What are Google Ads experiments and how do you run them?

Google Ads experiments (previously called drafts and experiments) let you test changes to your campaigns before committing your budget. Instead of making a change and hoping it works, you split your campaign traffic between the current version (the control) and a draft with your proposed change (the experiment). When the experiment ends, Google shows you the results with a confidence interval, and you decide whether to apply the change.

Most small business campaigns change too often, too fast. Every adjustment shifts the data and makes it hard to tell what actually improved performance. Experiments solve that by isolating one variable at a time so you can see what really drives results.

How experiments work

An experiment starts with a draft. You take an existing campaign and duplicate it as a draft, then make the changes you want to test. It could be a different bid strategy, new ad copy, a different audience, or a changed landing page. The draft stays in draft mode and does not serve ads until you turn it into an experiment.

When you create the experiment, you choose what share of the campaign's traffic goes to the experiment versus the control. A 50/50 split is the most common because it gives the cleanest statistical comparison. Google Ads randomly assigns each auction to one side or the other, so the only difference between the two groups is the change you made. That is the key principle of any good test: change one variable at a time.

What you can test with experiments

Ad copy and creative. Test two headlines against each other, compare different descriptions, or try a new image to see which gets the better click-through rate.

Bid strategies. Compare manual cost-per-click against automated Smart Bidding, or test Target CPA against Maximise Conversions. The guide on choosing a Google Ads bidding strategy covers the options in detail.

Audiences. Test a new audience segment against your usual targeting. For example, compare an in-market audience against your standard custom audience to see which group converts at a lower cost per lead.

Landing pages. Send experiment traffic to a different landing page to measure conversion rate. The landing page optimisation guide has more detail on what makes a page perform.

When to run an experiment

Run experiments when you need confidence that a change will actually improve results. If you have enough traffic, even a small improvement in conversion rate can be statistically significant. Google Ads reports the confidence level in the experiment results so you can see when a finding is reliable.

Do not use experiments for urgent operational fixes. If a keyword is wasting spend or a landing page link is broken, fix those directly and save experiments for strategic decisions you need data on.

What happens after the experiment ends

Google Ads summarises the results with the percentage difference between the control and the experiment, the confidence interval and a recommendation. If the experiment outperforms the control with statistical significance, you can apply the draft to the original campaign in one click. Everything carries over: the draft settings, budget and creative replace the control version. If the experiment did not win, you keep the original and plan the next test.

The data from experiments compounds. Each test tells you something about what your audience responds to, and the learning applies to the next campaign. That is the difference between a campaign that improves over time and one that flatlines.

Victory Digital builds experiments into every paid search campaign as standard. The Home Instead campaign, for example, delivered a verified +3,446% return on ad spend through structured testing. Digital marketing packages start from GBP 849 per month and include ongoing campaign testing and management.

Ready to improve your Google Ads performance with structured testing? Book a discovery call at https://victory.digital/book/discovery-call.