraints early.
  • Ask for support before hidden failure becomes a rescue.
  • Delegate meaningful ownership and credit contributors.
  • Reflect on behavior, not only project completion.
  • Say when the assignment no longer tests the agreed capability.

Use the fixed cadence without adding bureaucracy

The contract lives inside normal work and the fixed growth cadence. The standard 1:1 remains exactly 30 minutes with 10/10/10; its final segment checks one live question rather than reviewing the entire contract line by line.

MomentPurposePrompt
Weekly final segmentSmall adjustment“What did this week teach us about the capability or support?”
Day 30Check access and fit“Is this a fair test with real work and authority?”
Day 60Check evidence pattern“What is becoming reliable, and what explanation changed?”
Day 90Decide next cycle“Continue, deepen, redirect, or stop—and why?”

Keep routine project status asynchronous. Link the project evidence; use live time for interpretation, feedback, and decisions.

Worked example: platform adoption

A complete example shows how ambition and safety coexist. Jordan, a Senior engineer, wants to test Staff-scope direction across three developer teams.

FieldContract
OutcomeThree teams adopt one supported service bootstrap path; median setup time falls from two days to four hours
CapabilityFrame shared direction, negotiate constraints, and create distributed owners
OpportunityLead discovery, decision, and two-team pilot; third team reviews transferability
EvidenceSetup-time data, adopted decision, dissent log, owner independence, stakeholder observations
AuthorityConvene teams and own recommendation; directors retain staffing decisions
SupportManager introductions and scope protection; Staff peer reviews decision design
BoundariesNo mandatory migration; two-team pilot; pause if reliability declines

At day 60, one team rejects the proposed tool because of a regulated environment. That is not automatic failure. Jordan updates the direction to a shared interface with two compliant implementations, demonstrating better judgment than forcing uniformity.

Handle change and failure honestly

A growth contract is a learning device, so changed conditions require revision rather than retroactive blame. Separate project outcome, observed capability, and quality of the original opportunity.

SituationFair response
Business cancels the projectPreserve behavior evidence; find another test if capability still matters
Manager never secures authorityRecord opportunity failure; do not call it an engineer skill gap
Engineer avoids key stakeholdersGive specific feedback and redesign the next practice step
Outcome fails despite strong judgmentInspect decision quality with information available at the time
Outcome succeeds through rescue and burnoutDo not treat unsustainable heroics as the desired capability

Promotion may be discussed separately, but the contract must never say “complete this and you will be promoted.” It creates evidence and capability, not a guaranteed organizational decision.

Contract template

This one-page template is the working artifact for the quarter. Keep language observable and revise it when assumptions change.

# 90-Day Growth Contract

Engineer:
Manager:
Start / day 30 / day 60 / day 90:

## Growth hypothesis
I want to test [capability] because [evidence-based diagnosis].

## Real outcome and opportunity
For [beneficiary], change [baseline] toward [desired state] by [mechanism],
while protecting [constraint]. The bounded assignment is [scope].

## Evidence
- Outcome:
- Observed behavior:
- Corroboration:
- Context to record:

## Mutual obligations
- Engineer will:
- Manager will:
- Other support:

## Authority and boundaries
- Decisions / access:
- Capacity removed:
- Risk limit / stop condition:

## Review decision
Continue / deepen / redirect / stop, because:

Key takeaways

The contract turns growth intent into mutual, testable work while protecting uncertainty. It complements local career processes rather than replacing them.

  • Use one capability and one consequential opportunity per 90-day cycle.
  • Define outcomes, behavior evidence, corroboration, and context before work begins.
  • Write manager access, authority, sponsorship, and capacity obligations beside engineer actions.
  • Review at days 30, 60, and 90 while using the weekly final 1:1 segment for small adjustments.
  • Distinguish project failure, capability evidence, and opportunity quality.
  • Never promise promotion in exchange for completing a growth contract.

Practice: draft and pressure-test a contract

This exercise turns the diagnosis into a fair experiment. Draft independently, reconcile together, then ask a neutral senior peer to test the logic.

  • [ ] Outcome matters to actual customers, business, or engineering.
  • [ ] Capability is an observable behavior, not a title adjective.
  • [ ] Opportunity is real and bounded.
  • [ ] Evidence includes result, behavior, corroboration, and context.
  • [ ] Engineer has explicit authority and sustainable capacity.
  • [ ] Manager actions have owners and dates.
  • [ ] Stop conditions contain risk.
  • [ ] Contract does not imply a guaranteed promotion.

Pressure-test question: “If the outcome changed for reasons outside the engineer's control, could we still make a fair statement about capability and opportunity quality?”

Sources

These sources ground the opportunity and impact principles; the seven-field contract, review rhythm, and examples are course syntheses.