Google Says Changing Your Target CPA Or Budget Does Not Reset Learning

Google Ads confirms changing target CPA or budget does not reset the Smart Bidding learning period

A Google Ads product manager said on Google's own podcast that changing your target CPA or budget on an established campaign does not reset the learning period. If you have spent years refusing to touch a target in November for fear of resetting learning, that fear is costing you money, not the edit. Here is what was said, and the other takeaways worth knowing before Black Friday.

Where This Came From

This was not a blog post from Google. It came out of an episode of the Ads Decoded podcast, where Google Ads Product Liaison Ginny Marvin interviewed Product Manager Tal Akabas for about forty minutes. Frederick Vallaeys, who co-founded Optmyzr, wrote it up for Search Engine Journal and added his own analysis on top.

That matters for how much weight to put on it. A product manager explaining system behavior on a company podcast is good information, but it is not documentation you can point a client to. Nobody published a verbatim transcript of the key line, so what follows is a paraphrase.

Changing Targets And Budgets Does Not Reset Learning

Akabas said changing a target or budget on an existing campaign already running in a steady state does not erase your historical data or put the campaign back into Learning status.

The reason he gave is that the system already has bid landscape data for your campaign. It can scale bids in proportion to the new target without starting over, because it already understands the auction you compete in.

Learning status still gets triggered by real structural changes. A new or reactivated bid strategy does it, so does changing what the campaign is made of, and so does a significant change to your conversion goals. A genuinely new campaign still needs ramp-up time.

One widely repeated rule did not survive. The idea that you must keep daily budget changes under twenty percent is described as a common myth, not a Google threshold. I have heard that number quoted in client meetings for years as though it came off a help page. It did not.

Keep in mind, none of this means performance will hold steady when you change a target. It means your history is not wiped. Those are two different claims, and people conflate them. Repeated target changes inside one conversion cycle are still discouraged without a business reason.

Skip Seasonality Adjustments For Black Friday

This is the takeaway I expect the most pushback on, and it is the one I agree with most.

The guidance is to save seasonality adjustments for unforeseen events and skip them for routine holiday spikes. Black Friday is not unforeseen. Smart Bidding has seen your last several Black Fridays.

Optmyzr looked at Black Friday and Cyber Monday across 2022 to 2024 and found accounts using seasonality adjustments saw CPCs rise about twice as much, with ROAS falling each year, though revenue grew faster. That is observational, not proof of cause, which Vallaeys says plainly. Treat it as a reason to question the habit.

We broke down how this plays out across a year in our guide to how seasonality impacts Google Ads ROI.

Three More Things From The Episode

  • Use Promotion Mode for scheduled promotions where it is available. Smart Bidding Exploration is separate and needs eligible target ROAS campaigns with budgets that are not constraining spend. Google's FAQs put the effective ROAS flexibility at around ten percent.
  • Channel Prioritization only does anything on days your budget actually limits your spend. On every other day it is inert. And Budget Allocation percentages apply to your budget setting, not to what you really spent, which is a distinction that will bite you if you read the report quickly.
  • Display Expansion allocation behavior changes in February 2027. Setting Display to Low Priority at fifty percent preserves how things work today. Setting a Low Priority channel to zero percent means no spend at all on that channel. This is the same migration we covered when Google added channel priority controls.

The sixth takeaway covers aligning demand-led budgeting with your finance team. Useful at scale, less so on a single account.

What This Changes For You

If you have been sitting on an obviously wrong target CPA because you did not want to reset learning in Q4, change it. That was the point of the episode.

What I would not do is take this as permission to fiddle. The reason to leave a target alone is that constant changes inside one conversion cycle give the system a moving goalpost. The reason is not that you erase your data, because apparently you do not. For the groundwork, start with Smart Bidding and the full Google Ads guide.

Question to Answer:

Is there a target CPA or target ROAS in your account right now that you know is wrong and have left alone only because you were worried about resetting learning?

In Summary

A Google product manager said that changing targets and budgets on established campaigns does not reset learning or erase history, and that the twenty percent budget change rule is a myth. Performance can still move, so this is permission to fix a wrong target, not permission to keep adjusting one.

Before Black Friday, go fix the targets you have been avoiding and leave the seasonality adjustments alone unless something genuinely unexpected happens.

Read the source

1 comment

So, Google just keeps this fact to itself for years while advertisers act, or more pertinent, don’t act because they’re under the false (unnecessary) belief that CPA needs to “relearn” which seemed stupid on it’s face, given the exact data retention reasons stated in the podcast. Meanwhile, hundreds of agencies have undeliberately misled customers and, in effect, mismanaged thousands of accounts. Google can too often be the ultimate “fly in your soup” experience.

Sam •

Leave a comment

Please note, comments need to be approved before they are published.