Every lesson.
Each one is a single real product decision, told in frames. Filter by the shape of the story — an investigation reads differently from a dilemma.
01FoundationsThe Users Who Aren't There
Maya's survey ran as an in-app banner. Whose answers could it never have collected?
02FoundationsYou Asked Until You Heard Yes
Six of six said the shared inbox would help. What did that actually establish?
03CoreThe Feature That Caused Nothing
Boards users retained at 71% against 24%. Why did shipping Boards not move retention?
04CoreThree Angry Emails
The complaint rate is 0.07%. What is the right conclusion?
05FoundationsThe Faster Tool Nobody Switched To
The new importer is four times faster and 1,199 accounts stayed put. What does that tell you?
06CoreEveryone Loved It For Nine Days
Eleven users asked for the summary, praised it, and stopped opening it in nine days. What went wrong?
07FoundationsEighty-Nine Percent Chose Nothing
Reversing the pre-selection moved 71 percent onto compact. What does the original 89 percent measure?
08CoreSeven Ways To Say No
Conversion fell 31 percent for first visits and rose 2 percent for upgraders. What does that rule out?
09FoundationsIt Says Two Minutes
Exports still take up to 38 minutes after the change. Why did the complaints stop?
10PractitionerThe Third Apology
Sixty-eight percent read a genuinely good postmortem and churn stayed high. What were they responding to?
11FoundationsSame Rows, Different Question
Company size showed 42, 44 and 41 percent. The same accounts by first job showed 71, 63 and 18. What happened?
12CoreThe Competitor With No Website
The team wins 11 of 14 rows against the rival and loses 58 percent of deals to no decision. What is the matrix missing?
13FoundationsThe Tip That Worked And Cost Us
The tip cut errors 34 percent among readers and cut overall completion by four points. Which measurement is wrong?
14CoreForty Percent More Time
Time in app rose 40 percent and exports fell 22 percent. What made the duration number unreadable on its own?
15CoreNobody Turns On Beginner Mode
Beginner mode cuts ramp-up from three weeks to four days and 2 percent enable it. What is the barrier?
16PractitionerThe Tier That Would Work
The entry tier clears its target and may dilute what premium customers are buying. What settles it?
17FoundationsSix Weeks For Three Accounts
Six weeks of work, three accounts, 0.1 percent of the base. When is that ratio the wrong one to use?
18CoreForty-Nine Compared To What
Conversion went from 1.2 to 3.8 percent with no change to the price. What was actually wrong before?
19CoreThe Tickets That Moved
Tickets fell 91 percent and the same questions now stall 11 demos a week. What made this invisible for six weeks?
20PractitionerThree Ships Behind
A team ships weekly, reads a metric with a 21-day window, and swings 20 points a week. What is the cause?
21CoreNobody Cheated
Deals rose 31 percent, ninety-day churn rose 74 percent, and nobody broke a rule. What should change?
22PractitionerThree Date Pickers
Three teams built three date pickers, all for sensible reasons. What is the fix?
23PractitionerThree Stories, One Month
Three explanations fit the same churn spike. What is the most useful next step?
24PractitionerOne More Cut Of The Data
A one-day build, reversible in a day, has absorbed fifteen days of analysis. What went wrong?
25PractitionerEverything At Once
Escalations, a survey, an error spike and a deadline all point different ways. What decides the order?
26PractitionerRight For The Wrong Reason
The decision worked and the reasoning behind it was wrong. Why does that matter?