Facebook Ads Budgets: Getting the Structure Right Before the Numbers

The easiest mistake to make is also the hardest to notice

Ads Manager makes it very easy to change a number and watch what happens next. That ease is where a lot of Malaysian SME ad budgets quietly go to waste — not through bad numbers, but through good numbers interrupted before they had a chance to become reliable. Understanding why requires understanding what Meta's delivery system is actually doing in the days after a campaign launches, or after any meaningful change is made to it.

What the learning phase is measuring

When an ad set is new, or has just been changed in a way Meta considers significant, it enters what Meta calls the learning phase. Meta's own help centre describes it directly: "the delivery system is exploring the best way to deliver your ad set based on the optimization event you chose for your ad sets. This can lead to less stable performance and a higher cost per result" (Meta Business Help Center, Significant Edits and Learning Phase). The status shown inside Ads Manager during this period is described the same way: "the delivery system is exploring and optimizing the best way to deliver your ads. Performance is less stable during this time period and CPAs are usually worse" (Meta Business Help Center, View campaign, ad set or ad delivery status).

In plain terms: the system needs a batch of real outcomes to work out which audience segments, placements, and delivery patterns are actually producing the result you asked it to optimise for. Meta's guidance on when an ad set struggles to complete this process is specific — an ad set becomes "learning limited" when it is "unlikely to receive about 50 optimization events in the week after your last significant edit" (Meta Business Help Center, About Learning Limited). Interrupting that process before it gathers enough events doesn't just slow things down — it restarts the count.

What actually counts as a significant edit

This is the part most often guessed at rather than checked. Meta's own page lists exactly what always counts as a significant edit:

• Any change to targeting
• Any change to ad creative
• Any change to optimization event
• Adding a new ad to your ad set
• Pausing your ad set for 7 days or longer (the ad set reenters the learning phase once you unpause the ad set)
• Changing bid strategy

There's a second, less commonly quoted list of changes that "may or may not be significant, depending on the magnitude of the change": ad set spending limit amount, bid control or cost-per-result or ROAS goal amount, and budget amount itself. That distinction matters. A great deal of advice circulating in Malaysian SME ad groups treats every budget nudge as an automatic reset. Meta's own documentation is more specific than that — a small, incremental adjustment is a different action from doubling a budget overnight, even though both are technically edits to the same field.

Campaign-level versus ad-set-level budgeting

Meta's delivery system can be given a single budget to manage across every ad set in a campaign — what Meta calls Advantage+ campaign budget — or each ad set can be given its own fixed amount. The trade-off is about where the decision-making sits: campaign-level budgeting lets Meta shift spend toward whichever ad set inside it is currently performing better; ad-set-level budgeting keeps a guaranteed amount behind each specific audience regardless of how it's performing relative to the others.

Meta's own documentation flags one interaction worth knowing before choosing: when a campaign is running on Advantage+ campaign budget, "switching your campaign bid strategy might cause multiple ad sets within the campaign to reenter the learning phase" — a single edit affecting several ad sets at once rather than one.

This decision matters more, not less, for a small total budget. Splitting a modest daily budget across many ad sets means each one needs to gather roughly its own share of that 50-events-a-week threshold before it can stabilise — and a budget too thin to produce that volume anywhere is a documented cause of an ad set staying learning limited indefinitely. Meta's own troubleshooting guidance for an ad set stuck in this state is direct: "Combine ads sets and campaigns... Combining ad sets and campaigns will help you get the results you need faster, which means you'll see stable results sooner," and, separately, "Raise your budget. If your budget is too low to receive enough optimization events, the ad set is unlikely to exit the learning phase."

Testing and scaling are different jobs, not two speeds of the same job

Testing means deliberately comparing distinct variables — a different creative, a different audience, a different offer — in separate ad sets, and accepting the instability and higher cost of the learning phase as the price of finding out what actually works. Scaling means increasing budget behind something that already has a track record, and doing it in a way that respects the same magnitude principle Meta applies to what counts as significant: an incremental increase is a smaller disruption than an overnight jump.

Confusion tends to show up when both happen inside the same ad set at the same time — a new creative added while the budget is also being raised while the audience is also being tweaked. Each of those, on its own, is enough to trigger a fresh learning phase under Meta's own criteria; stacked together, there's no way to tell afterwards which change actually caused whatever happened to performance.

Telling a budget problem from a creative or offer problem

Before assuming a disappointing result is a budget problem, it's worth checking where in the sequence things are actually breaking down:

• Healthy reach, weak click-through — usually a creative or message problem. More budget buys more impressions of the same underperforming ad; it doesn't fix the ad.
• Healthy clicks, weak conversions — usually a landing page, offer, or follow-up problem, happening after the ad has already done its job.
• An ad set that never exits learning despite weeks of running — this is the case Meta's own documentation actually describes as a budget or bid problem: not enough volume reaching the ad set to gather the results the delivery system needs.

More budget cannot repair a weak offer or a slow follow-up process. It buys a weak offer more impressions, faster, and usually at a worse cost per result than a smaller, more patient test would have shown.

A grounded starting posture

There's no universal daily figure that fits every Malaysian SME, and any number offered without knowing your industry, offer, and margins would be a guess dressed up as advice. What Meta's own guidance does say plainly is to "set a budget large enough to get enough total results and avoid frequent budget changes (which can cause an ad set to re-enter the learning phase)" (Meta Business Help Center, About the Learning Phase). Structure comes before the number: decide whether you're testing or scaling, set the budget accordingly, and leave it stable long enough to actually find out.

Where JagaWeb fits

Ad budget decisions sit with the business or its media buyer, not with us — we don't manage ad accounts, and any pricing we do quote for other work is published on jagaweb.my. Where we can help is in ruling out the other side of the equation: if a campaign is being blamed for a poor result that's actually happening on the landing page or in the offer itself, JagaWeb's Essential System Review (RM1,500, currently RM999 until 16 September 2026) looks specifically at that side — an honest read of what's actually happening after the click, not a verdict on the ad budget. Reach us at sales@jagaweb.my or WhatsApp jagaweb.my.