Skip to content
flowss
Start freeChecking your session
Decision tree templates

Decision Tree — Test Market Before a National Launch

The value-of-information classic: a regional test with Bayes-consistent posteriors (0.694 after a favourable read, 0.118 after an unfavourable one) is worth $4,421k against $3,600k for launching blind, so the $800k test pays for itself and the tree abandons after a bad read. Illustrative values.

Decision treeBusiness & StrategyDecision tree syntax docs
Template previewDecision tree
Run a regional test market before the national launch?Run a regional test market before the national launch?Contribution NPV, US$ thousands — posterior probabilities by Bayes' rule3 decisions · 4 chance nodes · 9 outcomes · maximise expected valueOPTIMAL FIRST CHOICERun a 12-week test market·EV $4,421kLaunch nationally nowAbandon the productAbandonLaunch nationallySuccess0.118Failure0.882Success0.4Failure0.6$0net −$800kNational outcomeEV −$3,168k$18,000knet $17,200k−$6,000knet −$6,800kNational outcomeEV $3,600k$18,000k−$6,000k$0Run a 12-week test marketcost $800k · net $4,421kFavourable0.49Unfavourable0.51Launch nationallySuccess0.694Failure0.306AbandonTest first?EV $4,421kTest resultEV $5,221kAfter a favourable readEV $10,656kNational outcomeEV $10,656k$18,000knet $17,200k · P 0.34−$6,000knet −$6,800k · P 0.15After an unfavourable readEV $0$0net −$800k · P 0.51DecisionChanceOutcomeOptimal strategyRejected alternativeRisk profile·cumulative probability by first choice0%25%50%75%100%−$10,000k−$5,000k$0$5,000k$10,000k$15,000k$20,000kfinal outcome, net of costsFIRST CHOICEEVRun a 12-week test market$4,421kσ $9,398k · P(loss) 0.66Launch nationally now$3,600kσ $11,758k · P(loss) 0.6Abandon the product$0σ $0 · worst $0What the rollback showsBest first choice: “Run a 12-week test market”, EV $4,421k — $821k ahead of the next best, “Launch nationally now” ($3,600k).If the path reaches “After a favourable read”: “Launch nationally” ($10,656k) over “Abandon” ($0).If the path reaches “After an unfavourable read”: “Abandon” ($0) over “Launch nationally” (−$3,168k).Following it, the outcome ranges from −$6,800k to $17,200k across 3 possible results, standard deviation $9,398k; a 66% chance of a loss.Note: Prior P(success) = 0.40. The test reads favourable for 85% of eventual successes and 25% of failures, so P(favourable) = 0.49, P(success | favourable) = 0.694 and P(success | unfavourable) = 0.118.Note: Illustrative values for a fictional consumer-goods launch.

Make it your own.

title "Run a regional test market before the national launch?"
subtitle "Contribution NPV, US$ thousands — posterior probabilities by Bayes' rule"
currency $
unit k
risk-profile compare
note "Prior P(success) = 0.40. The test reads favourable for 85% of eventual successes and 25% of failures, so P(favourable) = 0.49, P(success | favourable) = 0.694 and P(success | unfavourable) = 0.118."
note "Illustrative values for a fictional consumer-goods launch."

decision "Test first?"
  "Run a 12-week test market" cost 800 -> chance "Test result"
    "Favourable" p 0.49
      decision "After a favourable read"
        "Launch nationally" -> chance "National outcome"
          "Success" p 0.694 -> 18,000
          "Failure" p rest -> -6,000
        "Abandon" -> 0
    "Unfavourable" p rest
      decision "After an unfavourable read"
        "Launch nationally" -> chance "National outcome"
          "Success" p 0.118 -> 18,000
          "Failure" p rest -> -6,000
        "Abandon" -> 0
  "Launch nationally now" -> chance "National outcome"
    "Success" p 0.40 -> 18,000
    "Failure" p rest -> -6,000
  "Abandon the product" -> 0