The Segment That Cannot See the Purchase
A retail marketing team wants a segment of visitors who viewed the clearance page and then purchased within the same visit. An analyst builds the segment with two hit containers joined by AND and gets 0.4% of visitors, against an expected 3%. The same team wants a Data Warehouse export of 4.2 million rows and a calculated metric for revenue per visitor. The report suite uses a purchase event and a unique visitors metric. A separate campaign report shows that direct traffic converts better than paid traffic, which the team finds surprising.
1. Why does the segment return 0.4%, and how should it be rebuilt?
Read the worked answer
Two hit containers joined by AND require both conditions to be true on the same hit. A clearance page view and a purchase are different hits, so the intersection is nearly empty. That explains 0.4% against the expected 3%.
Build one visit container with two conditions and use Then sequencing: clearance page view, then purchase within the same visit. If the requirement is any time before purchase, use a visitor container with the same sequence. Do not use two hit containers. Confirm that the purchase metric is set on the purchase hit and that the segment is not accidentally excluding direct traffic through a marketing channel condition. After rebuilding, validate the segment size against the visit count in a small date range and check a few visitor IDs in a data feed.
2. What integration approach and limits matter for the 4.2 million-row export?
Read the worked answer
Data Warehouse is the right tool for a 4.2 million-row scheduled export; a dashboard export is not. Data Warehouse supports large row counts and scheduled delivery to cloud storage or FTP, but request row limits, concurrent request limits, and segment compatibility still apply.
Visitor- and visit-container segments work, but a complex segment with Then logic can behave differently outside the report suite interface, so test a small date range first. Include visitor ID and a date dimension so the file can be joined to the order system. For recurring delivery, use a scheduled request and a stable filename convention. Do not email a file of that size. Confirm the recipient has access to the report suite and that the Data Warehouse user has the right permissions.
3. How should revenue per visitor be defined and validated?
Read the worked answer
Revenue per visitor equals revenue divided by unique visitors. In a segment, the denominator is the unique visitors who qualify for the segment, not all visitors. State that in the metric description or the number will be challenged.
In Data Warehouse, calculated metrics are computed from aggregated columns, so division works only if both metrics are included at the same granularity. Use revenue and unique visitors at the same breakdown; do not divide a visit-level metric by a hit-level metric. Validate against a small date range in Reports & Analytics before scheduling. For the direct-versus-paid comparison, check whether the campaign report is using a last-touch channel and whether direct visits are being credited with purchases that started in paid; that is an attribution question, not a data-quality question.