Mapping Your Own Growth: A Competency Self-Assessment That Isn't Wishful Thinking

A product manager competency self-assessment only works if every rating requires a specific piece of evidence — a decision, an artifact, a journal entry — behind it. Rate discovery, delivery, strategy, influence, and craft on a 1-5 scale, but reject any rating you can't back with a dated example from the last quarter.

Quick answer: Rate yourself across five competencies — discovery, delivery, strategy, influence, craft — on a 1-5 scale, but require a dated example for anything above a 2. No evidence, no rating. Pick the single weakest competency each quarter and design repeatable reps around it, then re-rate 90 days later.

Why Most Competency Self-Assessments Are Wishful Thinking

Most PM skills matrices fail because they ask for a gut-feel number with no evidentiary bar — and gut feel is exactly what's unreliable here. The Dunning-Kruger effect shows weaker performers consistently overrate themselves, while stronger performers often underrate themselves, so an unanchored 1-5 scale mostly measures confidence, not competence.

In Kruger and Dunning's original 1999 Cornell study, the weakest-performing quartile placed their own ability far closer to the middle or top of the pack than their actual scores supported — a gap of several dozen percentile points on some measures. The people least equipped to judge a skill are also the least equipped to judge their own judgment of it. A self-assessment that doesn't account for this isn't measuring growth; it's measuring how good you are at feeling confident.

The usual failure patterns show up the same way in almost every skills matrix a PM fills out alone:

• Rating "4/5 strategic thinking" with no memo, decision, or artifact attached to justify it.
• Copy-pasting a generic company competency matrix built for a different function and never adapted to PM work.
• Rating based on how a meeting felt rather than what you actually produced or decided.
• Never revisiting the rating, so it becomes a fossil instead of a working instrument.

Generic competency frameworks compound the problem. Marty Cagan and the Silicon Valley Product Group have argued for years that most company career ladders describe activities — write specs, run standups — rather than judgment, which is what actually separates a strong PM from a weak one. A framework built around activities lets you claim a high score for effort regardless of what that effort produced.

This exercise sits inside a broader discipline — the complete guide to PM craft reflection covers how self-assessment fits alongside retros, journals, and decision logs as one system, not an isolated worksheet you fill out once a year before a performance review.

The Five-Domain Framework: Discovery, Delivery, Strategy, Influence, Craft

A useful PM competency framework has five domains — discovery, delivery, strategy, influence, and craft — each answering a different question about how you create value. Discovery asks whether you find the right problems; delivery asks whether you ship them; strategy, influence, and craft round out judgment, persuasion, and technical fluency.

Public competency frameworks converge on a similar shape independently. GitLab's openly published product manager job family and the open-source progression.fyi project, which catalogs career frameworks from several tech companies, both separate "finding the right problem" from "shipping the solution" as distinct, separately-rated skills rather than one blended score.

Evidence for discovery usually comes from process artifacts you already run, not a separate exercise. If you don't have a live discovery practice generating that evidence yet, the complete guide to Jobs-to-be-Done and the complete guide to customer journey mapping are reasonable starting points before you try to rate this domain honestly at all.

Teresa Torres, in Continuous Discovery Habits, describes exactly this kind of evidence trail — weekly customer touchpoints, opportunity trees, assumption tests — as the difference between a PM who discovers continuously and one who only discovers when a launch forces it. That trail is what a 3 or 4 in the discovery row should point to.

Delivery, strategy, influence, and craft each need their own trail, not a single overall impression. A PM can be excellent at craft — clean specs, sharp data models — while still weak at influence, unable to get those specs prioritized. Collapsing five domains into one "how good a PM am I" number erases exactly the information a quarterly plan needs.

What Each Domain Looks Like in Practice

• Delivery is the domain most PMs default to over-rating, since shipped features feel like proof of competence on their own. Shipping through heroics — late nights, last-minute scope cuts — is different from shipping on a predictable cycle, and the evidence bar should reward the latter, not just the existence of a launch.
• Strategy competence shows up as reasoning that would survive being wrong. A prioritization call you can defend even after the outcome disappoints is stronger evidence than a call that only looked smart in hindsight.
• Influence is the domain PMs least accurately self-assess, since it depends on someone else's account of what changed their mind. A stakeholder alignment map, or a documented before/after position on a contested decision, is closer to real evidence than a private sense that "the meeting went well."
• Craft is the easiest domain to over-invest in relative to its leverage — a beautifully specified feature that never gets prioritized doesn't help anyone. Rate craft on whether the work holds up under someone else's scrutiny, not on how much of it you produced.

The Evidence-Based Rating Exercise That Fights Self-Deception

The fix for wishful self-rating is a five-point scale where each level has a written evidence bar, not just an adjective. A 3 requires a named artifact from the last 90 days; anything you can't cite drops to a 2 by default, regardless of how confident it feels.

If you can't fill in the evidence-bar column for a domain, the rating isn't a 3 — it's a 2 wearing a 3's number.

Running the Exercise Step by Step

1. List the five domains down a page — discovery, delivery, strategy, influence, craft.
2. Write the single most recent dated example of you exercising each one, before assigning any number. A past action, not a plan.
3. Assign the score only after the example exists. If nothing comes to mind within a minute or two, that's your answer.
4. Have one other person sanity-check at least any 4 or 5 — a manager, peer PM, or mentor. Self-serving inflation clusters at the top of the scale.
5. Store it somewhere you'll actually reopen. A rating that vanishes into a drawer doesn't fight Dunning-Kruger — it just delays the reckoning by a year.

Recency bias works against this exercise the same way hindsight bias does. The artifact from three days ago feels far more available than one from ten weeks earlier, even when the older one is the stronger piece of evidence. Scan the full quarter, not just the last sprint, before writing down the example.

Strategy ratings are the easiest to inflate in hindsight. A bet that happened to work can look prescient in retrospect even when the reasoning behind it was shaky at the time it was made. Cross-check any strategy score against your own decision log to counter hindsight bias before crediting yourself with judgment you didn't actually apply in the moment.

From Rating to Reps: Choosing One Competency Per Quarter

Once you have five honest ratings, resist the urge to fix all of them at once. Pick the single lowest-evidence competency, name one specific behavior inside it, and design a repeatable weekly rep around that behavior — the way a musician drills one passage, not the whole piece.

K. Anders Ericsson's research on expert performance, which introduced the term deliberate practice, found the differentiator among top performers wasn't raw hours logged. It was focused repetition on a specific weakness, with fast feedback attached. Sitting in more meetings doesn't build influence any more than playing more scales builds virtuosity without a teacher correcting the technique.

Designing a One-Quarter Rep Plan

1. Name the exact sub-skill, not the vague version. Not "be more strategic" — "write a one-page strategy memo before every roadmap review."
2. Pick a weekly cadence, not a one-time event. A rep that happens once isn't a rep.
3. Build in a feedback loop. Someone else reads the artifact and reacts, in writing or out loud.
4. Set the re-rating date now, 90 days out, on the calendar — not "sometime next quarter."

This is the same logic behind deliberate practice for product managers: a rep only counts if it's specific, effortful, and reviewed by someone other than you — not just repeated on your own.

Closing the Loop: Let Your Journal Prove or Disprove the Rating

Self-ratings drift from reality unless something outside your own head keeps score. A running journal of frictions, decisions, and small wins is the cheapest audit trail — when re-rating time comes, you check the number against entries written in the moment, not against memory, which is exactly what memory is bad at.

Building a friction journal habit gives you exactly this kind of raw material. The moments where something didn't work are often the clearest evidence of where a competency is still a 2, not the polished retro slide you present afterward.

At quarterly review time, tag each journal entry from the period against the competency it evidences, then look for the mismatches:

• Zero entries tagged to a domain in 90 days is data too — you're not exercising it, whatever number you wrote down.
• Entries that contradict a rating — you rated influence a 4 but every entry describes being overruled — are the most valuable signal. Trust the entries over the number.
• A repeating failure pattern across entries is your next quarter's rep design, already written for you.

This is also where hindsight bias does the most damage without a written trail: a quarter that ended well tends to get remembered as smoother and more strategic than the entries from the middle of it actually show.

Where Prodinja Fits: Growth Competencies as a Structured Mirror

None of this replaces the discipline of actually writing the example down and naming the weekly rep — a structured skill map only sharpens a habit you're already building, it doesn't manufacture the habit for you.

Key Takeaways

• Evidence beats confidence. A rating without a dated artifact is a guess wearing a number.
• Five domains, one question each. Discovery, delivery, strategy, influence, and craft each test a different way you create value — don't collapse them into a single vague "PM skills" score.
• The evidence bar is the anti-Dunning-Kruger mechanic. Kruger and Dunning's own research found weak performers overestimate themselves the most; an unanchored scale rewards exactly that blind spot.
• Pick one competency per quarter, not five. Deliberate practice research points to focused, reviewed reps beating broad, unreviewed effort every time.
• Your journal is the ground truth, not your memory. Re-rate against dated entries, not against how the quarter felt in hindsight.
• A missing entry is itself a data point. No journal evidence tagged to a domain in 90 days tells you it isn't being exercised, whatever the self-rating says.
• Structured tools can hold the map for you. Prodinja's Growth competencies inside the Leadership Suite give you a place to rate against and reflect from, instead of rebuilding the exercise from scratch every quarter.