Experiment Metrics
A test without a scoreboard is just a design change. In everyday language, a metric is “what we will count to decide if this worked” — clicks, purchases, signups. An event is the raw thing that happened on the page. A conversion is an event that met your success definition. In CROForge you do not invent metrics inside the campaign. You link metric definitions created in Data Vista. When those events fire on-site, Smart Code evaluates which metrics match and attaches the live experiment and variation ids so the Report tab can compare Control against variants.
Words used on this page
| Word | Everyday meaning | Meaning in CROForge |
|---|---|---|
| Metric | A number you watch to judge success (conversion rate, revenue, clicks). | A Data Vista definition: which event counts, optional conditions, how to calculate, and whether up is good. Campaigns link these definitions; they do not duplicate them. |
| Event | Something that happened (a click, a page view, a purchase). | The raw signal Smart Code sends (standard or custom). Metrics listen to events. An event is not automatically a conversion. |
| Conversion | A success moment you care about. | An event that satisfied a metric’s conditions. Reports treat that as progress toward the metric, not merely “something fired.” |
| Primary metric | The one number you will use to call a winner. | The main linked metric for the campaign. Decide this first so you do not get conflicting “winners.” |
| Secondary metric | Supporting numbers you check so you do not miss harm. | Additional linked metrics (for example revenue next to a click metric). Use them as evidence, not as a second independent winner. |
| Winner | The version that performed better. | A variation you promote after sample size and the primary metric support it. The product does not replace your judgment; the Report tab shows the comparison. |
| Data Vista | The catalog of measurement building blocks. | Create the metric under Data Vista → Metrics, then attach it on the campaign Metrics tab. |
Screenshot
Filename: experiment-metrics.png (campaign Metrics tab with linked primary and secondary metrics). Place the file under docs/assets/placeholders/. Paste the screenshot into WordPress after you copy this article; do not add an image tag here.
How to link metrics to a campaign
- Ensure the metric exists under Data Vista → Metrics (event, conditions, calculation type, direction).
- Open the campaign → Configuration → Metrics.
- Add / link metrics. Choose one clear primary metric; add secondary metrics only if they answer a real risk or supporting question.
- Save.
- When events fire on-site, Smart Code evaluates which metrics match and attaches live experiment / variation context so Report can attribute the conversion to the right bucket.
Design tips
| Tip | Why it matters |
|---|---|
| One clear primary metric | Two “main” metrics produce conflicting winners and stalled decisions. |
| Align the metric event with the variation change | A CTA color test should count clicks on that CTA, not every click on the page. |
| Use conditions (text, URL, amount) | Cuts noise so a “Book Demo” click is not mixed with footer clicks. |
| Choose unique visitor vs session vs event-count on purpose | The same events can tell different stories depending on calculation type. See Metric types in the reference. |
How this connects to events and reports
Smart Code sends events from the site. Independently, it applies a Running experiment’s variation and fires an impression when a visitor is shown that variation. When a later event matches a linked metric, the payload includes which experiment and variation were live. That is how the Report tab can say “Variant A converted more,” not merely “the site had more clicks.”
Link metrics before you set Running. A campaign that runs without a primary metric still assigns traffic and can change the page, but you will not have a trustworthy scoreboard. Do not call a winner from secondary metrics alone.
