Draft the campaign with honest limits

Question for this chapter

Which campaign should you propose, and what cannot be claimed from the current evidence?

Confirm campaign scale on the dashboard

Open the sales_overview dashboard under analytics. All five widgets use customer_analytics.

• Total Customers (statistic) — 22 customers
• Total Revenue (statistic) — $1,052.76 in total spend
• Customer Segments (donut) — the 7, 11, 4, and 0 distribution
• Revenue by Region (bar) — us-east leads at $475.18
• Avg Order Value by Tier (bar) — platinum leads at $15.03

Draft three campaign lanes:

1. Seven champion customers — recognition and early access rather than blanket discounting
2. Eleven loyal customers — a repeat-purchase suggestion tied to the latest order
3. Four new customers — product guidance and a second-purchase onboarding message

Hold the at_risk campaign until there is an audience.

Inspect purchase paths and the evidence boundary

Open ontology_materialization and select Run. On completion, the graph contains 22 customers, 88 orders, and 9 products. It also contains 88 customer-order relations and 88 order-product relations.

Run this query in Explorer to inspect CUST-017's purchase paths.

MATCH path=(c:ec_customer)-[:ec_purchased]->(o:ec_order)-[:ec_contains_product]->(p:ec_product)
WHERE c.customer_id = 'CUST-017'
RETURN path
LIMIT 25

The result should contain seven paths from the customer through orders to products.

Interpret the result

You can answer which segment receives which message and inspect the products bought by CUST-017. However, PRD-UNKNOWN does not represent a real product. Exclude it or resolve the source identifier before using it for product-level personalization.

Next decision

You have three campaigns to run and two scopes to hold: at_risk outreach and unknown-product personalization. Next, check the evidence needed for the final explanation in four questions.