Create Experiment

Documents / AB Testing

Create a campaign

Creating a campaign is how you turn a guess into a test. In everyday language, you write down what you think will improve the page (a hypothesis), name the effort, and start in a safe Draft so nothing changes on the live site yet. In CROForge, Create on the A/B list makes a new campaign (the UI word) which is stored as an experiment (the API word). You then walk Configuration — Pages, Variations, Metrics, Audience — review the summary, and only then start it.

Words used on this page

WordEveryday meaningMeaning in CROForge
CampaignA named effort you run for a period of time.What the A/B list and hub call this test. Same record as an experiment.
ExperimentA structured test with a method and a result.The API and route name. After create, you land on /workspace/{id}/ab-test/{experimentId}.
HypothesisA clear “if we change X, Y should improve because Z.”Not a separate field you must fill, but the title should state the idea and the surface (page or flow) so the list stays readable.
DraftA work-in-progress that is not published.Status DRAFT. Smart Code does not apply Draft campaigns on the live site.
ControlThe original you compare against.Created for you as the baseline variation, usually with empty JS and CSS.
All Traffic“Everyone who shows up.”The default audience segment with no conditions. Anyone who already passed domain and URL checks can enter until you pick a tighter segment.
ConfigurationThe setup steps.Hub tab with sub-tabs Pages, Variations, Metrics, and Audience. Use Prev, Next, Save, and Discard.
ReviewA last look before launch.Hub tab that restates targeting, variations, metrics, and audience so you can confirm before you set Running.

Screenshot

Filename: create-ab-test-modal.png (the Create campaign dialog with the title field). Place the file under docs/assets/placeholders/. Paste the screenshot into WordPress after you copy this article; do not add an image tag here.


Create a campaign step by step

  1. Open a workspace and go to A/B (the Campaigns list).
  2. Click Create.
  3. Enter a title that states the hypothesis and the surface, for example “Pricing page – annual toggle copy” or “Checkout – trust badges.” Avoid “Test 1.”
  4. Confirm. The campaign opens as Draft on /workspace/{id}/ab-test/{experimentId}.
  5. Work through Configuration in order: Pages (where), Variations (what, including traffic split), Metrics (how you measure), Audience (who).
  6. Open Review and confirm the summary matches what you intended.
  7. When targeting, code, metrics, and audience are ready, start the campaign (status Running). Until then, visitors never see it from Smart Code.

Naming tips

A good title is a sentence your future self can scan in the list. Put the page or flow first, then the change. That is how a hypothesis becomes searchable.

Good titleWhy it helpsAvoid
Homepage hero – green CTANames the page and the change.Test 1
Checkout – trust badgesNames the flow and the idea.New experiment
Pricing page – annual toggle copyStates surface plus hypothesis in one line.Untitled / copy of copy

What you get after create

CROForge does not leave you with an empty shell. Typical defaults:

  • A Control variation (baseline, often empty JS/CSS).
  • Starter or empty URL targeting on Pages — you still need to add Include rules before a real launch.
  • Audience often set to All Traffic (a segment with no conditions) until you choose a tighter segment from Data Vista.

You still need to add at least one alternative variation, link metrics, and validate URLs. Draft only means “not live yet,” not “ready to start.”

List actions after campaigns exist

From the campaigns list you can often bulk start, pause, archive, or delete. Treat bulk start as going live for every selected Draft. Treat bulk pause as stopping every selected Running test. Use bulk delete / trash only when you are sure those campaigns should leave the working list.

Do not start a campaign from the list until Pages, Variations, Metrics, and Audience have been reviewed. A Running campaign with empty targeting or no primary metric will either apply in surprising places or produce a report you cannot trust.

Suggested order after create

  1. Pages & URL targeting — decide where it may run.
  2. Variations — name Control and variants, write JS/CSS.
  3. Traffic split — Equal or Custom, percents summing to 100.
  4. Experiment metrics — link a clear primary metric.
  5. Experiment audience — keep All Traffic or attach a segment.
  6. Preview mode, then Experiment lifecycle & statuses to go Running.
  7. Campaign reports after you have sample size.

Related reading