In a Peak Season 2026 post dated October 21, Meta tells advertisers that holiday budgets are mostly lost by changing campaign settings too often, not by misreading the auction. The headline number: ad sets below roughly 50 conversions a week ran about 28% more expensive per purchase in Meta's own analysis.
What Meta published
The Meta for Business post, titled "Peak Season 2026: You Don't Beat The Auction. You Brief It.", frames campaign settings as a brief to the delivery system. Every time you rewrite that brief, the ad set re-enters learning, and the cost of that reset never shows up as a line item in Ads Manager.
Meta restates the familiar learning threshold of about 50 conversions per ad set per week and attaches a cost to missing it: around 28% more per purchase. PPC Land, which reported the post on October 25, notes that Meta did not publish a method, sample size or time period for that figure, so treat it as directional rather than a benchmark.
The ad set ceiling formula
The most practical part of the guidance is a simple sizing rule: weekly budget divided by (cost per result x 50) gives the maximum number of ad sets an account can feed properly. Using PPC Land's arithmetic, a $20,000 weekly budget at a $40 cost per purchase supports about 10 ad sets; at $100 per purchase it supports only 4.
- Budget: weekly spend across the campaign
- Cost per result: your realistic CPA for the optimisation event
- 50: Meta's weekly learning threshold per ad set
The three decisions Meta wants locked in
Meta groups the brief into three choices: what to optimise for (conversions, value, profit in beta, or predicted lifetime value), how to bid (spend-based, goal-based or manual, plus value rules for priority segments), and how results are counted (standard, incremental or custom attribution in beta). Its advice is to settle these, batch any edits so one learning phase replaces several, and validate the setup with an A/B test or Conversion Lift study held still for about a month.
Supporting figures in Meta's companion optimisation guide are mostly from 2024 and 2025 studies, which PPC Land flags as a limitation. Notably, the guide still cites older GEM numbers rather than the 18.9% conversion lift Meta published on October 13.
What to do before Black Friday
Run the ceiling formula on every account now. If you have more ad sets than the budget can feed, consolidate before BFCM rather than during it. Send currency and value with every purchase event so value optimisation has something to work with, and freeze structural edits once peak traffic starts. If you are scaling across several accounts for the season, Power Ads can supply agency accounts on corporate cards with a 24/7 operator, so a single restriction does not stall a whole launch.
What it means for you
- Max ad sets = weekly budget / (CPA x 50). Run it on every account this week.
- Treat the 28% figure as directional; Meta has not published how it was measured.
- Batch edits so you trigger one learning phase, not several, during Q4.
- Pass value and currency on every purchase event before switching to value bidding.
