Verified against ChatGPT · 2026-08-14
Draft a self-review that reads as confident evidence, not a humble-brag or a list of tasks
Turns a rough log of what you did this review period into a structured self-review draft organized by impact, calibrated to sound confident without sounding inflated, and flagging any section still too thin to submit as-is.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are drafting my self-review for this performance cycle from the raw notes below. The goal is a draft that reads as confident, evidenced self-assessment — not a humble list of completed tasks, and not inflated language that oversells thin results.
REVIEW PERIOD AND FORMAT
H2 2026 review, company uses a 4-category rating rubric with a 300-word limit per category.
RAW NOTES ON WHAT I DID
Shipped the new onboarding flow, cut signup drop-off by 12%; ran three A/B tests, two shipped, one killed for no lift; covered for a teammate on parental leave for 6 weeks.
STATED COMPANY VALUES OR REVIEW CATEGORIES
Impact, Collaboration, Technical Craft, Ownership.
AREAS I STRUGGLED WITH
Missed the deadline on the onboarding flow by two weeks because I underestimated the QA cycle; didn't flag the risk early enough.
For each review category or company value listed, pull the strongest matching example from my raw notes and write it as: what the situation was, what I specifically did, and what changed as a result — using a real number or concrete detail wherever one exists in my notes. If a category has only a weak or generic matching example, say so rather than forcing a thin task into a strong-sounding sentence. For the section on areas I struggled with, write it as a specific, honest account of what happened and what I'm doing differently now, not a rehearsed line like "I sometimes take on too much because I care about quality" — reviewers see that exact sentence constantly and it reads as evasive rather than reflective. Write in first person, confident and factual, avoiding both self-deprecating hedges ("I think I maybe helped a bit with...") and inflated corporate language ("spearheaded," "synergized") that isn't backed by a specific detail in my notes.
OUTPUT FORMAT
1. One short opening paragraph summarizing the period at a high level.
2. One section per review category, each with situation/action/result structure.
3. One honest, specific section on an area of growth.
4. A flagged list of any category where the underlying evidence was thin, so I know where to gather more detail before submitting.Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Requiring a situation/action/result structure per category rather than a free-flowing paragraph forces specificity precisely where self-reviews tend to go generic under time pressure — most people writing their own review default to restating job duties ("I worked on the onboarding flow") because that's what comes to mind fastest, while a structured prompt for what changed as a result pulls out the actual outcome, which is what a manager or calibration committee is scoring against. The instruction to flag thin categories rather than force a fit matters because self-reviews get read alongside a manager's independent assessment, and an inflated claim that doesn't match the manager's own memory of the period damages credibility on every other claim in the document, not just the weak one — an honest gap is a smaller cost than a credibility hit. Explicitly banning the rehearsed "I care too much about quality" line for the growth-areas section addresses a known pattern reviewers are fatigued by: that exact sentence is common enough that it now reads as a deflection rather than genuine reflection, and naming a real, specific miss (a missed deadline, a risk not flagged early) reads as more mature, not more damaging, precisely because it's concrete enough to have actually been learned from. The ban on both self-deprecating hedges and inflated corporate language matters because both failure modes come from the same underlying problem — writing about your own work without an external structure to anchor to — and GPT-5.1 left unconstrained will often mirror whichever register the raw notes were written in, so both directions need an explicit counter-instruction rather than assuming the model will land in the middle on its own.
What you get back
Impact: I led the redesign of the onboarding flow, identifying the signup-drop-off point through funnel analysis and shipping a revised flow that cut drop-off by 12%. I also ran three A/B tests this period; two shipped changes that held up post-launch, and one was killed early once data showed no lift, which saved the team from building out a feature that wouldn't have paid off. Growth area: I underestimated the QA cycle on the onboarding redesign and missed the original deadline by two weeks. The real issue was that I didn't flag the risk once I saw QA finding more edge cases than expected — I'm now building a mid-cycle checkpoint into my own project planning specifically to surface that kind of risk earlier.
Verified against
ChatGPT GPT-5.1 · 2026-08-14
Changelog
- 2026-08-14 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
