TL;DR
- Meta for Business published a peak-season best-practice post on 21 October 2026 saying ad sets under roughly 50 conversions a week ran about 28% more expensive per purchase in its analysis.
- The post publishes a formula: total weekly budget divided by (cost per result x 50) equals the most ad sets a budget can support; THB 35,000 a week at THB 350 CPA supports 2.
- Meta gives no sample size, period, region or vertical for the 28% figure and does not show the source of its 42% / 4.2% shopper statistic.
- Meta recommends one grouped change before peak, then holding the campaign still and measuring with an A/B test or Conversion Lift study over October.
Meta for Business published a Best Practice post on 21 October 2026 titled "Peak Season 2026: You Don't Beat The Auction. You brief it." Its most concrete claim is that ad sets receiving fewer than roughly 50 conversions a week cost more to run. In Meta's words: "Give each ad set enough volume to learn from. Roughly fifty conversions a week. Below that the system is guessing. In our analysis, ad sets under that threshold ran about 28% more expensive per purchase." The post also publishes a formula for how many ad sets an account can afford: "Total weekly budget ÷ (cost per result × 50) = the most ad sets your budget can support." Meta adds that "Most accounts are running more than that." The post gives no sample size, time period, region or vertical for the 28% figure, and it is tagged as advice for small businesses, not as a research release.
What Meta published on 21 October 2026
The post sits in the Best Practice section of Meta for Business news and carries the tags "Best practices" and "Small business". It is written as a peak-season essay rather than a product announcement, which matters for how much weight to put on it. Nothing in it changes how Ads Manager works.
The essay opens with a shopper statistic: "Last peak season, 42% of shoppers said they planned to spend less. They spent 4.2% more." That line carries a footnote marker, but the page does not show the footnote's source, so the survey behind it cannot be checked from the post itself. Meta uses the figure to argue that two kinds of buyers, the cautious planner and the confident buyer, sit in the same audience and see the same ads, and that an account "treated them as one person, because that's what it was told to do."
From there the post reframes every campaign as a brief to the auction. Meta writes: "The optimisation event is the ask. The budget is the constraint. The bid strategy is the definition of success. The attribution setting is how you agree to be judged." The point of the framing is that the auction, unlike a human colleague, never asks a clarifying question. "It reads what you wrote, decides you meant it, and goes and buys it."
The 50-conversion threshold and the 28% claim
The threshold itself is not new. Meta's learning phase help page says ad sets usually exit learning "after about 50 results in the week after the ad set's last significant edit." What is new in this post is a first-party cost figure attached to it: ad sets under the threshold "ran about 28% more expensive per purchase" in Meta's analysis.
Three things about that number deserve a plain statement. First, Meta says "In our analysis" and nothing more. There is no sample size, no date range, no list of countries, no vertical, no campaign objective and no split between Advantage+ and manual setups. Second, the post does not say whether the comparison controlled for budget. Ad sets that never reach 50 conversions a week are, almost by definition, smaller ad sets, and smaller ad sets can have higher costs per purchase for reasons unrelated to the learning phase. Third, the figure is "per purchase", which suggests the analysis looked at purchase-optimised ad sets, but the post does not say so explicitly. A Thai advertiser optimising for leads or messages should treat the 28% as an indicative first-party figure, not as a benchmark for their own account.
What the post is confident about is the direction: fewer, larger ad sets learn faster and cost less per result than many small ones.
The ad set budget formula, with a Thai example
The practical tool in the post is a single line of arithmetic. Meta writes: "Total weekly budget ÷ (cost per result × 50) = the most ad sets your budget can support." The logic follows directly from the threshold: if each ad set needs 50 results a week, and each result costs a known amount, then the weekly budget can only feed a certain number of ad sets to that level.
Worked through in baht, the ad set budget formula looks like this. An account spending THB 35,000 a week at a cost per purchase of THB 350 can support 35,000 ÷ (350 × 50) = 2 ad sets. An account spending THB 100,000 a week at THB 400 per purchase supports 100,000 ÷ (400 × 50) = 5. An account spending THB 14,000 a week at THB 350 per purchase supports 14,000 ÷ 17,500 = 0.8, which means it cannot bring even one purchase-optimised ad set to the threshold and may be better served by a cheaper optimisation event or a single consolidated ad set.
The table below shows how the formula behaves at a range of illustrative spend levels. The budget and cost-per-result inputs are illustrative; the formula and the 50-conversion threshold are Meta's.
| Weekly budget (THB) | Cost per purchase (THB) | Ad sets the budget supports |
|---|---|---|
| 14,000 | 350 | 0.8 (below one full ad set) |
| 35,000 | 350 | 2 |
| 70,000 | 500 | 2.8 (round down to 2) |
| 100,000 | 400 | 5 |
| 250,000 | 600 | 8.3 (round down to 8) |
Meta's own reading of the output is blunt: "Most accounts are running more than that. Which means most accounts have written a brief with more objectives than they have money to pursue." The post does not say what share of accounts exceed their supported count, only that most do.
Why Meta says every edit costs twice
A second argument in the post is about the cost of changing campaigns mid-season. Meta says an edit is paid for twice: "Once for the learning phase you restarted, while the system works out what the new instruction means. And once for the old instruction, which was live and spending at full price the whole time you had it wrong." Only the first of those appears in the account as a learning-phase warning. The second, the money spent while a mis-set campaign ran, "appears nowhere."
Meta's conclusion from this is behavioural. Advertisers "fear the visible cost and ignore the invisible one", leaving a poor brief running because changing it feels risky, and tinkering with a good one because activity feels like work. The post also notes that the auction is "rebriefed by every edit you make. Including all the ones you don't think of as edits." It does not list which edits trigger a learning phase to restart.
The three decisions Meta calls "the brief"
The post reduces a campaign brief to three settings and gives a one-line definition of each.
What to optimise for
Meta frames this as volume versus value: "Volume tells the auction to find you more cautious planners. Value tells it to find you more confident buyers. Same budget, same audience, different people at the checkout." It says most accounts "optimise for what's easy to count, not for what the business gets paid on."
How hard to chase
This is bid strategy. Meta's wording is that this is "where you admit that some customers are worth more than others" and where an advertiser bids up for valuable segments "instead of treating every buyer as the same buyer." The post does not name specific bid strategies.
How it is counted
Meta describes three attribution options: "Standard attribution counts what the ads touched. Incremental attribution counts what the ads added. Custom attribution counts it the way your business already counts it." It warns that if the number the account optimises against is not the one the finance team uses, "you're reading a different campaign from the one you're running." The post calls attribution "the decision most likely to be inherited rather than chosen."
Meta also says the three are "a blend" rather than a single choice, and that the brief decides which creative wins: "Brief the auction for volume and it finds the ad that brings cheap orders. Brief it for value and it finds the ad that brings big ones."
What Meta recommends doing this week
The post ends with a short list of actions, split by how much time an advertiser has.
- If one thing: review whether the largest campaign is optimising for the outcome the business values. If not, change it, then "Leave it alone." Measure it against an A/B test or a Conversion Lift study rather than against last month, and "Give it the month of October."
- If longer: run the ad set formula across the account, since "Every ad set is a separate instruction." Rewrite the brief in three lines (what to optimise for, how hard to chase, how to count). Group all changes into one edit: "Take one learning phase instead of five. Then hold it still."
The closing line captures the stance: "Peak season doesn't reward the advertiser who watches the dashboard most closely. It rewards the one who wrote the clearest instruction and then had the nerve to leave it alone."
What the post does not say
Because this is a best-practice essay, the gaps matter as much as the claims. The post does not disclose the method behind the 28% figure. It does not say whether the 42% and 4.2% shopper figures come from a Meta survey or a third party. It does not define "peak season" by date or market. It does not say which campaign types the 50-conversion threshold was measured on, and it does not discuss accounts whose weekly budget cannot reach the threshold at all, beyond implying they should consolidate. It also does not mention Advantage+ shopping campaigns, Andromeda or any specific ranking model, so the advice should be read as general rather than tied to a product change. Treat all of it as Meta's recommended practice, sourced from Meta, rather than as an independently verified result.
What this means for Thai marketers
The post gives Thai advertisers a Meta-sourced number to justify consolidating ad sets before peak season, and a formula that can be run on any account in minutes. An account that fragments a THB 35,000 weekly budget across eight audience-based ad sets is, by Meta's arithmetic, running six more instructions than it can fund. Meta's advice is to make one grouped change before the peak, then hold it for a measurable period. For teams that manage Facebook ads for Thai audiences, the three-line brief is a useful checklist to put in front of a client before the Q4 budget is released.
FAQ
Does Meta say ad sets under 50 conversions a week always cost 28% more?
No. Meta says that "In our analysis, ad sets under that threshold ran about 28% more expensive per purchase", with no sample size, period, region or vertical disclosed. It is a directional first-party figure, not a guarantee for any individual account.
What is Meta's ad set budget formula?
Meta's formula is "Total weekly budget ÷ (cost per result × 50) = the most ad sets your budget can support." For example, THB 35,000 a week at THB 350 per purchase supports 2 ad sets.
Is the 50-conversions-a-week threshold new?
No. Meta's learning phase help page says ad sets usually exit learning after about 50 results in the week after their last significant edit. What is new in the 21 October 2026 post is the 28% cost figure and the published formula built on the threshold.
Does this apply to Thai advertisers?
The post states no regional restriction. Whether the 28% penalty matches a Thai account is not stated by Meta.
What does Meta recommend doing before peak season?
Meta recommends checking that the largest campaign optimises for the outcome the business values, running the ad set formula, rewriting the brief in three lines, grouping all changes into one learning phase and then leaving the campaign alone for a measurable period such as the month of October.
If a Q4 Meta plan needs a second pair of eyes on ad set structure, optimisation event and attribution before budgets go live, Relevant Audience's Facebook ads team in Bangkok can review the account against the three-line brief described above.
