Before you commit budget, headcount, roadmap, or architecture, paste the plan into Faultline. It shows the part most likely to break, the test that would catch it, and the line that should make you change course.
Not a brainstorming tool. Not a second opinion that politely agrees.
A plan that fails this review should not be made yet.
Cut enterprise sales after a 10% lift in self-serve signups. If product-led growth is working, sales looks redundant.
Analytics records the conversion path, not the influence chain that preceded it. Sales-nurtured prospects register through self-serve and appear organic.
What happens when you submit a plan
Most plans do not fail because the team was stupid. They fail because one important assumption was never tested hard enough. Faultline finds that assumption and tells you what to check before you commit.
The team wants to reduce enterprise sales investment. Self-serve signups grew 10% last quarter. The logic seems reasonable: if product-led growth (PLG — meaning customers are finding, trying, and buying through the product instead of talking to a salesperson) is working, the company may not need as much sales coverage.
The sales cut is close to becoming the plan.
Primary finding
The 10% signup lift may be real, but the interpretation is fragile: enterprise prospects influenced by sales can still convert through self-serve and appear organic in the analytics stack.
Diagnostic
Cross-reference self-serve signup domains against CRM and marketing touchpoints from the prior 180 days.
Threshold
>25% prior sales contact: stop the cut. >40%: redesign attribution before personnel decisions.
The 10% self-serve growth is likely contaminated by prior sales activity. Enterprise prospects nurtured by sales are converting through the self-serve motion and being recorded as organic PLG demand. Your attribution system records the final action — not the influence chain that preceded it.
The signal may be real. The interpretation may be false.A prospect who received outreach, attended a demo, built internal consensus, and then self-registered appears indistinguishable from a genuinely organic user in a standard analytics stack. The internal read is self-reinforcing: leadership wants PLG to be working. The data, as reported, shows it is. The challenge does not come from the numbers — it comes from the attribution model behind them.
Sales headcount is cut based on apparent PLG momentum → awareness drops → self-serve growth plateaus → pipeline weakens → revenue misses appear one to two quarters later → the attribution error is discovered after the damage is done.
Pull all self-serve signups from the past 90 days. Classify email domains. Cross-reference CRM and marketing touchpoints from the prior 180 days. Calculate what percentage of “organic” self-serve signups had prior sales influence.
If >25% of self-serve signups share domains with prior sales contacts, do not cut sales. If >40%, halt the reallocation and redesign attribution before any personnel decision.
Do not commit before running the domain auditThe risk is not that PLG is false. The risk is cutting the sales engine before proving PLG is independent.
The summary above gives the call. The full diagnostic shows the reasoning chain behind it: threshold bands, decision asymmetry, known exceptions, safer path, and what would change the recommendation.
Decision asymmetry preview
Delay reallocation 60–90 days.
Defund a causal input to growth and discover the attribution error 1–2 quarters later.
Threshold band preview
| Below 20% | Independent PLG plausible → Proceed to no-touch cohort validation |
| 20–35% | Mixed signal → Hold reallocation until cohort evidence improves |
The full diagnostic includes the complete threshold table, known exceptions, safer path, and supporting analysis.
What Faultline produces
Faultline turns a submitted plan into a reviewable decision artifact. The packet preserves the plan, isolates the load-bearing assumptions, and states what would change the decision.
Captures the goal, assumption, constraint, and commitment point before the packet evaluates the decision.
View PLG packet → Assumption LedgerIdentifies the load-bearing assumptions and where the evidence path may be contaminated.
View sales packet → Wrong-If SignalStates what evidence would weaken, overturn, or release the hold before commitment.
View GovCon packet →The output is not a chatbot answer. It is a public-safe diagnostic packet with a visible hold / no-hold rationale.
When to use Faultline
Faultline is strongest when the plan is plausible, the stakes are material, and reversal would be slow, political, or expensive.
More Decision Hold Patterns
Causal contamination map
Most bad commitments do not start with obviously bad data. They start when a plausible assumption begins selecting evidence, moving thresholds, and normalizing exceptions before anyone has defined the line that would change the decision.
The team adopts a reasonable premise and starts planning around it.
Signals that support the plan get more attention than signals that would slow it down.
The proof needed to proceed becomes softer because the commitment path is already forming.
A known gap is treated as manageable without a written condition or owner.
Faultline marks the missing evidence, threshold, or condition that should pause the commitment.
Public demo library
Demo packets use public-source patterns to explain decision risk without publishing raw proof notes or making named-company accusations.
Browse Decision Hold Library → View Public Source Appendix →What makes this different
Faultline has no reason to protect the plan.
Request review
Send the decision in plain English. What are you considering— Why does it seem right— What would make being wrong expensive—
Early-access reviews are human-checked before delivery.
Questions: rob@faultline.me
What to include
What we're considering
The decision, the plan, or the move you are about to make.
Why it seems right
The evidence, logic, or signal making this look like the right call.
What would make being wrong expensive
Budget committed, headcount hired, architecture locked, contracts signed — whatever would be hard to undo.
Do not paste confidential documents. Strategic context in plain language is enough. Early-access reviews are returned by email. Timing depends on review volume.