How to create an experiment campaign
Test what works before you scale it. This guide walks you through experiment campaign setup in the Covatic platform, splitting your audience into controlled variants so you can compare creative, messaging, or attribution treatments against a like-for-like control. Give your advertisers evidence of what is driving performance, not assumptions, and use the results to justify the next spend commitment.
Step-by-step instructions
1. Start building your experiment
From the left-hand navigation menu, select Campaigns. Click the + (Plus) icon on the “Create a new campaign” banner at the top of the screen. Under the Experiment card, click Start.
2. Select your experiment split type
On the “Experiment Campaign” screen, choose one of the three available distribution models based on your testing goals:
- Standard A/B test (50/50): Splits your audience evenly between two variants (A: 50%, B: 50%) to determine which performs better.
- Multivariate A/B/C/D test (25% each): Splits your audience into four equal variants simultaneously to evaluate multiple creative or strategy options.
- Small trial (8.25% test): Allocates a small test group (8.25%) to receive a new variant while the vast majority (91.75%) serves as the control group.
Select your split option and click Next.
3. Enter your campaign details
On the “Campaign Details” screen, fill out the experiment metadata:
- Campaign Name: Enter a unique name for your experiment (for example, “Creative A/B test Q3”).
- Advertiser Name (Optional): Enter the brand or client name.
- Description (Optional): Add strategic notes on the test hypothesis.
Click Next.
4. Target your experiment audience
Define which users participate in the experiment split:
- All Audiences: Select this to run the experiment across your entire site or platform traffic.
- Select Audiences: Choose specific pre-existing audience segments from your library if you want to isolate the test to a targeted demographic or interest group.
Click Next.
5. Review and publish
On the Summary screen (Step 5 of 5):
- Review the distribution graphic and confirm the split percentages.
- Inspect the Experiment Audiences list, which displays auto-generated variant segment names (for example, [Campaign Name] – A and [Campaign Name] – B).
Click Save & Publish. A green banner reading “Campaign ‘[Name]’ created successfully” will appear at the top right of the dashboard, and your new campaign will show on the main Campaigns dashboard with a Live status.
6. Set up ad server targeting
Open the Campaign Details page for your new experiment. Under the Experiment Audiences section, you will see a table listing each variant alongside its unique Audience Code (key value). Target these exact key values in your ad server line items to deliver different creative variants or measure attribution split across your test groups.
Map creative line items 1:1 to the variant Audience Codes (for example, Line Item A targets Audience Code A, Line Item B targets Audience Code B) to guarantee accurate performance reporting.
Important: ad server dependency
The experiment will not deliver differentiated variants until the auto-generated key values are targeted in your ad server line items.
Action required: You (or your ad ops team) must add the generated key values to the relevant ad server line items to deliver different variants and capture the results.
Ready to run your first experiment?
Experiment campaigns run on the Covatic platform, alongside audience building, attribution tracking, and live campaign optimisation. To get set up you need a registered Covatic account, access to the Campaigns page in your dashboard, and access to your ad server to target the generated variant key values.
Don’t have an account yet? Request a demo or email [email protected] to see how the platform works with your inventory and data.
