Campaign Reports
After a campaign has been Running, you need a scoreboard — not a guess. In everyday language, a report is “how each version did.” In CROForge, reports are per campaign. There is no separate global BI dashboard for A/B. Open the campaign hub Report tab to see date range, charts by variation, metric graphs, and tables. Traffic source on that screen is a reporting filter (direct, referral, social, search, paid, and similar). It is not Audience. Do not call a winner until sample size per variation is large enough and you have looked at the primary metric first.
Words used on this page
| Word | Everyday meaning | Meaning in CROForge |
|---|---|---|
| Report | A results summary after something has been running. | The hub tab (and report route from the experiment sidebar) for this campaign only. Charts and tables by variation and linked metrics. |
| Traffic source | How someone arrived (typed the URL, came from Google, clicked a social post). | A reporting filter (all / direct / referral / social / search / paid / non-paid, depending on the build). It does not decide who enters the live test. That is Audience (a segment). |
| Sample size | How many people you observed before you trust the comparison. | Counts per variation (impressions and conversions). A “winning” line on a tiny sample is noise. |
| Winner | The version you decide to keep. | A human call after Running coverage, sample size, and the primary metric agree. Then mark Completed / Archived. The report does not ship the variant to 100% of traffic by itself. |
| Impression | A count of “this was shown.” | The event Smart Code sends when a visitor is assigned/shown a variation. Use it as the denominator for “how many people saw this version.” |
| Primary metric | The one number you will use to decide. | The main linked Data Vista metric. Read it first; use secondary metrics only as supporting evidence. |
| Audience | Who was allowed into the test. | The live segment on Configuration → Audience. Some builds also let you slice report data by segment; that slice does not change live assignment. |
Screenshots
Filename: campaign-report.png (Report tab with variation charts and metric tables). Place the file under docs/assets/placeholders/.
Filename: campaign-report-filters.png (date range and traffic source filters). Place the file under docs/assets/placeholders/. Paste both screenshots into WordPress after you copy this article; do not add image tags here.
Where to open a report
Reports are per campaign. Open the campaign → Report tab (or the report route from the experiment sidebar). The hub route is still /workspace/{id}/ab-test/{experimentId} with the Report tab selected.
What you typically see
- Date range filter — compare only the days the campaign was actually Running.
- Traffic source filter — all / direct / referral / social / search / paid / non-paid (labels depend on the build).
- Charts by variation over time.
- Metric graphs for attached Data Vista metrics.
- Tables of variation performance (impressions, conversions, rates).
Traffic source on the report is a reporting filter. It is not the same as Audience (the Data Vista segment used for who enters the test). Changing the filter never changes sticky assignment on the live site.
Segment filter on the report
Some builds allow filtering report data by segment (see the report UI / segment-filter). Use this to slice results — for example “how did mobile look inside an All Traffic test.” It does not change live assignment and it is not a substitute for setting Audience before you launch.
How to read results
- Confirm the campaign was Running for the selected date range. Draft, Paused, Completed, Archived, and Trash do not collect new live assignments.
- Check sample size per variation (impressions at minimum) before calling a winner. Tiny counts lie.
- Focus on your primary metric first. That is the decision number you chose on the Metrics tab.
- Use secondary metrics only as supporting evidence (did revenue fall while clicks rose?).
- Remember mid-flight JS/CSS/URL/segment changes can mix two experiences under the same variation name. If that happened, treat the report as two eras, not one clean test.
- After a decision, mark Completed then Archived so the working list stays honest. Completing the campaign does not automatically roll the winner out to 100% of traffic — ship that change in the site or a follow-up campaign.
