By the Mayab Media Team
A lot of advertisers are losing money because they're scared of a status label. They leave a winning campaign capped at a tiny budget for weeks because "raising it will reset learning." Google just said, in writing, that it won't.
If you've ever held off on scaling a campaign that was clearly working, this one's for you.
What Google actually said
Two things happened in October.
First, on October 15, Google's Ads Decoded podcast featured Google Ads product manager Tal Akabas. As PPC Land reported, he said changing the target or budget on an established, steady-state campaign does not trigger a learning period. His explanation: the system scales bids in line with the target change using auction simulation data it already has. He also said there's a lot of misinformation about this topic. We agree.
Second, on October 21, Search Engine Roundtable spotted that Google had rewritten its help document, now called "About Smart Bidding learning." The key points:
- The Learning status never lasts more than 7 days.
- Only three things trigger Learning: a new or reactivated bid strategy, a composition change (adding or removing campaigns, ad groups or keywords from the strategy), or a significant change to conversion goals.
- Changes to budgets, targets, creatives, or targeting (like location, audience or language) don't wipe historical data or trigger Learning.
- Google's best practices: one consistent primary conversion goal, budgets big enough to capture conversions, consolidated structures, and no stacking of multiple target changes inside one conversion cycle without a clear reason.
One caveat matters here. Akabas didn't say how big a target change can be before proportional scaling stops working. And the doc separates the visible Learning label from the time performance actually needs to settle. So "no Learning status" doesn't mean "no wobble."
Why this matters in money terms
Fear of learning has cost accounts real money in two ways.
Missed volume. You've got a campaign hitting a $40 CPA against a $60 goal, and it's "Limited by budget." Every day you don't raise that budget, you leave cheap conversions on the table for competitors. If the only thing stopping you was a reset, that excuse is gone.
Bad structure. People also do the opposite. They avoid adding a new campaign to a portfolio strategy, or swap conversion goals "just this once," without realizing those are the moves that do trigger learning. That's where performance actually resets.
So the real lesson is about knowing which levers are cheap to pull and which are expensive.
The cheap levers vs. expensive levers framework
Here's how we sort changes for our clients now.
Cheap levers (pull them when the data says so): - Daily budget increases or decreases - Target CPA or target ROAS adjustments - New ads, assets and creative - Location, audience and language targeting
Expensive levers (plan them, batch them, protect them): - Switching or creating a bid strategy - Adding or removing campaigns, ad groups or keywords from a portfolio strategy - Changing which conversion actions count as primary
What to do this week
- Pull every campaign marked "Limited by budget." Sort by CPA or ROAS against goal. Any campaign beating its target with room to grow is a candidate for a budget increase today. Google's doc says budget increases take effect immediately.
- Raise in steps you can measure. Learning won't reset, but you still need to read results. We like increases you can judge over one full conversion cycle before the next move. If your sales cycle is 10 days, don't judge a change on day 2.
- Make one target change per conversion cycle. This is Google's own advice and it's good. Moving tCPA from $50 to $45 on Monday and $40 on Wednesday makes it impossible to tell what worked.
- Batch your expensive changes. If you need to restructure a portfolio or change conversion goals, do it all in one go, on a quiet week, not the week before a big sales push.
- Lock down your primary conversion goal. If you've got five "primary" actions including page views and button clicks, fix that. A clean, consistent goal does more for Smart Bidding than any bid tweak.
- Stop reading the Learning label as a verdict. It's capped at 7 days. If a campaign is still underperforming after that, the problem is targeting, creative, offer or tracking, not "learning."
Our honest opinion
We're glad Google finally put this in the documentation. For years, "don't touch it, it's learning" was the most expensive sentence in PPC. It made people passive with budgets while they were happy to rip apart structure, which is exactly backwards.
That said, don't swing to the other extreme. Daily target fiddling still makes your data noisy, even if the label never shows up. The winning approach is boring: clean conversion data, consolidated campaigns, and confident budget moves when the numbers back you up.
Takeaway: Budgets and targets are cheap levers now, so use them, and save your caution for bid strategy, structure and conversion goal changes.
Want us to check how this affects your account? Book a free audit.
