Stop guessing.
Start experimenting.

One conversation can change everything.

Every engagement starts with a single session — try the method, try working with me, and see if it's a fit before you commit to anything.

1 Bring a Challenge A real situation you're facing right now
2 Design an Experiment A concrete behavior change to try
3 Get Results See what works, then decide your next step

Ways to Keep Going

Group Coaching

$225/month

Structured weekly practice with people who get it.

  • Weekly 1-hour group calls — bring a friction point, vote on what to tackle, and work through 3–4 real challenges together
  • A cohort of ~12 people who build on each other's ideas (experiments don't just come from me)
  • Async access via community platform between sessions
  • All session recordings + documented experiments
  • Community extras: text channels, book clubs, deliberate practice groups

Best for: You've read the books, done the trainings, tried the advice. It's helped, but you want less theory and more tactics. You're ready for structured practice and the support of people who share your struggles.

Private Coaching

$895/month

Dedicated 1:1 coaching tailored to your situation.

  • Monthly 1-hour 1:1 session for dedicated, concentrated attention on your challenges
  • Every experiment tailored to your specific situation, your relationships, your goals
  • Unlimited async access — when you're stuck, preparing, or celebrating, you're not alone
  • All group coaching benefits (weekly calls, recordings, community)
  • Support between sessions — low effort to stay involved, high value when you need it

Best for: You're focused and have a specific goal you're ready to crush. You want everything in group coaching, plus the depth and speed of dedicated 1:1 attention. You're ready to get results.

Bring your whole team.

Private cohorts give your engineering and leadership teams a structured way to problem solve. Run real experiments on real work for real results.

Get in touch →

Sample Session Report

Names and identifying details have been changed to protect privacy.

Session Notes: Alex

Date: January 5, 2026 | Duration: ~59 min Attendees: Andrea Goulet, Alex

Overview

Alex spent ~15–16 hours on his blog during the holiday break and feels proud of the progress. The session explored why budgeting 20 hours (vs. 1 hour) shifted his emotional experience from embarrassment to pride, identified a sunk cost fallacy pattern around rabbit holes, and established experiments for estimation and context capture.

Experimental Design

1. Friction Point

Estimating accurately and avoiding rabbit holes that delay launching the blog.

2. Context
Theme Description
Previous session Felt embarrassed about lack of progress when estimating 1 hour for tasks
This session Felt proud after budgeting 20 hours and spending ~15–16
Work blocks Morning sessions continue to be effective; context switching is expensive
Documentation Writing tickets/documentation "doesn't feel like work," contributing to sunk cost trap
Work situation Increased estimation requests at work
3. Variables
Variable Observed Outcome
Time budget (1 hr vs 20 hrs) 20 hrs → pride; 1 hr → embarrassment
Doom scrolling replacement Successfully swapped for blog work multiple times
Breaking work into pieces Easier to estimate smaller chunks
Claude Code for tedious tasks Content import, to-do list generation, redirect script creation worked well
4. Success Criteria
Criterion Status
Blog live In progress
First post published Pending
Proud of visual design Achieved locally
Able to publish new articles Pending deployment
5. Hypotheses
IF THEN
Alex quadruples his gut instinct for estimates He'll avoid disappointment and feel proud of progress
Alex uses Claude Code to capture thinking when hitting a blocker He can avoid sunk cost fallacy and context-switch more easily
Alex breaks big things into smaller pieces before estimating Estimates become more accurate
6. Measurement
Metric How to Track
Actual time vs estimated time Track for each task
Sunk cost trap occurrences Notice when falling in and capture the moment
Blog launch Binary: is it live?
7. Experiment
Behavior Change Expectation
Estimation approach: Before communicating any expectation (external or internal), break the big thing into smaller pieces, apply 4x multiplier to gut instinct, sum the pieces More accurate estimates, less disappointment
Avoiding rabbit holes: When a task exceeds its budget and isn't mission-critical, pause. Capture: goal, blockers, potential solutions. Create WIP commit with to-do list, push to branch, link in ticket Escape sunk cost trap, preserve context without full documentation overhead
Ask: "How essential is this for accomplishing the actual goal?" Prioritize mission-critical work
8. Follow What Works
Past Experiment Effectiveness
Budgeting more time (20 hrs vs 1 hr) Proud vs embarrassed
Morning work sessions Consistent progress
Doom scrolling → blog work swap Successful
Claude Code for tedious tasks Content import, to-do generation, redirect scripts all worked
"If I search and don't find the answer, publish when I figure it out" Content strategy in place

Action Items

Alex:

  • Hide non-essential elements (2 hrs budgeted)
  • Fix broken images in imported posts
  • Complete deployment with staging environment (10 hrs budgeted)
  • Set up redirect links (4 hrs budgeted)
  • Update essential content (remove stock/template text)
  • Write first blog post (favicon generation process)
  • Test everything works (images, RSS, URLs)

Andrea:

  • Post notes in Heartbeat

Planning

Task Estimate
Hide non-essential elements 2 hrs
Deployment + staging 10 hrs
DNS + redirect links 4 hrs
Content updates + testing ~4 hrs
Total remaining ~20 hrs

References

  • Content Collections library — separating content from display
  • Claude Code — content import, to-do list generation, redirect script creation
  • AWS S3 + CloudFront — hosting approach for redirect support

Next Session

  • Did the 4x estimation approach help with both personal project and work estimates?
  • How did the "Claude Code for context capture" experiment go when hitting blockers?
  • Is the blog live? If not, what blocked it?