Google Ads: changing budget and targets doesn't restart learning

For years you've avoided touching the budget and target of your Google campaigns for fear of resetting the algorithm. A Google product manager has just said that fear was unfounded, and it changes how you can optimize your campaigns starting today.

Does this change really affect you, or only large accounts?

This applies to anyone running Google Ads campaigns with Smart Bidding on, whatever the budget size. If you've ever put off a budget or target adjustment for fear of "breaking" the campaign, this news changes your day-to-day habits.

On October 15, 2026, on the official podcast Ads Decoded, a Google product manager stated openly that changing the target (CPA or ROAS) or the budget of a running campaign doesn't start a new learning period. The system scales bids in direct proportion to the change, using the historical data it has already collected.

The news matters for any business investing in Google Ads: e-commerce stores, professional firms, manufacturers, retail chains. Fear of an "algorithm reset" has often blocked useful optimizations, leaving poorly calibrated budgets running for weeks.

What exactly did Google say, and what's different from before?

The Google product manager made clear that there is "a lot of misinformation" on the topic: the system uses historical data from auctions already processed to recalculate bids proportionally, without resetting anything.

The statement comes from the Google Ads Decoded podcast, published on October 15, 2026. The product manager explained that the system uses so-called landscapes, simulations of bids and budget built on that campaign's auction history, to scale bids directly and in proportion to the requested change.

In plain terms: if you raise your target ROAS by 20%, the system doesn't start learning again. It uses what it already knows and recalculates bids accordingly. No reset, no weeks of unstable performance.

The key distinction is between already stabilized campaigns and new ones. For new campaigns, the learning period still exists, because there's no historical data to base the recalculation on.

How does the Google Ads learning period work, and how long does it really last?

Google has updated the very definition of the learning period: no longer "up to 3 weeks" but "about 50 conversions or 3 conversion cycles," a measure based on real results instead of the calendar.

In September 2026, Search Engine Land reported that Google had rewritten its official Smart Bidding guidelines, Smart Bidding being its automated bidding system. The old reference was "3 weeks or 1-2 conversion cycles." The new one is about 50 conversions or 3 conversion cycles.

This means a high-order-volume e-commerce store can exit the learning period in a few days. A B2B company with few leads per month can take much longer. Duration depends on real volume, not a fixed timer.

For Performance Max campaigns, the format that automatically combines all Google channels, industry operating guidelines point to 4-6 weeks to formally exit learning and 6-8 weeks to reach stable performance.

What changes in practice if you manage paid campaigns every day?

You can adjust budgets and targets more often and more quickly, without waiting weeks to see if the campaign "holds up." That speeds up the optimization cycle and cuts the time your budget spends working poorly.

Before this clarification, many ads managers waited weeks to touch an underperforming campaign, afraid of triggering a reset. Now the logic changes: if your CPA is too high against your goal, you can lower the target right away without losing the history you've built up.

The same goes for budget. If a campaign is converting well and you want to give it more fuel, you can raise the daily budget without waiting for the end of the month or a "safe" moment.

One caution, though: Google's statement doesn't specify how big a change can be before the system stops scaling proportionally. Very abrupt changes (e.g. doubling the budget in a single day) could still cause temporary instability, even if it isn't technically a "reset."

Three concrete scenarios: what changes for an e-commerce, a professional firm, and a manufacturing company?

The impact depends on conversion volume and how often campaigns need adjusting. High-volume accounts gain agility; low-volume accounts gain, above all, certainty.

Illustrative examples: numbers are meant to show the reasoning, they are not real customer data.

What are Promotion Mode and the other new tools announced alongside it?

Google has announced a series of new controls that complete the no-reset optimization logic: Promotion Mode lets you temporarily raise budget and targets during sales or launches without restructuring the campaign.

The October 15 podcast also previewed other upcoming tools. The most useful for anyone running seasonal promotions is Promotion Mode for Performance Max: it lets you add extra budget and temporarily change the ROAS target during a promotional period (sales, flash sales, product launches), without restructuring the campaign. Search Engine Land described it as one of the most anticipated updates of 2026.

This sits alongside the new Channel Prioritization and Budget Allocation controls announced on October 14, 2026 by the Google Ads & Commerce Blog: they let you specify which channels (Search, Shopping, Display, etc.) get priority budget and cap spend on secondary channels at a percentage of the daily budget. We covered them in detail in our article on Google Ads: channel priority and budget allocation.

Are there risks or limits to keep an eye on?

Google's statement has clear boundaries: it applies to stabilized campaigns, doesn't specify the maximum size of a change and doesn't cover structural changes. Some details from the podcast still contradict the written support pages.

The first limit is the size of the change. Google hasn't said how far the system scales proportionally. Doubling the budget in one day is different from raising it by 15%. Caution suggests gradual changes, even though technically there's no reset.

The second limit concerns discrepancies between the podcast and the written documentation. On some points, such as Demand Gen support for Budget Allocation and the exact end date of Display Expansion on Search (February or March 2027), the podcast and Google's official pages don't match. Until there's an official clarification, it's worth checking the settings directly in the interface.

The third point: the new Channel Prioritization and Budget Allocation controls are not yet supported by the Google Ads API or by Editor (the desktop tool for managing campaigns in bulk). If you run large accounts with automations, you'll need to wait for updates.

If you run low-volume campaigns, typical of Italian B2B, the learning period remains a real issue for new campaigns. This update doesn't eliminate the problem; it limits it to budget and target changes on campaigns that are already stabilized. It's also worth reading our analysis of how the Facebook algorithm is changing for campaigns, where the learning logic works differently.

Why is everyone talking about it now: is the paid market really changing?

In 2026, Google sped up its shift toward increasingly automated campaigns, where traditional manual controls are replaced by goal parameters. Knowing what you can touch and what you can't has become more important than ever.

The learning period clarification comes at a time when Google is redesigning its entire campaign control logic. The new tools announced in October 2026 (Channel Prioritization, Budget Allocation, Promotion Mode) all point the same way: giving advertisers more control levers without restructuring campaigns.

At the same time, Google has redefined the learning period itself: no longer a fixed timer of weeks, but a threshold based on actual conversions. As Delante reports in its October 2026 SEM roundup, the move from "3 weeks" to "50 conversions" changes how you plan the launch of new campaigns, especially if you have low volumes.

If you run paid campaigns in Italy, where average budgets are often tighter than in Anglo-Saxon markets, understanding these mechanisms matters. Every week of unstable performance has a real cost. Being able to optimize more often, without fear of a reset, is a concrete competitive advantage. If you want to dig deeper into budget management for seasonal peaks, the article on how to redistribute your paid media budget for Black Friday offers a complementary analysis.

Where to start: concrete steps to optimize your campaigns now

The first step is to identify campaigns that have already stabilized, meaning those with at least 50 accumulated conversions, and check whether budget or targets still align with your business goals.

There's no need to wait: you can start reviewing your active campaigns today with a more agile approach. The only precaution is to tell campaigns with enough history apart from those still in the learning phase.

What are the most common mistakes made at this stage?

The most common mistake is the opposite of the old one: now that we know there's no reset, it's tempting to change things too often and too fast, making it hard to tell what actually worked.

Google's clarification removes a brake, but it doesn't remove the need for method. Here are the most frequent mistakes we're already seeing in this transition phase.

How can you tell if your optimizations are really working?

The metrics to track are the same as ever, but now you can check them more often and react faster without worrying about compromising the campaign.

The measurement logic doesn't change. What changes is how often you can step in and how confidently you can read the signals. Keep an eye on these metrics in the days after each change.

If you run e-commerce campaigns, it's also worth reading our analysis of Google Universal Cart and how direct purchasing from Search is changing: the environment your campaigns run in is shifting on the shopping experience side too.

So, is it worth changing your approach for your business?

It's worth it if you have campaigns that have been running for at least 30-60 days with enough conversion history (close to 50), and you've put off optimizations for fear of a reset. Now you can act more often and more promptly.

A middle ground: you have campaigns with low or mixed volumes, some stabilized and others still learning. It makes sense to adopt the new logic on mature campaigns and stay cautious with new ones.

Not yet if you're about to launch new campaigns or have just changed your bidding strategy. In those cases the learning period still applies, and operational caution is still the right call.

How we see it at SHM

For us, this clarification from Google is an example of how technology should work: remove pointless operational constraints and give time and agility back to the people running campaigns. For years, the fear of an algorithmic reset slowed down decisions that could have improved results within weeks. Now that fear has a clear name, misinformation, and we can stop working around it.

We believe AI and automation should be integrated to free teams from routine, not to create new gray areas. Understanding exactly what the system does and doesn't touch is the first step to using it well, not being at its mercy.

In the paid campaign management projects we run, the optimization cycle is one of the most delicate points: too slow and you waste budget, too aggressive and you risk instability. This news lets us work with more precision on both fronts.

If you want to understand how to apply these principles to your campaigns, our team is available for a concrete analysis. You can explore our Google Ads campaign management services, our work on strategic AI integration in marketing processes, and our training on business process automation. You can also browse all the articles in the Marketing and ADV performance cluster to stay up to date on the news that really matters.

Frequently Asked Questions

The real news isn't that you can touch the budget: it's that for years you've optimized less than you should have, for fear of a reset that never existed.