How much extra revenue does your popup actually bring in? The only honest way to answer that is to keep a slice of your traffic away from the campaign and compare. That slice is called a control: a holdout bucket that sees nothing.

In OptiMonk you don't have to fake it with code. The campaign's A/B test has a built-in control arm:

  • Your design: visitors who see the popup.

  • Control (no popup): visitors who see nothing at all.

Traffic is split between the two, and the campaign's Overview tab shows the uplift and the chance to win. You don't need Google Analytics for the comparison, though you can still cross-check there if you want to.

Add the control arm

  1. Open Campaigns in the left menu and click the campaign you want to measure.

  2. Stay on the Overview tab. Under the metrics band you see the variants table.

  3. Click Add A/B test variant under the table.

  4. Choose Control (no popup): "A holdout bucket that sees nothing". While it is being created, you see "Adding control variant…" and "This can take a few seconds — it's syncing with OptiMonk."

  5. The control appears as a new row in the table. It has no preview image. Instead of a thumbnail it says No variant will be shown, which is exactly what it does.

You can only add one control per campaign. If one is already there, the menu item is greyed out and says Already added.

Make sure both rows are switched on. The Active switch in each row decides whether that arm takes traffic. A control that is switched off is not measuring anything. If you switch your design off, the control is switched off with it, because a control on its own would show nothing to anyone.

Pick the right goal

This step matters. Visitors in the control see no popup, so they can never submit one. With the default Submit goal, the control has nothing to count, and no uplift can be calculated.

Measure what happens on your site instead. In the metrics band, open the goal dropdown (the one with the target icon, next to the date range) and choose Order. The table columns change to Visitors · Orders · Order rate · Uplift. Now both arms are counted the same way: how many of their visitors placed an order.

How the traffic is split

By default, traffic is shared equally across all active variants. With one design and one control, half your visitors see the popup and half see nothing.

You can change that. Click the percentage in the Traffic column to open A/B test settings. Under Traffic share, type a share for each arm, for example 90% for your design and 10% for the control, and click Save. Each arm needs 1–99%, and the total must be 100%. To go back to an equal split, tick Evenly split.

A smaller control costs you less, but it takes longer to collect enough data. If the split can't be edited, the panel tells you why.

Read the result

Once both arms have traffic, the variants table on the Overview tab answers the question.

The banner above the table says A/B test running and names your design, for example:

Design A is leading by +14.20% (73% chance to win)

Before there is enough data, it says "Data collection in progress." instead.

In the variants table, the control is the baseline, so its Uplift cell shows a dash. Your design's row shows its uplift against the control.

A +18% uplift with the Order goal reads literally: visitors who could see the popup placed orders 18% more often than the ones who couldn't. That is the incremental effect of the campaign, with no attribution modelling in between.

End the test

You decide when to end a test against the control. OptiMonk's automatic check doesn't end it for you: it never stops or declares the control, and it only acts when there are at least two designs.

If your design wins, click the button in the banner (Declare winner at {threshold}%, or Declare now). Your design keeps running for everyone, and the control is switched off. If the design hasn't reached your threshold yet, OptiMonk asks first:

Declare winner now? "Design A" is leading with a 71% chance to win — below your 90% threshold. Ending the test now may pick a winner that isn't statistically reliable yet.
Keep testing / Declare anyway

If the control wins, that is a real answer too. It means this campaign is not paying for itself in its current form. The control itself can't be declared the winner. Pause the campaign, change the offer or the design, and test again.

Remove the control

Open the ⋮ menu at the end of the control's row and choose Delete. The campaign keeps running with the remaining variants. You can add the control back later from the same Add A/B test variant menu.