# How do I set up and run a variant A/B test? Canonical URL: https://support.optimonk.com/en/articles/how-do-i-set-up-and-run-a-variant-a-b-test If you want to know which version of your popup performs better, this is the feature for you. A variant A/B test shows two or more versions of the **same campaign** to different visitors, so the numbers — not your taste — decide which one stays. ## Why it is worth it Testing makes the same traffic produce more subscribers and more customers. And you can play it safe: if you run five tests and three of them improve conversion, you keep those three and drop the rest. Nothing is lost. --- ## Where the variants live Open **Campaigns** in the left menu and click a campaign. On the **Overview** tab, under the metrics band, is the variants table: | Traffic | Variants | Active | Impressions | Submits | Submit rate | Uplift | | --- | --- | --- | --- | --- | --- | --- | Every row is one arm of the test. Underneath it sits **+ Add A/B test variant**. --- ## Step 1: Add a second variant Click **+ Add A/B test variant**. You get three options: - **Control (no popup)** — *A holdout bucket that sees nothing.* Use this to measure what the campaign is worth at all (see A/B test with and without an OptiMonk campaign). Only one per campaign; once it's there the item reads *Already added*. - **Duplicate a variant** — *Clone an existing design to tweak.* Hover it and pick which existing variant to copy. This is the usual choice for a headline or button test: you get an identical copy and change one thing on it. - **Create with AI** — *Generate a new design in the designer.* OptiMonk builds a genuinely new design for the same campaign instead of copying one. If the campaign has no design session behind it, the item is greyed out and says *Unavailable — this campaign has no design session*. You can also start from an existing row: open the **⋮** menu at the end of it and choose **Duplicate**, or **New design from this variant**. While the variant is created you see *Duplicating variant…* with the note *This can take a few seconds — it's syncing with OptiMonk.* A variant created with AI takes longer — its row shows **Building · step 3 of 8** until the popup is finished. ### Adding a second design while you are still building the campaign If the campaign isn't finished yet, you meet the same decision earlier, on the **A/B test** step of the campaign worksheet. There you pick from the designs you already have: - **{count} designs go head to head** — *Visitors get one of them at random, in equal shares. You can follow the result in the campaign's analytics.* - **One design for now** — *You can go on like this. Pick one more, though, and the two run against each other — the numbers then tell you which one performs.* - Under **You could put these up against it**, click **Add it** on any other design from the concept board. Then **Continue to targeting**. With three or more variants, note the warning the step itself shows: *With three or more variants a reliable result takes longer to build up than with two. * --- ## Step 2: Switch the variants on Each row has an **Active** toggle. Only active variants take traffic — a variant that is switched off is not part of the test. Make sure every arm you want to measure is on. You can also **Rename** a variant from the **⋮** menu so the table reads like your test ("Headline — urgency", "Headline — discount") rather than "Variant 2". --- ## Step 3: Change one thing Click a variant's thumbnail or **Edit variant** to open it in the editor, then change what you want to test. Either edit the element directly on the panel, or describe the change on the **Edit with AI** tab ("make the CTA button much more prominent"). We recommend changing the elements in this order: - Headline - Subheading - Button text - Data asked - Background - Image - Colors - Other texts and descriptions Beyond the content of the popup, keep refining your main **offer**. Your customers might react ten times better to a different one — a discount may beat a free ebook, or the other way round. Every site has its own visitors with their own needs, and testing is the only way to find out. **Please note:** we do not recommend making too many changes at once, because then you won't be able to tell *why* a variant is performing better or worse. A sensible way to start: change one element, test it, keep the winner, then move on to the next element. ### Let OptiMonk suggest the test You don't have to come up with the idea yourself. In the chat panel next to the campaign, ask **"Suggest an A/B test I could run"**. OptiMonk first asks what kind of test you have in mind: - *Test the copy only (headline, CTA…)* - *Change the offer / incentive* - *Try a whole new design variant (takes longer)* - *Surprise me — mixed ideas* Then it comes back with concrete ideas, each with a preview. Click one to open **Variant comparison** — the **Control · current** version on the left, the **Suggested** one on the right, plus **Why this works** and **What changes**. **Add variant** turns the idea into a real variant on the table. If none of them fit, click **None of these — show me other variants**. --- ## Step 4: Traffic Traffic is shared **equally** across all active variants — two variants, half each; three variants, a third each. Click the number in the **Traffic** column to open **A/B test settings** and see the split per variant plus the total. The **Evenly split** checkbox is marked **Always on**: *Traffic is shared equally across all active variants.* Setting your own percentages is not available yet — clicking the checkbox says *Custom traffic split is still in progress. For now, traffic is always distributed evenly across active variants.* --- ## Step 5: Read the result and pick the winner While the test runs, the Overview tab shows a green **A/B test running** banner naming the leader — *… ****is leading by**** +12.40% (81% chance to win)* — or *Data collection in progress.* before there is enough data. Two numbers matter: - **Uplift** — how much better (or worse) a variant converts than the baseline. The baseline row shows a dash. - **Chance to win** — how confident the result is. Each row shows *{pct}% to win*. **Let it end itself.** In **A/B test settings**, **Auto-declare winner** is on by default: *Automatically end the test when a clear winner emerges.* It fires only when all three conditions are met — **Declare at 95% chance to win**, **Min conversions 200**, **Min days running 3**. Change any of them there. **Or call it yourself.** Click **Declare winner at 95%** in the banner. Below your threshold OptiMonk warns first — *Ending the test now may pick a winner that isn't statistically reliable yet* — and you can **Keep testing** or **Declare anyway**. Once the winner is declared, it stays active and the other arms stop. --- ## Housekeeping From a row's **⋮** menu you can **Rename**, **Duplicate**, create a **New design from this variant**, or **Delete** it. The last remaining variant cannot be deleted — a campaign needs a design to run.