This is real-life evidence into why you shouldn’t be making multiple major changes to budgets and/or bidding strategies across your automated Google Ads campaigns.

It’s easy to get into the habit of tweaking your Google Ads campaigns, particularly if you were in the industry where all that was available was a manual CPC bidding strategy. Now, almost all campaigns are automated (or at least they should be) and there are fewer things to do manually whilst pottering around in the account. When results are moving in either a positive or negative direction, advertisers typically turn to these actions:

Increasing budget on campaigns that are performing well
Decreasing Target CPA (or equivalent) on campaigns that have an unsustainable CPA for the business
Reducing budget on non Performance Max campaigns which are not performing well
Increasing Target CPA (or equivalent) on Performance Max campaigns to try and get more volume

Now, I’m here to tell you these actions have consequences, which you need to be prepared for.

Performance Max metrics after multiple changes

Here is a case study showing the decline in the results of a Performance Max campaign in less than three months. This Performance Max campaign in question had been changed numerous times, in a bid to reach budget and volume targets, and to attempt to mitigate the continuing decline in results.

Over the three month period, the daily budget was changed 13 times and the target ROAS was changed 15 times. The below graph shows the impact these changes had on conversion rate, and therefore CPA:

*Daily Performance Max CPA increased from £7 to £122 in less than three months*

What really happened with this campaign performance?

As this was a Performance Max campaign, the changes made directly to this campaign were way too frequent and large, which was dangerous. The campaign hadn’t been given enough time to understand the targets before being given a new one. It got confused and pressurised, and just tried to be a good boy by sending unqualified traffic, resulting in a drop in conversion rate.

The changes to other automated campaigns were also dangerous, particularly those already limited by budget. For example, the reduction in non-brand campaign budget and the increase in Performance Max target ROAS at the same time meant the Performance Max campaign didn’t just pick up the impressions, but was also able to experiment broadly with those impressions, leading to an influx of poorer traffic, and reduction in conversion rate.

So, what should have been the solution?

Performance Max should act as your ‘catch all’ approach to search, and a way for you to increase conversion volume and top of funnel traffic when other search campaigns have hit a limit.
Changes made to the Performance Max campaign should be limited in both frequency, and increments.
Decreasing volume on poor performing campaigns should be done by optimising from the ground up (i.e. keywords), rather than just cutting budget.

Why do budget and bidding changes impact Google Ads automated campaign performance?

By making large changes (i.e. change in budget or target by >10%) this causes volatility in the algorithm and tells Google to try a new way of generating results based on a new target. Although the history of the performance will be there to fall back on, increasing budget will likely increase the audience reach beyond users it already know could convert at the rate needed to achieve your goals.

How long is the Google Ads learning phase?

This is heavily disputed. In my opinion this should depend on the seasonality of your audience - are you an eCommerce business with a particular swing at the end of each month coinciding with payday? Are you B2B and have a weekly swing on any days? The logic should really be to give the algorithm at least a whole run through that ‘season’, and in my opinion should always be set at least two weeks, but I’d advise after a large change, to give the algorithm a whole month if the higher ups aren’t on your back.

How much should I tweak my automated budgets and bids?

To start off with, you need to work out as a business what results would be sustainable for you in order to reach overall targets. This will define your target CPA/ROAS, etc. You can then work out the daily budget, which should be enough for the algorithm to try and generate at least 5+ conversions a day. This might be a large change, but that’s OK, because chances are if that target is sensible and you wait patiently, results will come.

Beyond that initial big change, to ensure the algorithm doesn’t get knocked too much, I’d recommend only tweaking the budgets and targets by 5-10%. This figure is also heavily disputed, but in my experience, if you can, then the slower the better.

How Bid & Budget Changes Impact Performance Max