Build a 30/60/90 learning roadmap for a new domain
intermediateClaude SonnetPersonal ProductivityLearninglearningskill-developmentplanninggrowthroadmap
Use case
Use this when you're starting from scratch on something (new technical area, a domain you've never touched, a discipline you need to pick up). Beats the trap of reading 30 articles and feeling no closer to competent.
The prompt
You are designing a learning roadmap for me. Be concrete and ruthless about scope. Most learning roadmaps fail because they're too broad and have no project component. What I want to learn:{{learning_target}}Why I'm learning it (specific outcome):{{why_learning}}My current level (be honest):{{current_level}}Time I can realistically commit per week:{{weekly_hours}}My existing skills that transfer:{{transferable_skills}}How I learn best (reading, video, building, teaching, etc.):{{learning_style}}Design a 30/60/90 day roadmap structured as: DAYS 1-30: FOUNDATION - Goal — what I should be able to do by day 30 (concrete, demonstrable) - Core resources — 2-3 max, named specifically (book title, course, paper). No "explore the internet." - One small project — something I'll have built or done, not just read - Self-check — how I'll know I actually learned this and can move on DAYS 31-60: APPLICATION - Goal - Resources (2-3 max) - One mid-sized project - Self-check - One person to talk to (real practitioner, even if cold-outreach) DAYS 61-90: INTEGRATION - Goal - Resources - One real-world project tied to{{why_learning}}- Self-check — what does "competent" look like - How I'd teach this to someone else (the Feynman test) Then output: - WHAT TO SKIP — common rabbit holes I'll be tempted by but should avoid - HARD TRUTH — one honest thing about this learning journey most people don't acknowledge - WEEK 1 ACTION — the literal first thing I should do tomorrow morning Constraint: only suggest resources that exist and you're confident about. If you're not sure, say "find a current intro [topic] book" rather than inventing a title.
Variables
{{{{learning_target}}}}Replace with your {{learning target}}{{{{why_learning}}}}Replace with your {{why learning}}{{{{current_level}}}}Replace with your {{current level}}{{{{weekly_hours}}}}Replace with your {{weekly hours}}{{{{transferable_skills}}}}Replace with your {{transferable skills}}{{{{learning_style}}}}Replace with your {{learning style}}
Example
Input
learning_target: SQL — enough to do my own data pulls without bothering the data team why_learning: I'm a PM and I lose 1-2 days per week waiting on simple data pulls; I want to self-serve for funnel and cohort questions current_level: I can read SQL when it's in front of me; can't write a JOIN from scratch weekly_hours: 4 hours transferable_skills: Strong with spreadsheets, comfortable with logic and data structures, used to dashboards learning_style: I learn best by building real things on real data, not toy examples
Output
DAYS 1-30: FOUNDATION
Goal: Write a query from scratch with two JOINs and a GROUP BY against your real production data warehouse, get the right answer.
Core resources:
- "SQL for Data Analysis" by Cathy Tanimura (book, ~250 pages, written for analysts not engineers)
- SQLBolt (free interactive tutorial; do all lessons)
- Your company's actual data warehouse — get read access in week 1
Project: Pick one funnel question you've been bothering the data team about and write the query yourself. Compare your answer to theirs.
Self-check: Can you explain JOIN types (inner/left/right) without checking? Can you write GROUP BY + aggregate from scratch?
DAYS 31-60: APPLICATION
Goal: Build 3 real reports the data team currently maintains for you, on your own. Match their numbers.
Core resources:
- "Mode Analytics SQL Tutorial" (free, intermediate sections)
- Your company's existing query library — read 20 queries other people wrote
- Window functions chapter of any modern SQL book — this is the unlock for cohort analysis
Project: Build a self-serve cohort retention chart for your product area. Validate against the data team's number.
Self-check: Can you write a window function for "user's nth event"? Can you debug a query that returns the wrong row count?
Person to talk to: Buy a data analyst at your company a coffee. Ask them what bad SQL they see PMs write and why.
DAYS 61-90: INTEGRATION
Goal: SQL is no longer the bottleneck. You self-serve every funnel and cohort question and only escalate genuine modeling questions.
Resources:
- Advanced topics: CTEs, recursive queries (skim only what you need)
- Read your data warehouse's docs on dialect-specific functions (Snowflake/BigQuery/Postgres differ)
Project: For your current PM project ({{why_learning}} mentions onboarding — build the activation funnel dashboard yourself in your BI tool, owned by you).
Self-check: A peer PM asks you a data question — you don't punt to the analyst. You answer.
Teach: Explain to another PM what a window function does and when to use one. If you can't, you don't really know it.
WHAT TO SKIP
- Stored procedures, triggers, indexing internals — engineering territory, not yours
- Full database administration — not your job
- SQL performance optimization beyond "don't SELECT * on huge tables" — let the analyst review when needed
HARD TRUTH
Most PMs who "learn SQL" stop at JOIN + GROUP BY and never reach window functions. That's where the actual leverage is. The first 30 days will feel productive and addictive; the 31-60 stretch is where most people quit. Push through.
WEEK 1 ACTION
Tomorrow morning: request read-only access to your data warehouse and create one bookmark in SQLBolt. That's it. Don't read three books — open the tool.
Tips for best results
- 1Skip resources you can't find or remember the name of. The model will sometimes invent course titles — verify before committing.
- 2Pick a project that ties to a real work problem (your why_learning input). Toy projects don't survive your busy weeks.
- 3Build the 'person to talk to' step into the plan. Conversations compress weeks of learning into hours.
- 4If your roadmap doesn't have a teach-it-back milestone, add one. The Feynman test is the cheapest validation of real understanding.
- 5Re-run this prompt at day 30 with what you actually did. Roadmaps need recalibration; the original is a guess.
Related prompts
Test understanding with the Feynman technique
intermediateForce yourself to explain a concept simply, then have Claude poke holes in your explanation to find what you don't actually understand.
Personal Productivitylearningfeynmanunderstanding
Distill a book into 3 specific actions
intermediateTake a book you just finished and distill it into exactly 3 actions you'll take in the next 30 days — not a summary, not a takeaway list.
Personal Productivitylearningreadinghabits
Prep targeted questions for a mentor or skip-level
intermediatePrepare 3-5 sharp, specific questions for a mentor, skip-level, or coffee chat — questions only this person could uniquely help you with.
Personal Productivitylearningmentorshipcareer
Need help implementing this prompt in your workflow?
Book a call