How Long Does It Take for Facebook Ads to Start Working?

The most expensive thing a founder does in their first month of paid social is make decisions on day three. The numbers are right there, they are bad, and doing nothing feels like negligence.

Doing nothing is usually correct. Meta’s delivery system, its attribution reporting and the ad review queue each run on their own clock, and none of those clocks has finished by day three.

How long before Facebook ads start delivering?

Facebook ads start delivering within roughly 24 hours of being submitted, because Meta reviews most ads within that window before they can serve at all. That review time is Meta’s own stated target, reported in Foreplay’s 2026 guide to Facebook ad review, and while many ads clear in minutes, a new account or a regulated category can sit longer.

Delivery is not the same as results, and conflating the two causes most of the early panic. An approved ad starts spending immediately, so by hour 12 there is a dashboard with impressions, clicks and a cost per result that looks alarming. That figure is real spend against incomplete outcomes.

Two things make the first 72 hours unreadable. The delivery system has not yet worked out which people, placements and times respond to the ad, so early impressions are deliberately exploratory rather than optimised. And the attribution window has barely opened, so conversions that will eventually be credited to day one have mostly not happened yet.

The practical rule is that the first three days tell you about delivery mechanics, not performance. Use them to confirm the ad is approved, the pixel is firing, the landing page loads, and the budget is actually spending.

Those four checks are worth doing on day one. A verdict on the creative is not available yet, and looking for one anyway is how a working ad gets switched off before it has run.

How long does Facebook ad review take?

Facebook ad review takes under 24 hours in most cases, and Meta’s own ads review policy states that review is typically completed within 24 hours while adding that it can take longer and that an ad may be reviewed again after it is already live. Plenty of ads clear in minutes.

Treat the 24 hours as a planning figure and not as a promise, because the cases that break it break it by days rather than by hours.

What the queue inspects explains most of the variance. Review looks at the ad text, the images and video, the targeting, and the destination the ad points at, so a landing page that loads slowly, redirects, or changed after you submitted can hold up a decision on creative that is perfectly compliant.

A brand new ad account and a regulated category each add scrutiny on top of that, which is the pattern reported in Foreplay’s 2026 guide to Facebook ad review.

Some of the delay is self-inflicted. Editing an ad while it sits in review resubmits it, so the clock starts again, and editing the same ad twice in an afternoon is how a four hour approval becomes a two day one. Resubmitting a rejected ad unchanged, in the hope that a different reviewer sees it differently, wastes the appeal route that exists for that exact situation.

In the accounts we run, 48 hours is the point at which a review has stopped being normal and a support ticket is more use than more patience. Before raising one, check the account quality page for a notification, because a decision that looks like silence in Ads Manager is often a rejection sitting unread somewhere else.

The clause about re-review is the one to remember after launch. An ad that has delivered happily for a week can be pulled back into review and paused, which means a campaign that stops spending in week two is sometimes a policy event rather than a market one.

That difference is diagnosable in a minute from the delivery column, and we work through the rest of the causes in why your Facebook ads stopped working.

How long does it take for Facebook ads to start spending?

Facebook ads normally start spending inside the first hour after approval, and an approved ad set that has spent nothing after a full day has a fault rather than a slow start. The two things to hold apart are approval and delivery: nothing spends while an ad is in review, and almost everything spends soon after it is not.

Pacing is why the first few hours look wrong anyway.

Meta spreads a daily budget across the whole day instead of spending it as fast as the auction allows, so an account checked an hour after launch shows a small fraction of the daily number and an alarming cost per result built on a handful of impressions. Wevion’s 2026 guide to Meta budget pacing describes the same mechanism from the delivery side: pacing is designed to find the better opportunities later in the day rather than to clear the budget early.

The day being paced is not your day.

Daily budgets and pacing reset at midnight in the ad account’s time zone, per the same pacing guidance, and the account’s time zone is set when the account is created and effectively locked once it has recorded spend, according to Agrowth’s walkthrough of the Ads Manager time zone setting.

An account created in another region therefore starts and ends its day at hours that have nothing to do with your clock, which is the mundane explanation for most reports of a budget that stopped early or ran over.

Why would an approved ad set be spending nothing at all?

An approved ad set spending nothing at all is nearly always announcing the reason in its delivery column, which is worth reading before anything gets changed. These are the statuses that account for almost every case, drawn from the diagnostic sequence in AdLibrary’s 2026 playbook for an ad that is not delivering.

What the delivery column says	What it means	What to do
In Review	the ad has not been approved yet, so nothing can spend	wait out the 24 hours above, then raise a ticket
Not Delivering	the campaign, ad set or ad is off, or the start date is in the future	check all three levels are active and the schedule has started
Learning Limited	the budget cannot produce 50 events in 7 days, so delivery stays sporadic	raise the budget or optimise for a cheaper event
Account Spending Limit Reached	an account level cap has been hit, which pauses everything under it	clear or raise the limit in billing settings
Audience Too Small	targeting, exclusions and overlap have left too few people to deliver against	widen the audience or drop an exclusion
Rejected	a policy decision, which no amount of waiting changes	fix the ad or appeal, rather than resubmitting it unchanged

Two causes sit outside that column. A payment method that fails its first charge stops delivery without any obvious signal in the campaign itself, so billing is worth checking before targeting.

And a manual bid cap set below what the auction is clearing at produces an ad set that is active, compliant and losing every auction, which looks identical to a broken pixel from the dashboard and is fixed by raising the cap or switching to automatic bidding.

Meta does not publish a first-day spend cap for new accounts, so the ramp advice in warm-up guides is practitioner convention rather than a documented tier. What is documented is the ceiling on the other side: a daily budget can be exceeded on a single day and is bounded across the week, and we work through those limits in how much to spend on Facebook ads per day.

How long before a new ad in a running ad set starts spending?

A new ad in a running ad set can take days to start spending, and it may never spend at all, which Meta treats as the system working rather than as a fault.

There is no ad-level budget to appeal to. Terrads’ write-up of the behaviour notes that the Marketing API’s ad object carries no budget field, that Meta documents no per-ad budget, spend floor or spend cap anywhere, and that Meta describes the uneven delivery which follows as intended.

The same write-up describes new creatives added to a working ad set accumulating a few hundred impressions between them while the established ad keeps the money.

The mechanism is the same one that makes the learning phase work. Budget sits at the ad set, delivery goes to whichever ad the system expects to produce the cheapest result, and a new ad is bidding against an ad with weeks of history behind it.

The system is not withholding a verdict on your new creative. It has made one on the strength of very little evidence, and it is rational for it to do so.

What that costs is the thing the rest of this article measures. The day-seven and day-fourteen clocks start when an ad begins getting delivery, not when it is published, so an asset sitting at 300 impressions after three days has not been tested and is not evidence of anything.

Judging it as a loser retires creative that was never tried, which is the quiet version of the mistake this article opens with.

So read spend before you read performance. The first question about any new ad is what share of the ad set’s budget it received, and only an ad that got a meaningful share is a candidate for a verdict at all. An ad that got nothing is a delivery problem, and the fixes for it are different from the fixes for a bad hook.

How do you force spend onto a new ad on Meta?

You can force spend onto a new ad on Meta with the Push Delivery to This Ad setting, which Meta began rolling out in mid-2026, or by taking the ad out of the competition and giving it its own ad set. Neither is a documented guarantee, and the first is still behaving like a test.

Jon Loomer’s hands-on account of Push Delivery describes the mechanics: the option appears only once an ad set holds at least two ads, active or draft, it sits under Multi-Advertiser Ads in the ad-level settings behind the extra settings link, and it is off by default.

You enter a share of the campaign or ad set budget to dedicate to the ad and choose how long the push lasts, and the tooltip describes pushing delivery for up to 7 days or until the ad set ends, after which the ad returns to competing for the ordinary budget.

Per the same account, a test pushing 10% to each of five ads found Meta spending roughly in line with that over the first $75 or so, and Loomer adds that he cannot tell whether the requested share is a floor the system is allowed to exceed.

Two caveats before anyone builds a process on it. AdBeacon’s account of the same feature reports the seven days as a recommendation rather than the limit, with the real ceiling nearer 30 days and a one-day minimum, and says Meta advises against pushing several ads in one ad set at the same time.

And Loomer reports no official Meta documentation for any of it, so treat the specifics as current behaviour rather than a published rule. AdMake AI’s July 2026 Meta changelog puts the rollout at nearly all accounts by 10 July 2026, which is a single source for a date worth checking against your own account.

If a new ad is not spending	What it does	What it costs you
Wait	sometimes delivery arrives on its own, often it does not	days, with no guarantee at the end of them
Push Delivery to this ad	dedicates a stated share of budget for a set window	the delivery the other ads in the set would have had
Publish it with different settings	makes it less directly comparable to the incumbent ad	a confounded test, and still no guarantee of impressions
Give it its own ad set	removes the incumbent from the comparison entirely	a second ad set to fund through its own learning phase

Pick by what you are trying to learn.

Per AdBeacon, Push Delivery is built for a published ad that needs exposure rather than for comparing variants, and the creative testing tool remains the instrument for a comparison. Loomer’s earlier guide to one ad taking the whole budget sets out the pre-2026 options, publishing with different settings or moving the ad into its own campaign and ad set, and is blunt that neither obliges Meta to serve it.

How long does it take to set up a Facebook ad account from scratch?

Setting up a Facebook ad account from scratch takes about an afternoon to get the first ad submitted, and roughly three to four weeks before the account behaves like an established one. Creating the account itself is a form.

What takes the time is everything attached to it: a payment method, a pixel that fires, a verified domain, in some categories a verified business, and the delivery history the account has not earned yet.

Separate the steps that hold up your first ad from the steps that hold up your first scale, because founders usually discover them in the wrong order.

Step	How long it takes	What it holds up
Create the business portfolio and the ad account	minutes	nothing
Add a payment method	minutes	all delivery, since an ad set with no payment method never serves
Install the pixel and confirm events arrive	an hour or two of developer time	conversion optimisation, though not traffic campaigns
Verify the domain	instant by meta tag or file upload, up to 72 hours by DNS record	event priority configuration and conversion reporting
Verify the business, where your category requires it	Meta allows itself up to 14 business days	restricted categories and higher access tiers
Ad review	most ads clear within 24 hours	the first impression
Build delivery history	two to four weeks of uninterrupted spend	how fast you can raise the budget

Domain verification is the step that catches people out, because the fast route and the slow route look identical in the interface.

Meta accepts a meta tag in the homepage head, an uploaded HTML file or a DNS TXT record, and only the last of those waits on anything outside Meta: DNS changes usually propagate in minutes but can take up to 72 hours, according to Buffer’s guide to setting up Facebook domain verification.

If you can edit the site’s HTML, use the meta tag and skip the wait entirely.

Which setup steps can hold up a launch by days rather than minutes?

Two setup steps hold up a launch by days rather than minutes, and both are easy to discover late: business verification, and the delivery history a new account has not built yet. Everything else in the table above is an afternoon’s work.

Business verification is the step that can genuinely cost you a fortnight. Meta’s own help page on verifying your business in Meta Business Suite states that a decision on a submission may take up to 14 business days.

Clean submissions are frequently decided in a few days, but the ceiling is real and there is nothing you can do from your side to move it, so submit on day one rather than finding out in week three that a category you sell into requires it.

The last row of the table is the one with no published rule behind it. A new ad account has no payment history and no delivery history, so opening at your target budget on day one asks the system to trust a number it cannot price.

Raise it the way you would after any significant edit, by no more than 20% to 30% every three to five days, which is the convention in Cometly’s 2026 learning phase guidance.

Meta does not publish a starting spend cap for new accounts, and the figures quoted in warm-up guides are practitioner estimates rather than a documented tier, so treat the ramp as a precaution rather than a rule you can look up.

Read all of this as an offset to the rest of the article rather than an addition to it.

The timeline in the sections below starts when an ad is submitted for review, so a brand starting from nothing should add about a week to the front of it: a day or two of setup, then a few days of small spend while the account establishes that its payments clear.

The first honest read moves from day 14 to somewhere near day 21.

What is the learning phase and how long does it last?

The learning phase is the period when Meta’s delivery system is still testing who to show an ad to, and it ends when the ad set has recorded roughly 50 optimisation events in a rolling 7 days. That threshold is the published one, explained in AdLibrary’s 2026 guide to the 50-event rule, and it applies per ad set rather than per campaign or per ad.

Fifty events in seven days is an arithmetic requirement before it is a patience requirement. A brand optimising for purchases at a $40 target cost per acquisition needs about $2,000 of spend in a week, or roughly $286 a day, to reach it.

An ad set funded below that level does not learn more slowly. It sits in Learning Limited indefinitely, which Meta labels clearly and most advertisers read as a status rather than a warning.

We work the budget arithmetic out in full in how much should you spend on Facebook ads per day.

The other way accounts never leave the learning phase is editing. Per Meta’s own page on significant edits, budget changes above roughly 20%, changes to targeting, changes to the optimisation event, changes to bid strategy and changes to the attribution window each reset the ad set into learning, which restarts the seven days.

An advertiser adjusting an ad set every morning has an account that is permanently in exploration and never in exploitation.

Optimising for a cheaper event is the legitimate escape hatch when volume is genuinely out of reach. Fifty add-to-carts is far easier to reach than fifty purchases, and a well-correlated upper-funnel event gives the system enough signal to work with. The cost is that you are now optimising for a proxy, so watch the relationship between the proxy and revenue rather than assuming it holds.

What counts as a significant edit, and how does it affect the learning phase?

A significant edit is any change Meta treats as making the ad set a different thing to learn, and the price of one is the seven days of the learning phase over again. Meta’s own page on significant edits covers changes to targeting, creative, the optimisation event, the bid strategy and the attribution window, plus budget changes above roughly 20%.

It also covers the one most advertisers do not expect, which is adding a new ad to an ad set that is already running.

The wording of the exit condition is what makes editing expensive.

Meta frames it as roughly 50 optimisation events since the ad set’s last significant edit rather than 50 in a calendar week, and both Apogee’s 2026 breakdown of what resets learning and what does not and Scalemate’s version of the same list read the clause that way.

The counter is anchored to your last edit, not to Monday. An ad set sitting at 35 of the 50 events it needs loses those 35 when the creative changes, a point Adlio’s note on adding ads to a live ad set makes with the same arithmetic.

Change	Significant edit	Where that comes from
Targeting, geography, age or gender	yes	Meta’s significant edits page
Creative swap on a live ad	yes	Meta’s significant edits page
Optimisation event, bid strategy or attribution window	yes	Meta’s significant edits page
Budget change above roughly 20% at once	yes, threshold contested	practitioner consensus, not a published figure
Adding a new ad to a running ad set	on Meta’s list, reports vary	Meta’s list, as read by Apogee and Adlio
Pausing the ad set for 7 days or more	yes, a full restart	MagicBrief’s learning phase guide
Rewriting the caption or headline	not listed by Meta	no published rule either way
Replacing a custom audience’s members with the Replace Users API	no, a documented exception	Meta’s custom audience guide

Two rows carry qualifiers for a reason. The 20% budget threshold, repeated in Scalemate’s 2026 list of what resets learning and in most others, is a practitioner convention rather than a figure anyone can point to in Meta’s documentation, so treat it as a working ceiling rather than a line you can audit against.

Adding an ad is where practitioner reports diverge, since some accounts appear to absorb a new ad without visibly re-entering learning. Plan as though it restarts, because being wrong that way costs a week and being wrong the other way costs a quarter.

The consequence is to batch rather than abstain. Five changes on five consecutive days buy five restarts and about 35 days of unstable delivery, where the same five made in one sitting buy one.

Keep a list during the week, make the whole list at once, then read the ad set’s own progress in the Delivery column of Ads Manager, which shows the event count against 50 while calibration is still running.

One claim circulating in 2026 guidance is worth refusing: that the Andromeda update tightened what counts as a significant edit. No Meta source says so. Andromeda changed which ads get retrieved, which is a creative question rather than a delivery one, and it is covered in how many ad creatives to test.

How do you check whether an ad set is still learning, and when its clock restarted?

You can check whether an ad set is still learning, and read the exact moment its clock restarted, from a field Meta documents rather than from the label in the Delivery column.

The ad set carries a learning_stage_info object, described in Meta’s ad set reference as “info about whether the ranking or delivery system is still learning for this ad set”, and the learning stage reference defines what is inside it.

Field	What Meta’s reference says it holds
status	LEARNING, SUCCESS, or FAIL, where FAIL means the ad set “isn’t generating enough results to exit the learning phase”
conversions	conversions “since the time of its last significant edit during the learning phase”
last_sig_edit_ts	“timestamp of the last significant edit that caused ad set to reenter the learning phase”
dynamic_lp_conversions_threshold	“new conversions threshold for dynamic learning phase status”
dynamic_lp_days_threshold	“day to exit for dynamic learning phase”

Two of those settle arguments this page has had to leave to practitioner sources. The definition of conversions says in Meta’s own words that the count runs from the last significant edit rather than from the start of a calendar week, which is the reading the table above takes from Apogee and Adlio.

And last_sig_edit_ts makes the disputed question answerable in your own account: instead of arguing about whether adding an ad to a live ad set resets learning, add one and see whether that timestamp moves.

The audience row in the table above is the one case Meta’s own developer guide answers outright, and the answer is asymmetric.

The Replace Users API “does not reset your ad set’s learning phase when an audience is part of active ad sets”, per that guide, “unlike POST or DELETE API calls to the /<CUSTOM_AUDIENCE_ID>/users endpoint”.

So refreshing a retargeting list is free if the list is replaced wholesale and costly if it is patched member by member, and that choice is made inside whatever tool syncs the list rather than inside Ads Manager.

One trap sits inside the same object. Per the reference, conversions returns zero once an ad set has exited the learning phase successfully, so a zero means either nothing has happened yet or the phase is over, and the number only means what you think it means after you have read status.

The two dynamic fields are the more interesting find, because they say the famous threshold is a default rather than a constant.

Meta documents a conversions threshold and a days-to-exit value held per ad set under a dynamic learning phase status, which means an ad set can be working to its own exit conditions rather than to 50 events in 7 days.

The reference names the fields and does not publish the values or say how they are set, so treat 50 in 7 as the planning figure and the fields as the reason a particular ad set may not behave like it.

Read the attribution line in that reference with care, which is a useful warning about documentation generally. It states that the attribution window defaults to 1-day view and 28-day click, and that is not the default Ads Manager now applies or the one the next section describes, so the reference is accurate about the learning phase and out of date about the window around it.

How long before the results are worth reading?

Results become worth reading around day 14, and the reason is attribution rather than delivery. Meta’s default attribution setting is 7-day click plus 1-day view, which means a purchase counts if the buyer clicked within the previous seven days or saw the ad without clicking in the previous 24 hours.

A click on day 6 that converts on day 12 is credited back to day 6, so the reported performance of your first week keeps improving for a week after that week ends.

That window narrowed on 12 January 2026, when Meta removed the 7-day view and 28-day view options from the Ads Insights API, according to Dataslayer’s reporting on the change, leaving 7-day click and 1-day view as the reporting defaults.

The removed windows return empty data rather than an error, so a dashboard still asking for them shows blanks and says nothing about it. Longer consideration cycles are now systematically under-credited inside Meta’s own numbers, which matters for anything sold at a price people sleep on.

Day	What has happened	What you can read	What you should not do
1 to 3	ad approved, delivery exploratory	approval, spend pacing, whether the pixel fires	judge the creative, pause anything
4 to 7	ad set approaching 50 events	direction of click-through rate and CPM	change budget or optimisation event
7 to 14	learning phase complete, attribution catching up	first honest cost per acquisition	scale hard off one week of data
14 to 28	enough volume behind the figure	verdict on the creative and the audience	wait longer while the ad fatigues
21 onward	creative decay begins on fast placements	whether the next batch is ready	expect the same asset to keep working

Stack the two clocks and the timeline falls out. Day 7 is when delivery stabilises.

Day 14 is when the attribution window has caught up with the stable period.

Day 21 to 28 is when there is enough volume behind the figure to act on it with any confidence, which is also roughly the point at which the creative itself starts to decay: AdLibrary’s 2026 fatigue benchmarks put the burn window on Reels-heavy placements at two to three weeks.

The overlap is uncomfortable and worth saying plainly. On a fast-fatiguing placement, the window in which an ad is both stable and fresh can be as short as a week, which is why creative volume and patience are the same problem rather than competing ones.

Why do results get worse in week two?

Results commonly dip in week two because the ad set has left the learning phase and moved from exploration to exploitation, and the first honest reading of the ad is often worse than the flattering one from week one.

Early delivery frequently finds the warmest, cheapest slice of the audience first: people who already know the brand, are already in market, or sit in a retargeting pool that leaked into the campaign. Those convert cheaply and run out.

The second cause is frequency. As delivery concentrates on the segment Meta rates as most likely to convert, the same people see the ad repeatedly, engagement falls and the effective price rises. AdLibrary’s 2026 benchmarks put the decline threshold at a weekly prospecting frequency of 2.5, with a sharper drop past 4.0, and find that click-through rate and CPM diverge sharply once frequency passes 3.0.

The third is simple regression. A week one cost per acquisition built on nine conversions is not a stable estimate, and the week two figure built on thirty is closer to the truth. Nothing broke. The sample got bigger.

What separates these from an actual failure is the direction of the underlying metrics. Falling click-through rate with rising frequency is fatigue and calls for new creative.

Rising CPM with steady click-through rate is auction pressure, which we cover in why is my Facebook ads CPM so high. A collapse in delivery rather than a decay in efficiency is usually an edit, a rejection or a spending limit, not a market condition.

What should you do while the ads are learning?

While the ads are learning you should be producing the next round of creative, because that is the constraint that decides whether month two beats month one.

A typical win rate on new concepts is low enough that volume, not optimisation, produces winners, and the accounts that improve fastest are the ones that entered week three with a fresh batch ready rather than a set of tweaks to the batch that is fatiguing.

Leave the ad sets alone in the meantime. Every meaningful edit restarts the seven-day clock, so the instinct to improve things daily produces an account that never stabilises long enough to be measured. Set the budget you intend to run, confirm it is high enough to clear 50 events in 7 days, and then do not touch it.

Verify the measurement instead, since that costs nothing and catches the failures that look like creative problems. Confirm the pixel fires on purchase, that the conversion event matches what the campaign optimises for, that the landing page loads quickly on a phone, and that your own analytics roughly agrees with Ads Manager on order count. A broken checkout and a bad ad look identical in the dashboard.

Then set the review date in advance and hold yourself to it. Deciding on day 14 that you will judge the test on day 14 is much easier than deciding on day 3 that you will wait until day 14.

So what is the short version?

The short version is that Facebook ads deliver within about 24 hours, stabilise around day 7 once the ad set clears roughly 50 optimisation events in a rolling week, and become readable around day 14 once the default 7-day click attribution window has caught up. A confident verdict needs 21 to 28 days.

Two things break that timeline more often than anything else. Underfunding the ad set, so it never reaches the 50 events in 7 days that Meta requires according to AdLibrary’s 2026 guide, leaving it in Learning Limited permanently, and editing the ad set, since budget changes above roughly 20% and changes to the optimisation event, bid strategy or attribution window each restart the seven days.

Expect week two to look worse than week one. That is usually the sample growing and the cheap warm audience running out, not a decline. The account is in real trouble when click-through rate falls while frequency climbs, which is creative fatigue and needs new assets rather than more time.

All of those clocks start when an ad gets delivery rather than when it goes live, which matters most for a new ad added to a working ad set.

Budget sits at the ad set and there is no ad-level budget to set, per Terrads’ reading of the Marketing API, so a new creative can collect a few hundred impressions and no verdict.

Check the spend share first, and use Push Delivery to this ad or a separate ad set when the share is nil.

One limit worth stating. These are delivery timelines, not profitability timelines. Reaching a cost per acquisition that clears your break-even normally takes several creative cycles, because the first batch rarely contains the winner. Plan the budget over six to eight weeks and judge the channel on that, rather than deciding whether paid social works for you on the strength of one flight of ads.