Lead quality, before
the creative brief.

• One opportunity per property/project. Repeated contacts remain submission events, not extra opportunities. The all-source log has 19 events: 16 first inquiries, 2 duplicates, and 1 test. DEMO labels identify fictional records.
• 14 eligible Meta opportunities. DEMO-15 is an older Meta acquisition; DEMO-16 is organic. Both are excluded from these cost metrics. Genuine poor-fit inquiries remain included.
• 16 platform-reported leads are a separate measure. Platform CPL is $700 ÷ 16 = $43.75. Platform events and source-linked opportunities have different definitions; the file does not establish that their difference is solely duplicates.
• Qualification coverage: 8 ÷ 14 = 57.1% confirmed qualification yield. (8 + 3) ÷ 14 = 78.6% assessment coverage. The assessed-only rate is 8 ÷ 11 = 72.7%, which must be read with that coverage.
• Estimate completion under the latest schedule: 5 ÷ 6 due = 83.3%. DEMO-08 moved to Oct 8, beyond the snapshot; DEMO-10 is a no-show. The workbook uses the latest scheduled date, so rescheduling can change the due denominator.
• Customer verification: DEMO-11 is Won with unknown new-customer status; it is neither assumed new nor classified returning. The demo assumes this spend is for new-customer acquisition; real allocation requires verification.
• Activity is not acquisition. Five all-source wins by win date include an older Meta lead and an organic lead. They are not the denominator for this cohort’s CAC.
• Not job completion or revenue. A completed estimate is an assessment. A won opportunity is a recorded sale under the scenario; neither proves a finished installation or collected payment.

Source: S3, Summary and Submissions; counting definitions: S2 §§1, 6–8. Source linkage is synthetic, not a demonstrated Meta/CRM integration.

Fictional records from the lead tracker. “Meets” = Yes; “Fails” = No; “Unknown” = not yet confirmed. The table contains no real customer information.

Scroll horizontally for the full table. Source: fictional tracker (S3), Leads rows 7–20. The tracker uses “Closed” for disqualified records; the plan uses a different stage vocabulary. A shared stage mapping is required before live use.

What this means DEMO-05 and DEMO-13 lack intent confirmation; DEMO-12 lacks authority confirmation. Keeping these unknowns visible prevents premature creative judgments.

Review verified customer outcomes alongside qualification: an apparent CPQL improvement could still produce fewer customers. Attention and click metrics can diagnose an execution; they cannot establish a business winner on their own. No numeric benchmark or expected uplift is assumed.

What this means Each metric needs an agreed definition, a responsible data source, and a recording method. A formula cannot repair missing sales evidence.

Evidence boundary. Public research supplied the business context. A mock client interview supplied the operational scenario. Synthetic records supplied the numerical demonstration. No private account export, CRM access, live tracking audit, or business outcome verification was available.

• S1 — Home_Shield_Insulation_Account_Pack_v1.txt. Dated Oct 3, 2026. §§1–3 support public/proposed scope and unknown settings; §§6–9 document economics and creative limits.
• S2 — Home_Shield_Lead_Qualification_Measurement_Plan_v1.txt. Dated Oct 3, 2026. §§1–8 define counts, qualification, status evidence, sources, and formulas; §§9–11 cover proposed actions and approvals.
• S3 — Home_Shield_Fictional_Lead_Tracker.xlsx. Dated Oct 3, 2026. Summary A11:C35 supports counts and ratios; Leads rows 7–22 contain the 16 fictional opportunities; Submissions retains 19 events. This page shows the 14 selected-cohort records and explains the two exclusions.
• S4 — Marianne_Jaen_Home_Shield_Case_Study(1).pdf. Portfolio summary revised Oct 13, 2026. Pages 1–4 summarize the scope, Oct 3 fictional snapshot, priorities and proposed creative test. The revised PDF is a companion to this case, not an independent source of new results.

Source reconciliation. The account measurement brief records company performance as unknown. The later fictional tracker is a demonstration, not evidence of actual company results. The plan requires confirmed intent/contact; “Closed” in the tracker needs mapping to the agreed sales stages before live use. The Oct 13 revision recalculates the figures displayed in this case and reconciles them with the portfolio PDF. The original S1–S3 files were not re-opened for this revision; their references are retained from the existing case. Original workbook formulas, deduplication logic and source timestamps were not independently re-audited.

Public references recorded Oct 3, 2026: the brief records the Home Shield locations page and an indexed Upwork home-improvement brief, observed Oct 3, 2026. The listing’s advertised lead goals have unknown scope and effective dates and are not approved targets for this scenario. The job-brief reference was available only as an indexed excerpt. Public statements do not verify actual delivery or results.

Reporting method. The Sep 14–20 acquisition window and Oct 3 outcome cutoff remain unchanged in this revision. This is a single fictional snapshot with no comparable prior-period baseline. Use a source-linked acquisition cohort and a fixed outcome cutoff. The source plan proposes a first-known-source model, not Meta attribution or proof of incremental impact. Recompute ratios from totals; retain pending, exclusions, and unknown new-customer status. Changing a cutoff alone cannot reconstruct historical states from latest-status fields.

Contribution. I developed the qualification and measurement framework, checked the fictional calculations and translated the evidence into proposed creative applications. Research, analysis and report production used AI assistance; live client implementation was outside the project scope.