---
title: "Google Ads Learning Runs on Conversions, Not a Calendar"
description: "The Learning badge has a time limit. Smart Bidding’s performance depends on conversion volume and delay. Use your budget and CPA to plan a launch window."
url: "https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/92d83a96-dfb1-4ba3-a904-6ecd9d6e58e7.png"
published: "2026-11-03T06:47:06.426Z"
modified: "2026-11-03T06:47:06.487Z"
---

November 3, 2026 · 10 min read

# Google Ads Learning Runs on Conversions, Not a Calendar

[Alexander PerelmanHead Of Product @ groas](https://groas.com/author/alexander-perelman)[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![A golden cake sliced open to reveal a raw, oozing centre beside a kitchen timer at zero: the time is up, but the baking isn't finished.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/92d83a96-dfb1-4ba3-a904-6ecd9d6e58e7.png)

In this article

1. [The bidder needs answers, not elapsed days](#the-bidder-needs-answers-not-elapsed-days)
2. [Why seven days and 50 conversions answer different questions](#why-seven-days-and-50-conversions-answer-different-questions)
3. [An edit restarts learning when it changes the question](#an-edit-restarts-learning-when-it-changes-the-question)
4. [Learning (limited) is a volume problem you can calculate](#learning-limited-is-a-volume-problem-you-can-calculate)
5. [Build the launch window from volume, then add delay](#build-the-launch-window-from-volume-then-add-delay)
6. [Keep optimizing without moving the goalposts](#keep-optimizing-without-moving-the-goalposts)

Most agencies won’t tell you this: the Learning badge is not the learning. [Smart Bidding predicts conversion rates and sets a bid for each auction](https://support.google.com/google-ads/answer/10966879?hl=en-GB\&ref_topic=10964964), so its performance depends on the conversion data it receives, not how many days have passed since launch. [Google says the Learning status never lasts more than seven days](https://support.google.com/google-ads/answer/13020501?hl=en). **A badge can clear before the bids have settled.**

I covered what to check and what to leave alone in [_The First 28 Days of Google Ads: What to Check and What Not to Touch_](https://groas.com/post/the-first-28-days-checklist-what-to-chec). This second look explains why a fixed checklist cannot tell every account when to judge performance. The answer depends on how many conversions arrive and how long they take to arrive.

Keep one example in mind: you spend $3,000 a month and pay $60 per conversion. That gives the bidder about 50 conversions a month. Change the CPA while keeping the budget fixed, and nearly every sensible launch decision changes with it.

## The bidder needs answers, not elapsed days

Smart Bidding makes a prediction for each auction: how likely is this click to produce the conversion you asked for, and what might that conversion be worth? [Google says its auction-time bidding uses contextual signals to predict conversion rate and probable value, then sets a unique bid](https://support.google.com/google-ads/answer/10966879?hl=en-GB\&ref_topic=10964964). At launch, the prediction draws on what the system already knows, including account history. Reported conversions give it more direct evidence about this campaign.

Now do the division. At $3,000 a month and $60 CPA, you expect roughly 50 conversions a month, or 12 a week. At $30 CPA, the same spend produces about 100 a month. At $120 CPA, it produces 25, or roughly six a week. The accounts share a launch date and a budget. They do not give the bidder answers at the same rate.

Impressions and clicks still tell you something about delivery and interest. They cannot, by themselves, show which combinations of query, device, location, time and audience produce the conversion you care about. **For bidding toward conversions, conversion volume is the scarce input.**

Even that input arrives late. A click today might convert three, five or nine days from now. Until then, the account shows the cost without the eventual result. [Google recommends judging performance over two conversion cycles or two weeks, whichever is longer, while accounting for conversion delay](https://support.google.com/google-ads/answer/13020501?hl=en). If the buyer in our example typically takes seven days, grading the bids after the first week means grading a set of clicks before many of their outcomes arrive. I used to tell clients to wait for statistical significance. I was wrong about what to wait for: first, wait for the conversions to show up.

## Why seven days and 50 conversions answer different questions

The seven-day rule describes the _Learning_ label. [Google says the status marks a recent change, ads continue running while it shows, and the status never lasts more than seven days](https://support.google.com/google-ads/answer/13020501?hl=en). It is a cap on a visible status, not a certificate that performance has stabilized. [The same guidance says Smart Bidding keeps learning and can take one to two conversion cycles to converge on a new goal](https://support.google.com/google-ads/answer/13020501?hl=en).

I treat the badge like a download bar reaching 100 percent while the installer is still running. The analogy only goes so far: a bidder does not finish learning once and stop. But it captures the mistake of treating the end of a visible process as the end of the work underneath.

The 50-conversion figure addresses volume instead. [A September 2026 account of Google’s reframing described calibration in terms of up to around 50 conversion events or three conversion cycles, with history potentially speeding the process](https://searchengineland.com/google-ads-reframes-smart-bidding-learning-around-50-conversions-490382). That is an approximate benchmark, not an entry ticket the account must collect before a bid can work. [One practitioner guide uses at least 30 conversions in 30 days for Target CPA and 50 for Target ROAS as working thresholds](https://growwithvysta.com/blog/google-ads-conversion-tracking-for-ecommerce-how-to-feed-smart-bidding-clean-data/). Those are planning rules, not guarantees.

Our $3,000 campaign at $60 CPA sits at about 50 conversions a month: above that Target CPA rule of thumb and on the edge of the Target ROAS one. At $30 CPA, it gathers the same number in roughly half the time. At $120 CPA, it cannot gather 50 in a month at that spend. **Use the badge to understand the status; use conversion volume and delay to decide when to judge results.**

## An edit restarts learning when it changes the question

A common launch rule says a budget change of more than 20 percent restarts learning. I used to repeat it. [Google’s October 2026 guidance instead names three triggers for _Learning_ status: a new or reactivated bid strategy, a composition change involving campaigns, ad groups or keywords, and a significant goal change](https://support.google.com/google-ads/answer/13020501?hl=en). [It says budget, target, creative and targeting changes do not remove historical data or trigger that status](https://support.google.com/google-ads/answer/13020501?hl=en). The old budget-reset rule does not describe the badge anymore.

The three triggers make sense if you ask what the bidder is trying to predict:

- **A new bid strategy changes the task.** Maximize Conversions pursues conversion volume; Target ROAS pursues value against a return target. Old evidence can still help, but it does not answer the new question in quite the same way.
- **A significant conversion-goal change changes the correct answer.** Swap the outcome you optimize for, or change how it is counted, and yesterday’s successes no longer describe the goal in the same terms.
- **A composition change changes the auctions in the task.** Add 50 keywords to a 20-keyword campaign and the bidder faces a different mix of queries. History remains useful, but less of it applies directly.

**If you change the question, expect fresh calibration.** If you adjust what you will pay while pursuing the same answer, do not assume the badge must return.

That does not mean every edit is harmless or every badge-free change is small. An exact-match negative for obvious junk, a like-for-like replacement for a disapproved headline, or a repaired URL can fix a problem without changing the bid strategy or conversion goal. Tracking repairs are different: fixing duplicate tags or a blocked tag can make the before-and-after reports hard to compare. Leave the fault in place, though, and the bidder continues receiving bad conversion data. Repair it, log the date and read the numbers across that break with care.

Budget and target edits also deserve a distinction between **no reset** and **no effect**. [Google Ads PM Tal Akabas said established, steady-state campaigns do not enter a learning period after target or budget edits because the system uses existing bid and budget information](https://ppc.land/google-ads-pm-says-target-or-budget-changes-cause-no-learning-period/). A new campaign still has little of its own history. And [the 20 percent budget-reset rule is an agency rule of thumb, not a Google rule; large changes can still unsettle performance without showing _Learning_](https://www.sfdigital.co.uk/blog/google-ads-learning-phase-budget/). Double our example budget from $3,000 to $6,000 overnight and you can afford auctions you were not buying before. The badge may stay off while the results adjust to that new mix.

## _Learning (limited)_ is a volume problem you can calculate

_Learning (limited)_ is easier to understand without treating it as a penalty. [Practitioners describe the underlying problem as too little budget, targeting reach or conversion volume for the tests the bidder needs](https://www.sfdigital.co.uk/blog/google-ads-learning-phase-budget/). Start with the part you can calculate before launch: monthly budget divided by expected CPA.

At $3,000 and $60 CPA, our campaign expects 50 conversions a month. At $30 CPA, it expects 100. At $120 CPA, it expects 25. If that last campaign also has a seven-day sales cycle, the bidder receives about six delayed outcomes a week from which to price many auctions. Waiting for the seventh day on the calendar does not create more evidence. [One practitioner volume map puts 30 or more conversions a week at roughly a week to settle, 10 to 30 at one to two weeks, and fewer than 10 at three weeks or more, often with _Limited_ showing](https://www.sfdigital.co.uk/blog/google-ads-learning-phase-budget/). Treat those windows as estimates, not promises.

The arithmetic points to three ways to get more usable signal:

- Raise the budget so the same CPA can produce more conversions.
- Loosen a restrictive target, accepting a higher CPA in exchange for more conversion volume.
- Bid toward a higher-volume conversion, such as a qualified lead rather than a closed deal, while you build enough data to return to the stricter goal.

You can also consolidate campaigns when scattered volume is the problem; that changes the structure and deserves its own calibration window. What you cannot do is keep a $3,000 budget, a $120 CPA, a low-volume purchase goal and an unchanged structure, then refresh the status page until the arithmetic improves. **Budget divided by CPA is a better starting point than a date on the launch plan.**

![Budget divided by CPA, with three examples of expected monthly conversions](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/63d665c7-75b4-4bd4-a61c-485b734dfff2.png)

## Build the launch window from volume, then add delay

The first calculation gives expected conversions per month. The second asks when those conversions become visible. Together, they give you a defensible window for judging the campaign rather than a date borrowed from someone else’s account.

Take our three versions of the $3,000 campaign. At $60 CPA, you expect 50 conversions in a month, so a four-week read gives you a reasonable chance of seeing the 30 to 50 events used in the planning rules above. At $30 CPA, you expect 100 a month; about two weeks could give you 50, making a two-week read defensible if outcomes arrive quickly. At $120 CPA, you expect 25 in a month. Four weeks will not turn that into a high-volume account.

Then account for the conversion cycle. [Google’s recommendation is to evaluate over two cycles or two weeks, whichever is longer, with conversion delay included](https://support.google.com/google-ads/answer/13020501?hl=en). A same-day lead form with sufficient volume can be read sooner than a B2B demo with a 21-day cycle: at day 14, many of the latter campaign’s outcomes have not arrived. A seven-day cycle and 12 conversions a week point toward a three- to four-week read. A 21-day cycle and six conversions a week point past week six.

**The window must be long enough to collect outcomes and long enough for them to be reported.** If the calculation says you will not reach 30 events in 30 days, decide before launch whether to consolidate campaigns, bid to a higher-volume step, or treat month one as a period for collecting signal rather than demanding a settled CPA.

## Keep optimizing without moving the goalposts

This is the practical distinction behind the October rewrite. Clear junk negatives, disapproval fixes and tracking repairs can happen while the bidder works. A new strategy, a different conversion goal or a composition change asks it to solve a different problem. [Independent discussion of the rewrite makes the same distinction between the label clearing and results settling over one to two conversion cycles](https://www.jwcperformance.co.uk/blog/google-ads-smart-bidding-learning-period-50-conversions).

I log an edit with its date and ask: did I change what the bidder is trying to predict, or what I will pay for the same prediction? If it is the latter, I make the edit and allow enough time to read its effect. Constant attention is useful. Constantly changing the question is not.

Once the window you calculated has passed, the next decision follows the evidence. If CPA is within 20 percent of target after allowing for lag and you have seen 30 to 50 events, scale in 20 percent budget steps a week apart and measure each step. If volume never arrived, consolidate or move to a higher-volume conversion: the same issue comes up when [moving from manual to AI bidding](https://groas.com/post/from-manual-bidding-to-ai-powered-biddin). If the auction mix itself was wrong, restructure once and give the changed campaign a fresh window. For a closer look at what happens during that window, see [the first 28 days as three clocks rather than one launch schedule](https://groas.com/post/what-google-is-actually-doing-in-your-first-28-days-a-new).

That is where groas’s always-on execution earns its place. Small, safe fixes can happen continuously while a human strategist owns the direction, guardrails and timing. The account does not need a weekly strategy reset dressed up as optimization.

Now try a case that was not in the running example. You spend $2,000 a month pursuing $150 SQLs with a 14-day cycle. Division gives you about 13 conversions a month, roughly three a week, and many arrive late. No seven-day badge rule fixes that. You can raise the budget, bid toward a $50 demo-booked goal until you collect more events, or accept a window beyond six weeks without moving the target every few days. **Pick the window from the math, then give the bidder time to receive the answers.**

## Frequently Asked Questions

### How long does Google Ads Smart Bidding take to learn after launch?

Smart Bidding's performance depends on the conversion data it receives, not on how many days have passed since launch. Google says the Learning status never lasts more than seven days, but the badge can clear before the bids have settled, so elapsed days alone do not tell you when results have stabilized.

### Does the Learning badge disappearing mean Smart Bidding is done learning?

No. The Learning badge only marks a recent change, and Google says the status never lasts more than seven days. It is a cap on a visible status, not a certificate that performance has stabilized, and Smart Bidding can take one to two conversion cycles to converge on a new goal.

### How many conversions per month does a Google Ads campaign need for Smart Bidding?

A practical starting point is monthly budget divided by target CPA. A $3,000 budget at a $60 CPA gives about 50 conversions a month, while the same budget at a $120 CPA gives only 25. Common planning rules use at least 30 conversions in 30 days for Target CPA and 50 for Target ROAS, but these are estimates, not guarantees.

### When should I judge the performance of a new Google Ads campaign?

Google recommends evaluating performance over two conversion cycles or two weeks, whichever is longer, while accounting for conversion delay. A click today might convert several days later, so judging too early grades clicks before their outcomes have arrived. A seven-day cycle at 12 conversions a week points toward a three- to four-week read.

### Does changing the budget in Google Ads restart the learning phase?

No. Google names three triggers for Learning status: a new or reactivated bid strategy, a composition change involving campaigns, ad groups or keywords, and a significant conversion-goal change. Budget, target, creative and targeting changes do not remove historical data or trigger Learning status. The 20 percent budget-reset rule is an agency rule of thumb, not a Google rule.

### What should I do if my Google Ads campaign shows Learning (limited)?

Learning (limited) reflects too little budget, targeting reach or conversion volume for the tests the bidder needs. Three ways to add usable signal are to raise the budget, loosen a restrictive target to accept a higher CPA in exchange for more volume, or bid toward a higher-volume conversion such as a qualified lead instead of a closed deal.

### Can I edit my Google Ads campaign while Smart Bidding is still learning?

Yes, as long as the edit does not change what the bidder is trying to predict. Exact-match negatives for junk queries, like-for-like replacements for disapproved headlines and repaired URLs can be made while the bidder works. Tracking repairs such as fixing duplicate or blocked tags should be logged with the date, because they break before-and-after comparisons.

### How do I decide how long to wait before evaluating a new Google Ads campaign?

Calculate expected conversions by dividing monthly budget by target CPA, then add the conversion delay. At a $30 CPA with 100 conversions a month and fast outcomes, a two-week read can be defensible. At a $120 CPA with 25 conversions a month and a 21-day cycle, the read points past week six. The window must be long enough for outcomes to arrive and be reported.

## Paid Ads & SEO, Run For You

Businesses pay one flat monthly fee per service, agencies resell it under their own brand, and groas does the work either way.

[See Pricing](https://groas.com/pricing)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","description":"groas is a done-for-you marketing service, not software: elite people working with proprietary technology systems run search ads, Meta ads, SEO and AI search for businesses, and white label for agencies, with a dedicated senior strategist owning every account.","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the#webpage","url":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the","name":"Google Ads Learning Runs on Conversions, Not a Calendar","description":"The Learning badge has a time limit. Smart Bidding’s performance depends on conversion volume and delay. Use your budget and CPA to plan a launch window.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/92d83a96-dfb1-4ba3-a904-6ecd9d6e58e7.png"},"breadcrumb":{"@id":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the#breadcrumb"},"dateModified":"2026-11-03T06:47:06.487Z","mainEntity":{"@id":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"Google Ads Learning Runs on Conversions, Not a Calendar","item":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the#article","url":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the","mainEntityOfPage":{"@id":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"Google Ads Learning Runs on Conversions, Not a Calendar","description":"The Learning badge has a time limit. Smart Bidding’s performance depends on conversion volume and delay. Use your budget and CPA to plan a launch window.","datePublished":"2026-11-03T06:47:06.426Z","dateModified":"2026-11-03T06:47:06.487Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/92d83a96-dfb1-4ba3-a904-6ecd9d6e58e7.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"conversion, conversions, cpa, learning, budget, google, volume, month","wordCount":2100,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"My background includes Goldman Sachs and Stanford Computer Science.","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/why-google-ads-learning-phase-runs-on-conversions-not-the#faq","mainEntity":[{"@type":"Question","name":"How long does Google Ads Smart Bidding take to learn after launch?","acceptedAnswer":{"@type":"Answer","text":"Smart Bidding's performance depends on the conversion data it receives, not on how many days have passed since launch. Google says the Learning status never lasts more than seven days, but the badge can clear before the bids have settled, so elapsed days alone do not tell you when results have stabilized."}},{"@type":"Question","name":"Does the Learning badge disappearing mean Smart Bidding is done learning?","acceptedAnswer":{"@type":"Answer","text":"No. The Learning badge only marks a recent change, and Google says the status never lasts more than seven days. It is a cap on a visible status, not a certificate that performance has stabilized, and Smart Bidding can take one to two conversion cycles to converge on a new goal."}},{"@type":"Question","name":"How many conversions per month does a Google Ads campaign need for Smart Bidding?","acceptedAnswer":{"@type":"Answer","text":"A practical starting point is monthly budget divided by target CPA. A $3,000 budget at a $60 CPA gives about 50 conversions a month, while the same budget at a $120 CPA gives only 25. Common planning rules use at least 30 conversions in 30 days for Target CPA and 50 for Target ROAS, but these are estimates, not guarantees."}},{"@type":"Question","name":"When should I judge the performance of a new Google Ads campaign?","acceptedAnswer":{"@type":"Answer","text":"Google recommends evaluating performance over two conversion cycles or two weeks, whichever is longer, while accounting for conversion delay. A click today might convert several days later, so judging too early grades clicks before their outcomes have arrived. A seven-day cycle at 12 conversions a week points toward a three- to four-week read."}},{"@type":"Question","name":"Does changing the budget in Google Ads restart the learning phase?","acceptedAnswer":{"@type":"Answer","text":"No. Google names three triggers for Learning status: a new or reactivated bid strategy, a composition change involving campaigns, ad groups or keywords, and a significant conversion-goal change. Budget, target, creative and targeting changes do not remove historical data or trigger Learning status. The 20 percent budget-reset rule is an agency rule of thumb, not a Google rule."}},{"@type":"Question","name":"What should I do if my Google Ads campaign shows Learning (limited)?","acceptedAnswer":{"@type":"Answer","text":"Learning (limited) reflects too little budget, targeting reach or conversion volume for the tests the bidder needs. Three ways to add usable signal are to raise the budget, loosen a restrictive target to accept a higher CPA in exchange for more volume, or bid toward a higher-volume conversion such as a qualified lead instead of a closed deal."}},{"@type":"Question","name":"Can I edit my Google Ads campaign while Smart Bidding is still learning?","acceptedAnswer":{"@type":"Answer","text":"Yes, as long as the edit does not change what the bidder is trying to predict. Exact-match negatives for junk queries, like-for-like replacements for disapproved headlines and repaired URLs can be made while the bidder works. Tracking repairs such as fixing duplicate or blocked tags should be logged with the date, because they break before-and-after comparisons."}},{"@type":"Question","name":"How do I decide how long to wait before evaluating a new Google Ads campaign?","acceptedAnswer":{"@type":"Answer","text":"Calculate expected conversions by dividing monthly budget by target CPA, then add the conversion delay. At a $30 CPA with 100 conversions a month and fast outcomes, a two-week read can be defensible. At a $120 CPA with 25 conversions a month and a 21-day cycle, the read points past week six. The window must be long enough for outcomes to arrive and be reported."}}]}]}
```