The 20% rule appears in none of the platform’s documentation. Not on the page about significant edits. Not on the learning phase page. Not on limited learning, not on the last significant edit metric, not on the cost reduction guidance.

What makes that absence meaningful rather than merely unproven is that the same documents are full of precise numbers. Fifty results in seven days. Seventeen purchases plus five. The platform is perfectly willing to publish a threshold when it has one. On budget percentages it publishes nothing, and it is not for want of the habit.

This page sets out what is actually documented, what the one budget example says, and how to scale without a rule nobody wrote.

What the platform does publish, precisely

Numbers, in the same pages where the 20% is supposed to live.

The learning phase, verbatim. The period when the delivery system still needs to learn about how an ad set may deliver and perform. Ad sets exit it as soon as they can deliver stably, which usually occurs after about 50 results in the week after the ad set’s last significant edit.

The level it applies to. The ad set, not the campaign.

A second, more specific threshold. For shops advertisements, a minimum of 17 purchases through your website and 5 through the platform is needed for the learning phase to complete.

Limited learning, verbatim. Described as not a penalty, but an indication that your budget is not being spent effectively because the delivery system cannot optimise performance with your current setup. An ad set becomes learning limited when it is unlikely to receive about 50 optimization events in the week after your last significant edit.

What that establishes. The platform publishes exact figures, including an oddly specific 17 plus 5, when it has one to publish.

Which is why the missing percentage matters. An organisation that will tell you it needs 17 website purchases and 5 in-app purchases is not being coy about numbers in general.

The one thing it says about budget changes

There is exactly one worked example, and it is a very wide interval.

The classification. Some edits are always significant: any change to targeting, any change to creative, any change to the optimization event, adding a new ad to the ad set, pausing for seven days or longer, and changing bid strategy.

The conditional ones. Ad set spending limit amount, bid control or cost per result goal or return goal amount, and budget amount. These are significant or not depending on the magnitude of the change.

The example, verbatim. That if you increase your budget from $100 to $101, that is not likely to cause one or more ad sets to reenter the learning phase. However, if you change your budget from $100 to $1,000, one or more ad sets may reenter the learning phase.

What that gives you. A 1% change described as probably safe, and a 900% change described as possibly not. Nothing between.

What it does not give you. 20%. Nor 15%, nor 30%. The interval where the answer changes is undefined, and the platform has chosen to leave it that way.

One exception it does state. Under automatic campaign budget, ad sets within the campaign will not reenter the learning phase as budget is distributed between them.

And the qualitative guidance. Use realistic budgets, set a budget large enough to get enough total results, and avoid frequent budget changes, which can cause an ad set to reenter the learning phase. Frequent is not defined either.

Published numeric thresholds on the advertising platform compared with the undefined budget change intervalTable contrasting the precise numeric thresholds the largest social advertising platform publishes in its delivery documentation with the deliberately undefined interval it leaves around budget changes. The platform states that the learning phase is the period when the delivery system still needs to learn how an ad set may deliver and perform, that ad sets exit it as soon as they can deliver stably, and that this usually occurs after about fifty results in the week following the ad set’s last significant edit, applied at ad set level rather than campaign level. For shops advertisements it publishes a more specific requirement of a minimum of seventeen purchases through the advertiser’s website and five through the platform for the learning phase to complete. It describes limited learning as not a penalty but an indication that budget is not being spent effectively because the delivery system cannot optimise performance with the current setup, and states that an ad set becomes learning limited when it is unlikely to receive about fifty optimization events in the week after the last significant edit. Against those precise figures, the documentation on budget changes classifies them as significant or not depending on the magnitude of the change, and provides a single worked example stating that increasing a budget from one hundred to one hundred and one dollars is not likely to cause re-entry into the learning phase, while changing it from one hundred to one thousand dollars may cause it. This provides a one percent change described as probably safe and a nine hundred percent change described as possibly not, with nothing specified between them, and the widely followed twenty percent figure appears nowhere. The platform additionally states that ad sets within a campaign using automatic campaign budget will not re-enter the learning phase as budget is distributed between them, and advises avoiding frequent budget changes without defining frequent.Precise everywhere else, silent herePublished thresholds, in the same documentsExit the learning phase~50 results in 7 days”as soon as they can deliver stably”, at ad set levelShops ads specifically17 website purchases + 5 on-platformBecome “learning limited”Unlikely to reach ~50 in 7 daysThe one budget example, and the gap it leaves$100 to $101 (+1%)“isn’t likely to cause… reenter the learning phase”$100 to $1,000 (+900%)“one or more ad sets may reenter the learning phase”Everything between +1% and +900% is undefined. Including 20%.And under automatic campaign budget, redistribution between ad sets does not restart learning at all.
An organization that publishes '17 purchases plus 5' is not being shy about numbers. It declined to publish this one. Source : Platform learning phase and significant edits documentation, via web archive (2026)

Where the number probably came from

No primary source exists, so the origin has to be inferred. The mechanism is visible if you look at how the claim circulates.

What the platform offers. A qualitative rule, “depending on the magnitude of the change”, illustrated by two extremes.

What an operator needs. A number they can put in a process document and hand to a junior buyer.

What fills that gap. Someone rounds the vague guidance into a memorable figure. It gets repeated, and repetition becomes authority.

How you can watch it happening. Search for the rule and you will find pages that assert it as established practice and, further down the same page, note that it does not appear in the platform’s documentation. Both claims, in one article, without the author noticing the contradiction.

Why 20% specifically. Probably because it is round, feels cautious, and is small enough to be safe under almost any interpretation of “magnitude”. It is a reasonable guess dressed as a policy.

Which is not the same as saying it is bad advice. Increasing budgets gradually is defensible. It is just not a rule, and it should not be presented to a client as one.

Why no fixed percentage could be correct anyway

Even if the platform wanted to publish one, the number could not be a constant, and the reason is in the definition of the learning phase itself.

What the threshold is actually about. Fifty results in a week. Not fifty percent of anything, and not a budget figure. A volume of optimization events.

What follows arithmetically. An ad set producing 200 results a week has enormous headroom. One producing 55 is a single bad week from limited learning. The same percentage budget increase puts those two accounts in completely different positions.

And cost per result moves too. Doubling budget in a thin auction may double results. Doubling it in a saturated one may raise cost per result and deliver fewer additional events than the arithmetic suggests.

So the meaningful question is not the percentage. It is whether the ad set will still clear roughly 50 results a week after the change, at the cost per result the higher budget produces.

Which is computable. Take current weekly results, current cost per result, the new budget, and an allowance for cost per result rising. If the answer stays comfortably above 50, the change is safe in the sense that matters.

Why that is better than any rule. It is specific to your account, it uses numbers you already have, and it explains why the same 30% increase is fine on one ad set and destabilising on another.

Why a fixed percentage budget threshold cannot be correct given a volume-based learning thresholdWorked comparison demonstrating why a fixed percentage threshold for budget increases cannot be correct, given that the platform’s documented learning phase threshold is expressed as a volume rather than a percentage. The documented threshold is approximately fifty results in a week, meaning a count of optimization events rather than any proportion of a budget. Consider two ad sets each receiving the same percentage budget increase. The first currently produces two hundred results per week and therefore possesses substantial headroom above the fifty-result threshold, so even if cost per result rises following the increase it remains comfortably clear of limited learning. The second currently produces fifty-five results per week and sits a single poor week away from limited learning, so the same percentage increase places it at genuine risk if cost per result rises at all. Cost per result is itself variable: doubling budget in a thin auction may approximately double results, whereas doubling it in a saturated auction may raise cost per result and deliver fewer additional events than simple arithmetic would suggest. The meaningful question for an advertiser is therefore not what percentage the budget increased by, but whether the ad set will still clear approximately fifty results per week after the change, at whatever cost per result the higher budget produces. That question is computable from figures the advertiser already possesses: current weekly results, current cost per result, the proposed new budget, and an allowance for cost per result increasing. If the projected figure remains comfortably above fifty, the change is safe in the sense that actually matters. This approach is superior to any fixed rule because it is specific to the account, uses existing data, and explains why an identical thirty percent increase can be harmless on one ad set and destabilising on another.Same increase, two ad sets, opposite riskBecause the documented threshold is a volume, not a percentage.Ad set A200 results a week todayThreshold is at 50. Enormous headroom.A 30% increase is a non-event, even ifcost per result rises.Ad set B55 results a week todaythreshold at 50One bad week from limited learning.The same 30% is a real risk if cost perresult moves at all.The question worth asking insteadWill this ad set still clear ~50 results a week after the change, at the cost per result thehigher budget produces?Computable from what you already haveCurrent weekly results, current cost per result, the new budget, and an allowance for CPA rising.
The documented threshold is a volume, not a percentage. So the safe increase depends on where you are starting from. Source : Applying the platform's own 50-results threshold (2026)

What actually restarts the learning phase

The always-significant list is short, checkable, and more useful than the percentage argument.

Always significant. Any change to targeting. Any change to ad creative. Any change to the optimization event. Adding a new ad to your ad set. Pausing your ad set for seven days or longer. Changing bid strategy.

Note what is on that list. Adding a new ad. Which means a creative refresh, the thing most accounts do routinely, is guaranteed to restart learning in a way a budget change may not.

Note the seven-day pause. Turning campaigns off over a holiday period costs you the learning state if the pause runs a week.

Conditionally significant. Spending limit, bid control or cost goal, and budget amount, all depending on magnitude.

What this reorders. If you are worried about learning resets, the budget percentage is the least of it. Your creative rotation and your pause behaviour matter more, and both are documented as always significant.

The practical consequence. Batch your changes. If you are going to refresh creative and raise budget, do both at once rather than a week apart, because you are paying the learning cost either way and you may as well pay it once.

Edits classified by the platform as always significant compared with those significant only by magnitudeTable separating the advertising platform’s list of edits that always constitute a significant edit, and therefore always restart the learning phase, from those that are significant only depending on the magnitude of the change. The always significant list comprises any change to targeting, any change to advertisement creative, any change to the optimization event, adding a new advertisement to the ad set, pausing the ad set for seven days or longer, and changing the bid strategy. Two entries on that list carry practical consequences that most advertisers overlook: adding a new advertisement is always significant, which means that a routine creative refresh guarantees a learning phase restart in a way that a budget change only might; and pausing for seven days or longer is always significant, which means that switching campaigns off across a holiday period costs the learning state once the pause reaches a week. The conditionally significant list comprises the ad set spending limit amount, the bid control or cost per result goal or return on ad spend goal amount, and the budget amount, each of which is significant or not depending on the magnitude of the change. The practical reordering this implies is that an advertiser concerned about learning phase resets should attend to creative rotation and pause behaviour before attending to budget percentages, since the former are documented as always significant while the latter is documented as conditional. The corresponding operational recommendation is to batch changes together, so that if creative is to be refreshed and budget raised, both are done simultaneously rather than a week apart, because the learning cost is incurred either way and incurring it once is preferable to incurring it twice.Most accounts worry about the wrong editAlways significantAny change to targetingAny change to ad creativeAny change to optimization eventChanging bid strategyAdding a new ad to your ad setAnd: pausing for 7 days or longerSignificant “depending on magnitude”Ad set spending limit amountBid control, cost goal or return goalBudget amountThe one everybody argues about is here,in the conditional column, with nothreshold attached.Two consequences people missA routine creative refresh always restarts learning. So does switching off for a week over a holiday.So batch your changesRefresh creative and raise budget together. You pay the learning cost either way; pay it once.
Adding a new ad is always significant. A budget change only might be. Most accounts worry about the wrong one. Source : Platform significant edits documentation, via web archive (2026)
The propagation of a qualitative platform guideline into an unsourced numeric ruleDiagram tracing how a qualitative guideline published by an advertising platform becomes a specific numeric rule that no source supports. At the origin the platform publishes a qualitative statement, namely that a budget change is significant or not depending on the magnitude of the change, illustrated only by two extreme examples: an increase from one hundred to one hundred and one dollars described as not likely to restart the learning phase, and a change from one hundred to one thousand dollars described as one that may restart it. An advertising operator requires something different from this: a specific figure that can be written into a process document and given to a junior media buyer to follow. That gap is filled when someone rounds the vague guidance into a memorable number, after which repetition supplies the authority that evidence never did. The propagation can be observed directly by searching for the rule, which returns pages that assert the twenty percent figure as established practice while also noting, sometimes within the same article, that the figure does not appear anywhere in the platform’s documentation, with the author apparently not registering the contradiction between the two statements. The specific choice of twenty percent is plausibly explained by the figure being round, feeling cautious, and being small enough to remain safe under almost any interpretation of the word magnitude, which makes it a reasonable guess presented as a policy. None of this establishes that gradual budget increases are poor practice, since increasing budgets gradually remains defensible on its own terms; it establishes only that the practice is not a rule and should not be presented to a client as though the platform had published it.How a guideline becomes a ruleWhat the platform publishes”depending on the magnitude of the change”, plus two extreme examples.+1% probably fine. +900% maybe not. Nothing between.↓What an operator needsA number that fits in a process document and can be handed to a junior buyer.↓What fills the gapSomebody rounds the guidance. Repetition supplies the authority.↓What you can watch happening todayPages that state the rule as established practice, and note it appears in no documentation, in the same article.None of which makes gradual increases bad practice. It makes them a choice, not a rule.
Search for it and you will find pages asserting the rule and denying its existence, in the same article. Source : Citation pattern observed while tracing the claim (2026)

How to scale without a rule

Five practices that survive the absence of a threshold, because none of them depends on one.

Change in increments that matter. A 10% budget increase on a small account is rounding error and teaches you nothing. If you are going to disturb delivery, disturb it enough to observe something.

Leave enough time between changes to reach stability. The documented marker is roughly 50 results in seven days. If your ad set does not produce 50 results in a week at any budget, learning phase discussions are not your problem; volume is.

Judge on cost per meeting held, not on the status label. The learning phase indicator tells you the system is still calibrating. It does not tell you whether the campaign is working.

Batch changes deliberately. One change window per week beats continuous fiddling, because continuous fiddling is exactly the “frequent budget changes” the platform warns against without quantifying.

Write down what you changed and when. Almost nobody does this, and it is the only way to attribute a performance shift to an action rather than to a story invented afterwards.

And if your ad set cannot reach 50 results a week. Consolidate. Fewer ad sets with more volume each is the documented direction, and it is also what the platform’s own limited-learning guidance recommends: combining ad sets, broadening the audience, raising budget or changing the optimization event.

Five operating practices for scaling advertising budgets in the absence of a published thresholdDiagram setting out five operating practices for scaling advertising budgets that do not depend on any published percentage threshold, since no such threshold exists in the platform’s documentation. The first practice is to change budgets in increments large enough to matter, on the grounds that a ten percent increase on a small account is rounding error that teaches the advertiser nothing, so if delivery is going to be disturbed it should be disturbed sufficiently for something to be observable. The second is to leave enough time between changes for the ad set to reach stability, using the documented marker of approximately fifty results in seven days, with the corollary that if an ad set cannot produce fifty results in a week at any budget then the advertiser’s problem is volume rather than the learning phase. The third is to judge outcomes on cost per meeting held rather than on the learning phase status label, since that indicator reports that the delivery system is still calibrating and says nothing about whether the campaign is achieving business results. The fourth is to batch changes deliberately into a single weekly change window rather than adjusting continuously, since continuous adjustment is precisely the frequent budget changes the platform warns against without ever quantifying. The fifth is to record what was changed and on what date, which almost no advertiser does and which is the only means of attributing a subsequent performance shift to a specific action rather than to a narrative constructed after the fact. Where an ad set cannot reach fifty results per week, the platform’s own limited learning guidance recommends consolidating by combining ad sets, broadening the audience, raising the budget or changing the optimization event.Five practices that need no threshold1. Increments that matter10% on a small account teaches you nothing.2. Time to stabiliseThe documented marker: ~50 results in 7 days.3. Judge on cost per meeting heldNot on whether a status label reappeared.4. One change window a weekContinuous fiddling is the thing they warn about.5. Write down what you changed, and whenThe only way to attribute a shift to an action rather than to a story invented afterwards.If your ad set cannot reach 50 results a week at any budgetThen the learning phase is not your problem. Volume is. Consolidate.The platform’s own remedy list for limited learning:combine ad sets, broaden the audience, raise budget, change the optimization event.
If your ad set cannot reach 50 results a week at any budget, the learning phase is not your problem. Source : Platform learning phase guidance, via web archive (2026)

Where to go next

You are deciding how many creatives to run. How many creatives to test.

Your returns fall as you spend more. Why ROAS drops when scaling.

You are weighing the automatic campaign type. Meta Advantage+.

You want the account structure around this. Meta ads account structure.

Your creative has been running a long time. Creative fatigue metrics.

You want the buying mechanics underneath. Media buying explained.

In short

  • The 20% rule appears in no platform documentation: not on significant edits, the learning phase, limited learning, the last-significant-edit metric or cost reduction guidance.
  • The same documents publish precise numbers, including 50 results in seven days and, for shops ads, 17 website purchases plus 5 on-platform.
  • On budget the platform gives one example: $100 to $101 probably will not restart learning, $100 to $1,000 may. Everything between is undefined.
  • Budget changes are only conditionally significant, “depending on the magnitude of the change”.
  • Adding a new ad is always significant, so a routine creative refresh restarts learning where a budget change might not.
  • So is pausing for seven days or longer, which catches anyone switching off over a holiday.
  • Automatic campaign budget redistribution does not restart learning between ad sets in a campaign.
  • The guidance on budget is qualitative: realistic budgets, large enough for enough results, and avoid frequent changes. Frequent is not defined.

Batch your changes, record them, and judge on cost per meeting held. Book a diagnostic, or see how we approach B2B paid acquisition.