How a fair test works
- Pick one changeA headline, offer, page or bid strategy.
- Split trafficSend half to each version at once.
- WaitCollect enough conversions on both sides.
- ApplyKeep the winner and plan the next test.
Tools inside Google Ads
Google's custom experiments copy a campaign, change what you choose and split traffic between them. Google recommends a 50% split for the clearest comparison. A cookie-based split shows each person only one version. When the test ends, you can apply the changes to the original campaign.
What makes a test worth running
- It changes something big enough to matter.
- It changes only one thing at a time.
- Both versions run during the same days.
- It runs until each side has enough conversions.
Common testing mistakes
Ending a test after a few days because one side looks ahead. Judging by click-through rate when leads are the goal. Testing on a campaign with only a handful of conversions a month, where results never become clear.
Good example. A law firm tests a page with a short form against its long form for six weeks. The short form brings 30% more consultations.
Bad example. A firm changes the headline, offer and landing page at once. Leads rise, but nobody knows which change worked.
Not sure where to start? See what to test first. Our PPC management runs at least one test in every account each month.