Meta on October 21, 2026 published a peak season best-practice post on its Meta for Business site arguing that advertisers lose money during the holiday quarter not because they misjudge the auction but because they keep rewriting the instructions they give it. The post, titled "Peak Season 2026: You Don't Beat The Auction. You Brief It.", puts a number on one of those instructions: according to Meta, ad sets that receive fewer than roughly 50 conversions a week ran about 28% more expensive per purchase in its own analysis. It also offers a formula that sets a ceiling on how many ad sets a budget can support, and asks advertisers to pick one campaign, change its optimisation if needed, and then leave it untouched while an experiment measures it.

In Short

Meta wrote a guide saying its ad system does exactly what advertisers tell it to, so the settings they choose (what to aim for, how much to bid, how to count sales) act like instructions to a worker who never asks questions. It matters for any business spending heavily on Facebook and Instagram ads before Black Friday, because Meta says splitting money across too many small ad groups, or changing them every week, makes each sale cost more. Meta's suggested fix is fewer ad groups, a goal tied to what the business actually earns, and fewer edits, though most of the evidence for it comes from Meta's own tests.

What the post says

The post opens with a statistic. "Last peak season, 42% of shoppers said they planned to spend less. They spent 4.2% more," according to Meta. The figure carries a footnote marker, but the post does not show the footnote text or link to a study, so neither the survey behind the 42% nor the spending measure behind the 4.2% is identified.

From there, Meta builds its argument around a single analogy. Every campaign setting, the post says, is a brief. "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," according to the post.

The difference between briefing a person and briefing an ad system, in Meta's telling, is that the system never pushes back. "The auction never asks. It accepts everything. It reads what you wrote, decides you meant it, and goes and buys it." Results are therefore the only feedback, and they arrive after the money has been spent.

Meta then describes what it calls a flawed feedback loop. Performance dips, the advertiser changes something, performance recovers, and the change gets the credit. "You've learned nothing. You'll do the same thing again next month," the post states. Results, according to Meta, move with "the season, the competition, a creative that has fatigued, someone else's campaign, the weather," and the campaign brief is only one input among dozens - but the only one advertisers control.

Two costs, one visible

The most specific claim in the narrative section concerns the cost of edits. Changing a campaign, according to Meta, costs twice: once for the restarted learning phase, and once for the period the old, wrong instruction was running "at full price." An illustration in the post labels the first cost "What you see" and the second "What you paid," the latter described as "Three weeks of the wrong brief, at full price." According to the caption, "Only one of these two costs appears anywhere in your account."

The post argues that this asymmetry produces two opposite errors. Advertisers "leave a bad brief running because changing it feels risky. Then they tinker with a good one because tinkering feels like work." No measurement of how often either happens is given.

Meta also names two places where its own analogy fails. A brief to a person is handed over once, whereas "the auction is rebriefed by every edit you make." And a poor brief to a person starts an argument, while a poor brief to the auction "produces a perfectly executed wrong answer, delivered on time and inside budget."

Three decisions

At the centre of the post is a graphic headed "Achieve the best outcomes for your business," which divides campaign settings into three dimensions. The KPI column lists conversion optimisation, value optimisation, profit (beta) and predicted lifetime value. The bidding column lists spend-based, goal-based and manual bid strategies, plus value rules for priority segments. The attribution column lists standard attribution, incremental attribution and custom attribution (beta).

Meta frames each dimension as a question. On the first - what the campaign optimises for - the post claims most accounts "optimise for what's easy to count, not for what the business gets paid on." Volume-based optimisation, according to Meta, finds "more cautious planners," while value-based optimisation finds "more confident buyers." The second question, how hard the system should chase, is where advertisers "admit that some customers are worth more than others" by bidding up for chosen segments.

The third question concerns attribution, which Meta calls "the decision most likely to be inherited rather than chosen." The post distinguishes the three options plainly: "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." The consequence, according to Meta, is that an advertiser whose optimisation target differs from the measure its finance team uses is "reading a different campaign from the one you're running."

The post adds that the three choices are a blend rather than a single selection, and that they determine 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."

The 50-conversion rule and the formula

The section with the most direct operational consequence is short. "Give each ad set enough volume to learn from. Roughly fifty conversions a week. Below that the system is guessing," the post states. "In our analysis, ad sets under that threshold ran about 28% more expensive per purchase."

Meta does not describe the analysis. The post gives no sample size, time period, region, vertical or campaign type, and it does not say whether the comparison controls for budget, which is the obvious confounder: ad sets with smaller budgets are both less likely to reach 50 conversions and more likely to have different cost structures for reasons unrelated to learning.

The threshold itself is long-standing. Meta's Business Help Centre generally expects about 50 optimisation events within seven days of an ad set's last significant edit before it exits learning, as PPC Land's explainer on the learning phase sets out; the same piece notes vendor reports, without supporting Meta documentation, that the threshold had dropped to 10 events for some campaign types in 2024 and that Advantage+ shopping campaigns could exit learning after 25 purchases in 2026. Meta's peak season post restates the round figure of fifty without addressing those reports.

The formula that follows is presented in bold in the post:

Total weekly budget ÷ (cost per result × 50) = the most ad sets your budget can support

To illustrate with PPC Land's own arithmetic rather than Meta's: an account with a weekly budget of $20,000 and a cost per purchase of $40 needs $2,000 per ad set per week to buy 50 conversions, so the formula yields a maximum of 10 ad sets. At a $100 cost per purchase the same budget supports four. Because peak season usually raises cost per result, the ceiling falls exactly when many advertisers add campaigns.

"Most accounts are running more than that," according to Meta. "Which means most accounts have written a brief with more objectives than they have money to pursue." The post does not say how Meta determined that most accounts exceed the limit.

Hold one campaign still

The post closes with recommendations sorted by available time. The single action, according to Meta, is to check whether the largest campaign optimises for the outcome the business values, change it if not, and then stop. Meta suggests an A/B test or a Conversion Lift study to read the result "against something rather than against last month," and adds: "Give it the month of October. One brief, held still, measured."

That instruction sits awkwardly with the publication date. The post went up on October 21, leaving ten days of October for a measurement window Meta describes as a month.

For advertisers with more time, the post lists three steps: run the ad set formula on the account, rewrite the brief in three lines (what the campaign optimises for, how hard it chases, how results are counted), and group any changes so that one learning phase replaces five. The final paragraph links to Meta's optimisation tools page. The post is tagged "Best practices" and "Small business."

The optimisation tools guide

A six-page Meta document titled "Optimisation tools on Meta technologies" sets out the product detail behind the post's three dimensions. The document carries no publication date; its most recent footnoted source is a Meta case study dated August 2026.

The figures and their age

The guide's opening page credits Meta's AI systems with two gains. Meta Andromeda, its ad retrieval system, is credited with an 8% increase in ad quality, and Meta GEM, its large ads recommendation model, with a 5% increase in conversions. The GEM figure comes, according to the footnote, from "a randomised experiment on Facebook Reels" run over one week in March 2025, with "a sample size of approximately 22.9 billion impressions," measuring ad conversion rate normalised by impression. The Andromeda figure is footnoted to a Meta for Business blog post from March 2025, and the page links to Meta's December 2024 engineering post on the system, which PPC Land covered when it reported advertiser skepticism of Meta's automation claims.

Both numbers are older than Meta's latest public claims. On October 13, 2026, Meta published new GEM figures, and PPC Land reported that Meta credited GEM with an 18.9% conversion gain across its apps, without a disclosed baseline or method. The guide still uses the 5% figure from March 2025.

The three headline product figures are similarly dated:

  • Incremental attribution: a 46% increase in incremental conversions compared with business-as-usual campaigns, based on 37 conversion lift studies run between July and October 2024 for 30 advertisers across 8 verticals in North America, EMEA and APAC. The footnote says only studies "where statistically significant incrementality was detected in at least one cell" were included, and that performance "isn't guaranteed."
  • Value optimisation: 12% higher return on ad spend than conversion optimisation, from four user-randomised experiments between April 17 and April 23, 2025, and a 37% ROAS increase when paired with Advantage+ sales campaigns, from a meta-analysis of 24 conversion lift studies run from September to December 2024.
  • Value rules: twice as many conversions on targeted segments compared with the same segments without value rules, based on 17 split studies run in March and April 2025 among e-commerce, retail, gaming and CPG advertisers in APAC, EMEA and North America, with rules applied to gender and age.

The 46% and 12% figures are the same ones Meta used in mid-2025, when PPC Land reported that Meta was testing profit-based ROAS optimisation and expanding incremental attribution globally. The selection rule on the 46% figure matters: excluding studies that found no significant incrementality in any cell removes the cases most likely to show no difference.

Measurement, KPIs and audiences

The guide organises its tools into three areas that mirror the post: measurement approach, KPIs and preferred audiences.

On measurement, Meta describes incremental attribution as a toggle that optimises toward "conversions caused by your ad," informed by "insights from thousands of global conversion lift studies." For businesses that rely on multi-touch attribution, the guide points to standard attribution combined with analytics integrations. Meta uses a navigation metaphor here: "Imagine if your GPS's compass was off by 30 degrees. You're moving, but slightly off course from where you actually want to go." The case study for this section is Boots, which according to Meta saw 15X higher incremental ROAS and a 15X sales lift by pairing incremental attribution with Advantage+ sales campaigns, in a case study dated February 2025. The guide does not state the baseline for either multiple.

On KPIs, the guide separates value optimisation for purchase and non-purchase events, which maximises conversion value, from value optimisation for profit, a beta that "prioritises high-margin sales" and is described as best suited to businesses "with wide margin variation e.g. in-house products vs third-party resell items." An illustration compares a $95 product carrying $75 profit with a $150 product carrying $30 profit. Level Shoes, according to a Meta case study from February 2025, recorded 24% higher total purchase value with value optimisation.

On preferred audiences, the guide covers value rules, which let advertisers bid more for selected groups. The guide's own examples include "prioritising female customers who typically spend more or in-app purchasers who tend to have greater lifetime value." Available attributes are audience (age, gender, location, operating system), placement across Facebook and Instagram Feeds, Reels and Stories plus Audience Network, and conversion location (website, app, instant forms). Ulla Popken, according to a Meta case study dated August 2026, halved its cost per new customer using value rules with Advantage+ sales campaigns.

Caveats in Meta's own text

The guide includes warnings that the peak season post leaves out. Value rules should be used "sparingly, only when you have strong signals to suggest certain audiences are worth more to your business," according to Meta, which adds: "Value rules may reduce your campaign's short-term performance, but it will help you drive results that matter most to your business over the long term." That echoes the setup warning PPC Land reported in January, when it found that value rules might cost more than they are worth for many accounts.

The best-practice page restates the learning threshold as "the minimum 50+ conversions per week to exit learning phase" and asks advertisers to consolidate campaigns, use Advantage+, diversify creative across formats including Reels and partnership ads, and send currency and revenue with every event. It also gives a data rule of thumb: "the top 10% of conversions should be at least 3 times more valuable than the lowest 10%." For measurement, Meta recommends single-cell conversion lift studies "to calibrate your measurement model on a quarterly basis."

Four worked examples

The final pages set out four use cases with specific settings:

  • A direct-to-consumer beauty brand measuring through incrementality studies: incremental attribution, value optimisation for purchase, highest value bidding, no value rules.
  • A "mature search-led marketplace" measuring ROAS in Google Analytics: standard attribution with analytics integrations, value optimisation for purchase, highest value bidding.
  • A luxury brand moving up the funnel with a 5X ROAS target on add-to-cart: incremental attribution, value optimisation on the add-to-cart event alongside existing conversion-optimised campaigns, starting with highest value bidding and then setting a target ROAS bid "10-20% below your observed performance."
  • A department store with a mainly female customer base seeking male gift buyers: incremental attribution, conversion optimisation, highest volume bidding, and a value rule to "bid 50% more for men."

The marketplace example is notable because Meta presents Google Analytics as the measurement platform of record, and its own tools as a way to align with it rather than replace it.

Why the post matters for advertisers

The timing explains the tone. Adobe forecast US online holiday spending of $275.1 billion for November 1 to December 31, 2026, up 6.7% on the prior season, with social media projected to gain 17% in revenue share. Nine days later, a Simon-Kucher survey of 1,501 consumers found US households planned to cut holiday budgets by 5.5%. Meta's opening line - shoppers who said they would spend less, then spent more - speaks directly to that gap between intention and behaviour, although the post does not say where its own numbers come from.

The post is also consistent with a direction Meta has pushed for several years: fewer, larger campaigns with automated targeting and placements, and less manual control. PPC Land has documented the friction that creates. In August, some advertisers reported that Meta removed placement controls and offered value rules as a substitute, with bid cuts capped at 90%, so a placement could be reduced but not switched off. Earlier, the expansion of value rules to placement-specific bidding in July 2025 had been presented as added control. A brief that asks advertisers to hold campaigns still sits within the same pattern.

Meta's own default settings also change during the quarter it describes. PPC Land reported on October 15 that Meta will turn on product browsing in existing ads on October 28 unless advertisers opt out. Under the post's own logic, a platform-initiated change to live ads is also a rebrief, though one advertisers did not write.

On measurement, the post's distinction between what ads "touched" and what they "added" is the core of the incrementality debate. Meta's position has been laid out before in its white paper on a hybrid "suite of truth" for measuring incrementality, which argued that last-click attribution misallocates incremental conversions. Independent tests reported by PPC Land have produced different readings; one incrementality test cited in the November 2025 coverage of advertiser skepticism found Advantage+ delivered only 17% of the conversions Meta's attribution reported.

That leaves a basic question about the evidence. The most quotable claim in the post, the 28% penalty for ad sets under the learning threshold, is Meta's figure without a method. The supporting guide's numbers are older, mostly from 2024 and the first half of 2025, and some were selected from studies that found an effect. The recommendation to measure with an A/B test or a lift study is, in effect, a recommendation that advertisers generate their own evidence - which is also the only way to test whether the claims in the guide hold for a particular account.

Meta's previous seasonal guidance followed a similar pattern. Its 2025 festive season guide, released in November 2025, linked running 15 A/B tests to 30% higher ad performance and credited Advantage+ creative enhancements with 8% more conversions per dollar. The 2026 post contains fewer product claims and more argument, but the underlying message - consolidate, automate, optimise for value, and test - is unchanged.

Timeline

Summary

Who: Meta, through its Meta for Business site, addressing advertisers running Facebook, Instagram and Audience Network campaigns, with a supporting six-page optimisation tools guide citing case studies from Boots, Level Shoes and Ulla Popken.

What: A peak season best-practice post arguing that campaign settings act as a brief to the ad auction. It states that ad sets under roughly 50 conversions a week ran about 28% more expensive per purchase in Meta's own analysis, gives a formula for the maximum number of ad sets a budget can support, and asks advertisers to fix one campaign's optimisation and then hold it still while an A/B test or Conversion Lift study measures it. The guide's supporting figures (46% more incremental conversions, 12% higher ROAS, 2X segment conversions) come from Meta studies run in 2024 and 2025.

When: The post was published on October 21, 2026, ahead of a holiday season Adobe expects to run from November 1 to December 31. The guide is undated; its latest cited source is from August 2026.

Where: The Meta for Business site, in its UK English edition, tagged for best practices and small business; the guide's studies span North America, EMEA, LATAM and APAC.

Why: Meta argues that frequent edits restart the learning phase and hide the cost of running the wrong instruction, and that splitting budgets across too many ad sets leaves each below the volume its system needs. For advertisers, the claims shape how budgets are structured in the most expensive weeks of the year, though the central 28% figure comes without a published method.