Google Ads Smart Bidding: How Long to Wait Before Changing Anything

After you launch or change a Google Ads Smart Bidding strategy, wait at least one to two weeks, or until it has recorded enough conversions to judge, before making any major change. Google describes an initial learning period during which the system adjusts to a new strategy, and changes made during that time can restart it. For a small account, patience is usually cheaper than fiddling.

Below is how to decide how long to wait, which changes restart learning, which are safe, and when waiting becomes the wrong decision.

What is the Smart Bidding learning period?

Smart Bidding strategies such as Maximize Conversions, Maximize Conversion Value, Target CPA and Target ROAS use machine learning to set bids for each auction. When you create a strategy, switch to one or make a significant change, the system has to learn how your conversions relate to signals such as device, location, time of day and audience. Google’s help documentation says a strategy may show a “learning” status while it calibrates, and that this commonly takes around seven days, or a few conversion cycles, whichever is longer. Treat that as a guide, not a guarantee, and check the current wording in Google’s help pages before relying on any specific number.

The key point for small advertisers is that learning depends on conversions, not days. An account that gets two conversions a week learns far more slowly than one that gets 40.

How long should you wait before changing anything?

Use two thresholds together: time and data.

These bands are our working rules from managing accounts, not Google figures. The logic is that you need enough conversions for a pattern to be distinguishable from luck. With five conversions, a “good week” and a “bad week” can be nothing more than chance.

If your account is on the low end, our guide on Target ROAS bidding on a small budget explains why target-based strategies can backfire when conversion data is thin.

Which changes restart learning?

Google lists several changes that can trigger a new learning period. The common ones are:

• Changing the bid strategy type, for example from Maximize Clicks to Maximize Conversions.
• Making large changes to a target CPA or target ROAS value.
• Making large budget changes.
• Changing the conversion actions the strategy optimises for.
• Adding or removing a large share of the campaign’s keywords, ads or audience signals.
• Pausing a campaign for a long time and re-enabling it.

“Large” is the important word. A budget change of 10 to 20% generally gives the system room to adapt, while doubling a budget overnight is more likely to disturb it. When we need to scale, we raise budgets in steps of around 20% every few days instead of in one jump. Again, that is our practice, not a rule published by Google.

Which changes are safe during learning?

You do not have to freeze the account completely. These are generally safe and often necessary:

• Adding negative keywords. This removes waste without changing what the strategy is optimising for. Our weekly search terms routine fits here.
• Fixing disapproved ads or broken landing pages.
• Correcting conversion tracking errors. If tracking was wrong, the system has been learning from bad data, so fix it as fast as possible and accept that learning may restart.
• Pausing individual ads that are clearly poor. Keep at least two good ads active in each ad group.
• Adding a new ad variation while leaving existing ones alone.

What should you look at during the waiting period?

Waiting does not mean doing nothing. Use the time to watch signals that tell you whether the strategy has a chance:

1. Impressions and impression share. If the campaign barely gets impressions, the target is probably too strict or the budget too small.
2. Click-through rate. A very low rate suggests the ads do not match the queries.
3. Search terms. Look for irrelevant queries and add negatives.
4. Conversion tracking. Place a test conversion and confirm it records.
5. Landing page behaviour. Check sessions, bounce and form starts in analytics.

If clicks are arriving but nobody is enquiring, the strategy is probably not the problem. Our seven-day diagnostic for ads that get clicks but no leads helps separate bidding issues from page and offer issues.

When is waiting the wrong decision?

Patience has limits. Intervene sooner if you see any of the following:

• Spend is far above plan with no conversions. If a campaign has spent several times your target cost per lead without a single conversion, something is wrong with targeting, the offer or tracking.
• The strategy is limited by a target that cannot be met. A CPA target of €10 in a market where a click costs €4 and the conversion rate is 5% will starve the campaign.
• Conversions are recorded but are low quality. If the system optimises for newsletter signups while you want sales enquiries, it will do exactly what you asked, not what you meant.
• A clear external event occurs. A product goes out of stock, a price changes or a campaign period ends.

The cost of being wrong is asymmetrical. Changing too early wastes learning, but waiting too long wastes budget. Setting a spend ceiling in advance, such as “if we spend €300 with zero conversions, we review”, turns the decision into a rule instead of a mood.

A simple decision rule

Before touching a Smart Bidding campaign, ask three questions:

1. Has it been at least 7 days since the last significant change?
2. Has the campaign recorded enough conversions to see a pattern?
3. Is the issue one that the bidding strategy can fix?

If the answer to any of them is no, make only the safe changes listed above. Write down the date and what you changed. In a month, you will have a record that tells you what worked, which is more valuable than any single optimisation. For a realistic idea of how long paid campaigns take to mature, our 90-day timeline for paid ads sets expectations by phase.

What if you do not have enough conversions?

Small accounts often lack the volume that Smart Bidding wants. You have three realistic options. First, move to a simpler strategy such as Maximize Clicks with a maximum CPC limit while you collect data. Second, optimise for a higher-funnel conversion that occurs more often, such as a quote request started, but only if it correlates with real business. Third, combine campaigns so that data is not spread over many small ones. Whichever you choose, our guide to the minimum budget to start Google Ads or Meta Ads is a useful check on whether the budget can support the strategy at all.

The takeaway is to treat the learning period as a cost of doing business with automation. Give the system time, keep your safe maintenance going, set rules for when to intervene and write down every change.