Conversion credit is not incremental value.

1The owner question

Should Juniper cut the content program based on attributed conversions?

2What each team sees

Each team's report looks fine. Connecting them changes the decision.

3Connect the evidence

4What does not reconcile

• RECORD8 of 10 wins credited to branded search; 7 of 10 touched content first.
• INFERENCEJourney overlap does not prove causality either way.
• SCENARIOIllustrative branches only, not estimated treatment effects.

A difference between systems is not necessarily an error.

5Financial bridge

Client economics · one mature cohort (illustrative branches)

• SCENARIOKeep current mix: 10 × $20,000 − $150,000 = $50,000
• SCENARIOBounded experiment, favorable branch: 12 × $20,000 − $158,000 = $82,000
• SCENARIOBounded experiment, unfavorable branch: 9 × $20,000 − $158,000 = $22,000

6Compare decisions

7Why we might be wrong

INFERENCEContent has negligible incremental value and branded demand would occur anyway.

Evidence strength

Journeys recorded; causality untested.

Execution dependencies

Client willingness to test.

Financial sensitivity

Each win is worth $20,000 to the client.

Client or contractual approval

Client must approve experiment.

Timing uncertainty

Sales cycle length.

8Next action

Recommendation

Run a bounded, informative experiment before a broad channel cut.

Accountable role

Client Strategy Lead

Verification needed

Design a holdout test with the client.

Reversal trigger

Credible experimental evidence indicates negligible incremental value.

Excluded benefits

• Client economics are not agency revenue.
• No treatment effect is estimated.