You can explain the evidence, the assumption underneath it, and what would make you change the recommendation.
FOR PEOPLE WHO HAVE TO DEFEND THE ANSWER
Walk into the meeting knowing why you trust the answer.
AI can produce a polished analysis in minutes. Your job is deciding whether the question, evidence, definitions, assumptions, and recommendation are strong enough to act on.
AI suggests. Analyst decides.
◯ AI selects an answer. ✕ You challenge it. ◯ You decide what actually matters.
01Work faster
Get to a credible first answer sooner, without rebuilding every first pass by hand.
02Catch the weak assumption
Find the definition, exclusion, context, or missing fact that could overturn the result.
03Make the call
Move from an interesting pattern to a decision the business can actually use.
04Defend it
Explain what you know, what you checked, and what could still change your mind.
THE MOMENT THIS IS FOR
The analysis is finished. The decision is not.
You do not need another dashboard to tell you the result looks convincing. You need a reliable way to challenge the answer before somebody else does.
Before it goes into the deck, you know what the model could not know unless you supplied it.
You check whether a definition, process, source system, or reporting rule changed before you explain the pattern.
THE MECHANISM
Seven checkpoints between a plausible answer and a defensible decision.
The Seven C's of Successful Data Analysis gives you a repeatable place to look when an answer feels finished but the decision still carries risk.
01
THE FAILURE TO CATCH
Choose the right question.
The decision the analysis must support is never stated, so everyone can answer a nearby question beautifully.
AI CAN HELPGenerate candidate questions, summarize context, and propose measures.
ANALYST DECIDESWhich decision actually matters, who will act, and what would change the call.
JUDGMENT CHECK
What decision will change if this answer changes?
THE METHOD IN REAL LIFE
My Netflix model was mathematically right and commercially wrong.
The deciding cost lived outside the dataset. More analysis of the data I had would only have made the wrong answer more convincing.
Watch analyses like this free →
01 / MODEL
Heavy streamers looked extraordinarily profitable.
Streaming delivery cost pennies compared with mailing DVDs, so the model valued them highly.
02 / CONTEXT
The relevant fact lived in studio contracts in Beverly Hills.
Some streaming agreements triggered content payments that my dataset could not see.
03 / JUDGMENT
The answer changed because the frame changed.
No amount of additional analysis on the data I had would have found it.
WHY THIS IS MORE THAN AN AI WORKFLOW
The analytical method predates the current AI cycle. The Accidental Analyst was already being taught as a way to think.
EXTERNAL VALIDATION
Tableau co-founder and Stanford professor Pat Hanrahan put the book at the center of a Tableau Conference presentation and discussed its influence on how he approached teaching analytics at Stanford. The tools are changing. The need to know why you trust the answer is not.
WHY THIS MATTERSBetter tools do not eliminate analytical judgment. They make it easier to produce an answer before you have earned confidence in it.
EXPERIENCE3+ decadessolving analytical problems
TEACHINGThousandstaught live, in person
ORGANIZATIONS50+organizations and teams served
WHEN YOU WANT ANOTHER SET OF EYES
Bring the analysis you need to defend.
The free series teaches the method. The Working Lab applies it repeatedly to the forecast, dashboard, recommendation, or AI-assisted answer already on your desk, while the decision can still change.
01Bring the current answerReal project. Real deadline.
→
02Show why you believe itEvidence, logic, uncertainty.
→
03Pressure-test the frameAssumptions, gaps, definitions.
→
04Find what changes the callThe issue worth checking now.
→
05Leave with a decisionOr the exact test that settles it.
ILLUSTRATIVE LAB REVIEWExample only · not a participant result
AI explanation: slower pipeline conversion is the primary driver.
-8%forecast vs. plan
- Stage definition changed in July.
- Two enterprise renewals slipped in timing but are not lost.
- The AI summary treated both effects as conversion decline.
Rebuild the comparison on a consistent stage definition, separate timing slippage from true loss, then decide whether conversion actually deteriorated.
3 MONTHS
$2,450
Two live 75-minute sessions every week. Attend either or both.
NEXT ENTRYSeptember 1
LIVE SESSIONSTue + Thu · 1 PM CT
CAPACITY15 per session
EARLY ENROLLMENT2 private project reviews
Bring Your Work to the Lab
See exactly what you are buying →
Attend the first two live sessions. If it is not right for your work, request a full refund before Session 3 begins.
COMPANY PAYING?
Need manager approval or purchasing paperwork?
Company-card checkout and itemized receipts are available through enrollment.
FROM THE WORKING LAB
What participants found once they went looking outside the frame.
YOUR NEXT ANALYSIS
Before the answer leaves your hands, know what would change your mind.
Learn the Seven C's free. If you want live judgment applied to the work you already own, bring it to the Working Lab.
ONE-PAGE JUDGMENT CHECK
Eight questions before the answer leaves your hands.
No course gate. The complete method stays free.