-
A/B Experiment with Portfolio Bid Strategies in Google Ads
Hi everyone,
Is it possible to run an A/B experiment for 2 portfolio bid strategies (Max Conv. vs Max Conv Value)? The main reason for using Portofolio bid strategies is that they allow for max cpc bids on smart bidding campaigns. When I tried creating such an experiment Google Ads tells me A/B experiments do not support shared budgets/bid strategies and I was wondering if anyone has found a way around it.
Cheers!
-
1 Reply
-
You’d be asking the bid strategy to manage two different goals against the same inventory, same cut of your program. They would be trying to compete on high value auctions with no way of knowing which strategy to prioritize, while also splitting traffic ‘evenly.’ Just can’t effectively optimize with all those constraints. One of them would suffer.
A/B testing in general on Auction-Time bidding barely works, very hard to keep front end metrics in line with each other (like cost), making performance very difficult to compare.
You gotta just make two unique strategies and deploy against either different Geo’s or different cuts of your program. Campaign duplication with unique geos is probably the best way to get long term answers but probably needs support from someone who can model out matched markets & predicted conversion behavior for each segment to compare against predicted. Just comparing the two against each other can be misleading.
I’d also recommend running without Min / Max CPC thresholds for a while to understand your baseline performance before dropping them in. That said, I do see Min / Max as some semblance of a control lever to try to manage CPC’s, but you’ll be balancing against quality of traffic.
TL;DR – You should Geo-test if possible.
-
There is no workaround. Remove the max bid cap and run your test.
-
It sounds like your problem is that you want two bud strategies in the same budget, which would leave them mia configured.
The question is what is your overall goal? If it’s just as many conversions as possible, then max conversion is the way. If it’s as many of the higher value conversions you’ve established (meaning typically fewer overall conversions) then MCV is the way. Choose the one that makes the most sense for you and stick with it.
-
You can’t use shared budgets as you called out. Your test structure is flawed: comparing apples and oranges.
-
I used to run portfolio bid strategies for the exact same reason you mention here.
And I had also been looking for a work around to set up experiments.
Unfortunately I didn’t find any.
So I don’t think there is a way to combine these two – sorry…
Log in to reply.