The corpse on the table (composite)
Concept: “Unified inbox intelligence” for SMB service firms — AI triage, routing, sentiment flags — ambitious integrations Week One fantasy.
Energy: founder conviction sky-high — micro-case studies, slick prototype.
Kill catalyst: not cruelty — collision.
What died first — distribution physics
Growth persona shredded CAC realism: outbound + SMB + noisy differentiation = long payback on thin margin once inference/support counted.
Not haters — math schedule.
Second wound — buyer specificity rot
“SMB service firms” hid incompatible workflows — agencies ≠ HVAC dispatch psychology.
PM optimism bundled personas — segment liquefaction.
Kill verdict subsection: pick one vertical bloodily or stop pretending software solves sociology.
Third wound — technical spine honesty
CTO lane surfaced integration reliability trap — promises became ops theater.
Maintenance avalanche probability > novelty retention.
Synthesis — verbal verdict flavor
Narrative collapsed toward:
Stop broad wedge — if revived: vertical choke + concierge proving ROI weeks — else abandon.
No participation trophy — gratitude privately after denial faded.
Emotional timeline (honest)
Hour zero: defensive adrenaline
Hour six: quiet recognition — spared churn hire
Week two: sharper narrow experiment spawned — healthy lineage
Killing an idea ≠ killing founder velocity — often unlocks honest next bet.
After-action artifacts worth stealing
| Artifact | Purpose |
|---|---|
| Written kill criteria pre-session | Prevents moving goalposts |
| Segment razor — max three hypotheses | Stops blob ICP |
| 14-day falsifier experiment | Cheap burial |
Why one session sufficed
Parallel specialist pressure exposes compound failure faster than sequential doubt.
Same engine documented for stress-testing — run it before code becomes emotional concrete.
Related reading
Disclaimer
Composite educational scenario — not business counsel.
Kill early so your calendar dies of old age building the right thing — not the comfy thing.