Home › Guides › Google Ads learning period

Google Ads learning period: what actually resets it

By Abhishek Thakral · abhi.performs, Dubai · Updated October 2026 · 8 min read

Short answer

The Google Ads learning period starts when you create or reactivate a bid strategy, change one of its settings, or add or remove campaigns, ad groups or keywords. A Google Ads PM said on 15 Oct 2026 that changing the target or budget of a steady campaign does not trigger learning. Structural changes still do.

As of 17 Oct 2026, the Google Ads learning period has an official answer to its most argued-about question. On Google's Ads Decoded podcast, product manager Tal Akabas said that adjusting the target or budget of an established campaign does not send it back into learning. If you have ever been told never to move a budget by more than 20%, this matters.

This guide puts three things side by side: what Google's help pages say, what was said on the podcast, and what is just agency folklore. Then it covers what I do in UAE lead-gen accounts, where long sales cycles make the learning period harder to read.

What Google says triggers the Google Ads learning period

Google's help page on the topic lists the reasons a bid strategy shows the "Learning" status (Duration of the learning period, Google Ads Help):

You can check which one applies by hovering over the status in the bid strategy column. Google's guidance on bid strategy statuses is simple: keep running the account as normal and do not judge performance until the status clears (About bid strategy statuses, Google Ads Help).

Notice what is not on the list. Neither page gives a percentage threshold for budget changes, and neither says the learning period lasts seven days.

What a Google Ads PM said on 15 October

The episode is "How to guide campaign spend with Google Ads budget and bidding controls", hosted by Ginny Marvin, Google's Ads Product Liaison, with Tal Akabas (Ads Decoded, 15 Oct 2026). It is mostly about the new channel prioritization and budget allocation controls Google announced the day before.

According to PPC Land's write-up of the episode, Akabas said that changing the target or budget of an established, steady-state campaign does not trigger a learning period. Google's systems scale bids in proportion to a target change, using the auction data the campaign already has. He said a new campaign with no prior data can have a learning period, and called the topic one with "a lot of misinformation" (PPC Land, 16 Oct 2026).

Two limits are worth holding on to. PPC Land notes that Akabas did not say how large a target change can be before proportional scaling stops working, and that the claim does not cover structural changes. So "small, steady target moves are fine" is supported. "Halve your target CPA overnight and nothing happens" is not.

Is this a contradiction?

Not really. Google's help page lists "setting change" as a learning trigger, and a target is a bid strategy setting. The podcast adds detail: for a campaign with plenty of history, Google adjusts bids from data it already has, so a target change does not need a fresh calibration in the way a new strategy does.

In practice you may still see the "Learning" label after a big edit. Treat the label as a reason to wait before judging results, as Google suggests, not as proof that the campaign has forgotten everything.

How long the learning period lasts

Google does not give a number of days. It says calibration can take up to around 50 conversion events or 3 conversion cycles, and can be faster with more conversion data. The help page names three things that affect duration: how many conversions the campaign gets, how long your conversion cycle is (time from click to conversion), and the bid strategy type. Manual CPC has no learning period at all.

The "7 days" figure you see on many blogs is a rule of thumb. For a high-volume e-commerce account it may be roughly right. For a lead-gen account that optimises to an offline outcome, it can be far off.

Why it is harder to read in UAE lead-gen accounts

Most of the accounts I run sell offline: off-plan property, aesthetic clinics, business setup. In those accounts the conversion I want Google to optimise to is a qualified lead or a booked consultation imported from the CRM, not a form-fill. That is the right choice, but it stretches the conversion cycle.

Do the maths with Google's own guide. If it typically takes ten days from click to a qualified lead in your CRM, three conversion cycles is about a month. A low-volume account may also need weeks to reach 50 conversions. Judging a change after three days in that account tells you very little. Meta works differently: an ad set usually exits after about 50 results in the week after its last significant edit, as I explain in the guide to the Meta ads learning phase.

The expensive mistake I see is not the learning period itself. It is the second and third edits made because someone panicked on day three. In the clinic account I wrote about in my location targeting mistake, one structural change took ROAS from 3.5x to 1.4x, and recovery came from reversing it and then leaving the account alone.

Offline imports also break. A CRM sync fails, a field changes, uploads stop for a week. Google has a separate tool for that case: data exclusions, which are meant for conversion data outages such as being unable to upload offline conversions. Google says they are not meant for frequent or long use, and that fluctuations can last one to two conversion cycles after a week or more of affected clicks (Use data exclusions for conversion data outages, Google Ads Help).

What still resets learning, and what probably does not

Based on Google's documented triggers and the podcast, this is how I sort changes in a mature campaign:

The new channel prioritization and budget allocation settings are rolling out "over the next months", according to Google's 14 October announcement, and the post does not say whether turning them on affects learning (Google Ads & Commerce blog, 14 Oct 2026). I cover those controls in my guide to Search Partners and the new channel controls. Until Google says more, I treat them like any setting change: one at a time, with a note of the date.

What to do now

  1. Check the status reason. Hover over "Learning" before reacting. It tells you whether it was a new strategy, a setting change or a composition change.
  2. Stop rationing budget changes by a 20% rule. On a steady campaign, move the budget to where demand is. Keep target changes moderate and give each one time.
  3. Batch structural changes. If you need to add keywords, restructure ad groups or change strategy, do it in one go rather than drip-feeding edits every few days.
  4. Size your waiting time to your conversion cycle. Use Google's 3-cycle guide. For offline lead-gen, that often means weeks, not days.
  5. Protect the conversion data. If CRM uploads fail, apply a data exclusion for the affected dates instead of changing targets to compensate.
  6. Log every change with a date. When results move, you will know what caused it.

Before any of this, make sure Google is optimising to something real. If your primary conversion is still a form-fill, fix your ad tracking first. It is the same order I follow in Google Ads management: tracking, then bidding, then budget.

Frequently asked questions

How long is the Google Ads learning period?+
Google does not set a number of days. Its help page says calibration can take up to around 50 conversion events or 3 conversion cycles, and can be faster with more conversion data. Accounts with few conversions or long click-to-conversion times take longer.
Does changing the budget reset the learning phase in Google Ads?+
Not for an established campaign, according to Google Ads product manager Tal Akabas on the Ads Decoded podcast on 15 October 2026. He said target or budget changes on a steady-state campaign do not trigger a learning period. New campaigns and structural changes are a different case.
Does changing target CPA or target ROAS trigger learning?+
Google's help page lists a bid strategy setting change as a learning trigger, and a target is a setting. On the podcast, Akabas said Google scales bids in proportion to target changes using existing data, so a steady campaign does not need to relearn. He did not say how large a change stays safe, so keep moves moderate.
What should I do while a campaign is in learning?+
Google's advice is to keep running the account as normal and not judge performance until the status clears. Avoid stacking more edits, check the status reason by hovering over it, and size your waiting time to your conversion cycle.
Is the 20% budget rule from Google?+
Not that I can find. Google's help pages on the learning period and bid strategy statuses do not give any percentage threshold for budget changes. It is a rule of thumb repeated by agencies and blogs.

Related

Not sure if your campaigns are learning or just stuck?

Apply for a free strategy call. We'll look at your bid strategies, conversion setup and change history, and tell you what to leave alone and what to fix.