AdWords Campaign Experiments are a bit of an AdWords secret. Launching over 4 years ago they never seemed to have been widely adopted by the industry and to some extent that’s a bit of a shame. However, it’s also a great way to get one up on the competition!
[/text_output][image type=”none” float=”none” info=”none” info_place=”top” info_trigger=”hover” src=”4915″][text_output]Why use them?
This feature lets you split test various parts of a campaign, from keywords and match type to bids and creatives. The real benefit of these is that the test runs as a split test so any weird market force changes that affect your control and your experiment meaning the learnings are a bit more solid.
Where do I start?
Start with a question and work from there. “What would happen if I increased by bids my 10%?”, “What would happen if I changed my ads from proper case to lower case?”.
What can I test?
- Match types
- Landing pages
What can’t I test?
- Any campaign settings (geographic bids, daily budgets, ad scheduling)
- One campaign versus another – it all has to be held within the campaign
[/text_output][image type=”none” float=”none” info=”none” info_place=”top” info_trigger=”hover” src=”4912″][text_output]We’ve recently started an experiment after we noticed that our conversion rate was negatively affected when average position increased. By using AdWords Campaign Experiments we were able to trial a small chunk of traffic at a higher position with a higher bid and watch what happens to the conversion rate.If our hypothesis was wrong we only affected a small chunk of traffic. If we were right then we can expand the test further.
- If you want to try bidding differently there is an option to just apply a bid modifier like you do with mobile devices. However, if you use 3rd party tracking it’s better to clone the ad groups and run the test between the ad groups. That way the 3rd party tracking system sees two keywords and can track the performance.
- Keep a very close eye on performance – use Hour Parting on the Dimensions tab – especially early doors. Although you might think your new creative is the best thing since sliced bread it might collapse.
- If you do have to clone ad groups (maybe you want to do a landing page split test) make sure you clean up afterwards. If the experiment worked better change all the control ad groups over and delete the test ones. Clean as you go!
- Find a statistical significance calculator you are happy with and use it. That way the test doesn’t have to run for 30 days or two weeks; it can run just as long as it needs and then you can plan the next test.
- Start small. Pick one keyword or ad and change it. Monitor that change, record it and work out whether it’s having a positive impact. If it is think about another test. Small iterations are better than trying to shift an entire campaign over to a new landing page!
Happy testing and if you’ve any questions give us a call.
Martin[/text_output][/vc_column][vc_column width=”1/4″][visibility type=”hidden-desktop”][gap size=”20px”][/visibility][vc_widget_sidebar sidebar_id=”ups-sidebar-2″][/vc_column][/vc_row]