-
Thoughts on google ads experiments?!
We have been running a standard shopping campaign with enhanced cpc for a while, and we have decided to give one of the ROAS experiments a go – however, what we have found (only a few hours in mind!), is that the ROAS campaign appears to have cannibalised the performance of the standard campaign off which the experiment is based – so obviously it is going to do better in performance. The standard campaigns performance has suddenly ground to a halt with minimal impressions and clicks whereas the ROAS experiment is taking all the clicks/impressions.
How can this be considered a true and proper test to judge performance because obviously we are going to see better results on the ROAS campaign if the standard one just dries up (gets cannibalised!).
Am I missing something? Are these experiments worthwhile? May be im judging too soon but it just seems odd how the standard campaign has suddenly dropped in performance so much.
-
1 Reply
-
I would imagine that your tROAS is effectively a higher bid than your manual or ECPC bids. You could increase your tROAS or raise your manual bids to make the baseline more competitive. You might also consider running Max Conv. Value without a target to have a more “fair” experiment.
-
Hi, have you split the budget 50/50 for the experiment? also what kind of ROAS target have you set and is it around the recommended are past ROAS you have been achieving?
I did exactly what you are doing a month or two ago ( except my shopping campaign was using manual CPC not eCPC ) and although the TROAS experiment didn’t cannibalise the standard shopping campaign it just wasn’t profitable in terms of ROAS after a few weeks.
I then thought that maybe i just wasn’t giving the TROAS experiment enough budget so recently tried switching the original campaign to TROAS bidding, but after a fantastic start it just went downhill after that and was costing me more and more money each day with less sales.
Maybe it was the campaign structure, maybe the ROAS target, the lack of conversion data or i just didn’t leave it long enough to optimise, but I couldn’t afford to keep spending more with no/few sales, so really you might need a spare couple of £k to run the test ( depending on the size of your account )
-
You are only a few hours in, you can not judge anything and ask if this is a proper test. You need to run your test longer.
Log in to reply.