Smart Bidding learning period: which changes trigger it?

Short answer: Google’s rewritten help article says Smart Bidding Learning status never lasts more than 7 days. Three things trigger it: a new or reactivated bid strategy, adding or removing campaigns, ad groups or keywords from a strategy, and a significant change to conversion goals. Edits to budgets, targets, creatives and targeting do not trigger it.

What did Google change about Smart Bidding learning?

In October 2026 Google rewrote its help article on this topic and renamed it About Smart Bidding learning. On the Learning status, it now says: “It never lasts more than 7 days.”

Smart Bidding is Google’s automated bidding: Maximize conversions, Maximize conversion value, target CPA and target ROAS. CPA is cost per acquisition, meaning ad spend divided by conversions. ROAS is return on ad spend, meaning conversion value divided by ad spend.

For daily work, the most useful part is the clear list of what triggers Learning status and what does not. Many account owners I talk to treat every edit as dangerous because they were told it “resets learning”. The new wording draws a much sharper line.

Which changes trigger Learning status and which do not?

According to Google, three things put a bid strategy into Learning status:

• A new or reactivated bid strategy.
• Campaigns, ad groups or keywords being added to or removed from a strategy.
• A significant change to the conversion goals the strategy optimizes for.

Changes to budgets, targets, creatives or targeting do not trigger it. Here is how that applies to the edits you actually make:

Change you make	Triggers Learning status?	What to watch afterwards
Raise or lower the daily budget	No	Spend pacing, and whether CPA holds as volume grows
Change target CPA or target ROAS	No	Volume and efficiency over the next 1-2 conversion cycles
New ads, headlines or images	No	Conversion rate of the new creatives
Location or audience targeting	No	Whether the new traffic converts like the old traffic
Switch to a new bid strategy or reactivate one	Yes	Results only after Learning ends and conversions have caught up
Add or remove campaigns, ad groups or keywords from a strategy	Yes	Batch these into one planned change
Significant change to the primary conversion goal	Yes	Whether conversion counts and values still make sense

Does “no Learning status” mean the change has no effect?

No. Learning status is a label in the interface. It does not promise that results stay flat. Say you lower a target ROAS from 600% to 400% (an illustrative figure). The strategy will bid harder, and you should expect more volume at lower efficiency. That is what you asked for, not a learning problem.

Google says performance typically converges to a new goal within 1 to 2 conversion cycles, and that you should account for conversion delay. A conversion cycle is the usual time between the ad click and the conversion.

How to estimate your own conversion cycle

1. In Google Ads, segment your campaign report by days to conversion (Segment, then Conversions).
2. Find the number of days within which most of your conversions arrive.
3. Double it. That is your minimum evaluation window after a target or goal change.

Illustrative example: say most purchases in your e-shop arrive within 3 days of the click. Then judge a target change after about a week, not after one bad Tuesday. For a B2B company that imports qualified leads from its CRM weeks after the form fill, the window can be a month or more.

How to change a Smart Bidding campaign without hurting it, step by step

1. Log the change with the date and the reason. Google Ads keeps a change history, but it does not tell you why you made a change or what you expected to happen.
2. Sort the edit into one of two groups. Budgets, targets, creatives and targeting go on one side. Bid strategy, structure and conversion goals go on the other.
3. Move budgets and targets in moderate steps. They do not trigger Learning status, but the strategy still reacts. Smaller steps let you see which change caused which result.
4. Batch the structural changes. If a restructure means new keywords, moving campaigns into a shared strategy and a new goal, do it all in one go. That way the strategy learns once rather than three weeks in a row.
5. Fix the conversion goal before anything else. Google recommends one consistent primary conversion goal. If yours is wrong, every other optimisation rests on a bad signal. My guide to conversion value setup for profit covers how to get this right.
6. Wait out Learning and at least two conversion cycles. Then compare matching periods, for example the same days of the week.
7. Remember that recent days always look worse. Google Ads assigns conversions to the click date by default, so the last few days are still filling in.

What I see in the accounts I run

Across the 500+ campaigns I have launched, the edits that hurt results are rarely budget changes. They are casual changes to conversion goals. Someone sets “add to cart” or a newsletter signup as a primary conversion “just for a week”. Or a new cookie banner changes what gets tracked, and the strategy starts optimising for something different.

A European example: switching conversion values from prices with VAT to prices without VAT. It feels like an accounting detail, but the strategy sees every value drop at once. I treat it as a significant goal change and plan it outside the busiest weeks.

The opposite mistake is just as common in smaller EU e-shops: owners freeze the account for weeks out of fear of resetting learning. Meanwhile, a campaign that converts well is held back by its budget, or a target set in spring no longer fits Q4 margins. In national markets like Slovakia or Czechia, where conversion volumes are small, waiting too long often costs more than a well-timed edit.

My rule: I adjust budgets and targets whenever the business needs it and note each change in the log. Bid strategy switches, structural moves and goal changes I plan, batch and keep away from peak selling weeks.

Most common mistakes with Smart Bidding learning

• Treating Learning status as the only risk. A big target change does not trigger Learning status, but it can still shift volume and CPA for 1-2 conversion cycles.
• Judging results after 2-3 days. Conversion delay makes recent days look weak, so many good changes get rolled back too early.
• Spreading one restructure over several weeks. Each step that triggers Learning status starts the clock again.
• Combining a planned restructure with a test you want to measure. If you are also changing how Performance Max spends across channels (see Google Ads channel prioritization), run that in a separate window so you can tell which change did what.
• Splitting a small budget into many campaigns. Google recommends consolidating campaigns and ad groups and keeping daily budgets sufficient. Thin data across many campaigns gives each strategy less to learn from.
• Blaming bidding for a website problem. If conversion rate dropped because of stock-outs, a slow checkout or a price increase, no bid strategy change will fix it.

When not to make changes that trigger Learning status: in the week before a major sale such as Black Friday. Budget increases are fine then, but a strategy switch or a new conversion goal is not.

Start by checking your primary conversion goal, then work out your conversion cycle so you know how long to wait after each edit. After that, budget and target changes become routine adjustments rather than risks. If you want a second pair of eyes on your bidding setup, see how I work as a PPC specialist including a free advertising audit. Email me and I’ll reply within 24 hours.