Pillar 03: Amplification
Meta is quietly moving its ad features from opt-in to on by default, and three of the four switches live outside Ads Manager.
This week added product showcase to that list. From API v27.0, live since November 4, Meta can put catalog products into an ad you briefed as a single-product hero, and it decides which products. Separately, Meta can read a discount off your website and show it in ads where promotions are switched off.
Nobody builds a messy permissions setup on purpose. It gets that way one reasonable-looking default at a time, in four different screens. The job before the Black Friday freeze is 45 minutes: find all four switches, decide each one on purpose, and clear your feed errors, because your feed is now ad copy.
What changed
Delivery
v27.0
Meta's new API version can turn product showcase on for ads built without showcase settings, acting only when it has what Meta calls high confidence that a product set matches. If your tooling creates ads, set the opt-out yourself. If you build by hand, go look at the previews on your newest ads.
PPC Land, Nov 4, 2026, quoting Meta's developer blog. Release date confirmed on Meta's own changelog. (opens in a new tab)Nov 4, 2026Reported
Risk
Four switches
Product browsing flipped on in late October. Enhancements turned up in accounts without anyone opting in. Product showcase defaults on from November 4. Detected offers was found ticked on November 6. Device expansion is queued for around February. One shift, four screens: Advantage+ Creative toggles, account Advertising Settings, Commerce Manager, and the API call your tool makes.
PPC Land and Adweek, Oct 15 to Nov 6, 2026, plus Meta's v27.0 release notes via PPC Land. (opens in a new tab)Oct 15 to Nov 6, 2026Reported
Terms
Still paused
Meta's Business Help Center says the four updated agreements "are now paused and will not take effect on November 20, 2026," with no new date. The useful part for an operator is what is already in force. Under the 2024 terms, the advertiser is already solely responsible for AI-generated content it uses. The paused update would have widened that. It did not invent it.
Search Engine Journal, Nov 10, 2026, quoting Meta's Business Help Center notice. (opens in a new tab)Nov 10, 2026Confirmed
The numbers
What peak week did to the numbers, in two separate 2025 datasets
Change during the Black Friday period against the rest of the year. Two independent samples, both 2025 data. No 2026 peak data has been published yet.
- Confect, 1,362 advertisers
- Billy Grace
Source: Confect study of 1,362 advertisers (published Sep 25, 2026) and Billy Grace study via PPC Land (opens in a new tab), both using 2025 dataReported
See the data as a table
| Metric | Confect, 1,362 advertisers | Billy Grace | Source |
|---|---|---|---|
| CPM | 45% | 36% | Confect, Black Month 2025 data, published Sep 25, 2026; Billy Grace, Black Friday week 2025, EUR medians, via PPC Land |
| Cost per order | -10% | -18% | Confect, Black Month 2025 data, published Sep 25, 2026; Billy Grace, Black Friday week 2025, EUR medians, via PPC Land |
| ROAS | 34% | 26% | Confect, Black Month 2025 data, published Sep 25, 2026; Billy Grace, Black Friday week 2025, EUR medians, via PPC Land |
Meta price per ad against impression growth
Year-over-year change, worldwide. Price per ad is Meta's own earnings metric, not CPM.
- Price per ad
- Ad impressions
Source: Meta quarterly earnings, Q4 2025 to Q3 2026Reported
See the data as a table
| Quarter | Price per ad | Ad impressions | Source |
|---|---|---|---|
| Q4 2025 | 6% | 18% | Meta Q4 2025 filing (Jan 28, 2026) |
| Q1 2026 | 12% | 19% | Meta Q1 2026 earnings |
| Q2 2026 | 12% | 14% | Meta Q2 2026 earnings (Jul 29, 2026) |
| Q3 2026 | 14% | 10% | Meta Q3 2026 results (Oct 21, 2026) |
Benchmarks
Each figure with its source and the date or window it covers. A range means the sources disagree or the figure is a band.
| Metric | Figure | Source | Note |
|---|---|---|---|
| Price per ad, worldwide, Q3 2026 | 14% | Meta Q3 2026 results (opens in a new tab)Oct 21, 2026Reported | Year over year. Impressions up 10% over the same period |
| Price per ad, US and Canada, Q3 2026 | 23% | Meta Q3 2026 results (opens in a new tab)Oct 21, 2026Reported | Year over year. Impressions up 8% over the same period |
| Peak period CPM lift, two 2025 samples | 36% to 45% | Confect (1,362 advertisers) and Billy Grace via PPC Land (opens in a new tab)2025 data, published Sep 25, 2026 and earlierReported | Black Friday period against the rest of the year. Confect puts the CPM at $10.90 against $7.50 |
| Peak period cost per order, two 2025 samples | -18% to -10% | Confect (1,362 advertisers) and Billy Grace via PPC Land (opens in a new tab)2025 data, published Sep 25, 2026 and earlierReported | Black Friday period against the rest of the year. Costlier media, cheaper orders |
| Peak period cost per purchase, absolute median | 24.51 USD | Confect, 2.6B catalog impressions, 1,362 advertisers (opens in a new tab)Black Month 2025 data, published Sep 25, 2026Reported | Advertiser-balanced median, catalog ads only, countries and currency not stated. Alongside it: ROAS 5.08, CTR 2.70%, conversion rate 1.49%, CPM $10.85, AOV $128 |
| Thin ad set cost penalty, under about 50 conversions a week | 28% | Meta Peak Season 2026 guide, via PPC Land and Social Media Today (opens in a new tab)Meta post Oct 21, 2026Reported | More expensive per purchase than ad sets above the threshold. No sample size, period or category given |
No 2026 Black Friday cost data has been published anywhere as of Nov 16, 2026. Everything on peak is 2025. Meta publishes no CPM, so price per ad from its earnings is the official proxy. No traceable 2026 dataset exists for hook rate, hold rate or CTR bands in CPG and DTC, so the circulating 25, 30 and 40 percent hook bands have nothing behind them.
The evidence
What each headline claim rests on, and whether the sample behind it was disclosed.
| Claim | Figure | Evidence | Period | Sample disclosed |
|---|---|---|---|---|
| Peak period CPM, cost per purchase, ROAS, CTR, conversion rate and AOV | CPM +45%, cost per purchase -10%, ROAS +34%, CTR +21%, CVR +25%, AOV +16% | 2.6 billion catalog ad impressions, $24 million spend, 17,390 ads, 1,362 brands and retailers. Advertiser-balanced medians. Ads under $30 spend and extreme outliers excluded. The study calls these correlations across a large dataset, not controlled tests Confect study, 1,362 advertisers (opens in a new tab)Black Month 2025, published Sep 25, 2026Reported | Black Month 2025, published Sep 25, 2026 | Yes |
| Price shown with a voucher code in the creative | +64% ROAS, -38% cost per purchase | Same dataset. Against the average ad in the period. Correlational, catalog ads only Confect study, 1,362 advertisers (opens in a new tab)Black Month 2025Reported | Black Month 2025 | Yes |
| Ads built for the period against evergreen ads left running | ROAS lift +37% against +15% | Same dataset. Baseline not clearly specified in the study Confect study, 1,362 advertisers (opens in a new tab)Black Month 2025Reported | Black Month 2025 | Yes |
| Manual audience targeting against broad ad sets | -19% ROAS | Same dataset, same period Confect study, 1,362 advertisers (opens in a new tab)Black Month 2025Reported | Black Month 2025 | Yes |
| Product browsing, conversions | +3% | About 117,700 advertisers across three or more verticals, significant above 99% confidence. Meta did not disclose which conversions, the attribution window, or any cost data Meta notice via PPC Land (opens in a new tab)Aug 6 to 21, 2026Reported | Aug 6 to 21, 2026 | Partly |
| Advantage+ sales campaigns, ROAS | +37% | 24 cross-vertical Conversion Lift studies Meta optimization guide via PPC Land (opens in a new tab)Sep to Dec 2024Reported | Sep to Dec 2024 | Yes |
| Claim | Figure | Evidence | Period | Sample disclosed |
|---|---|---|---|---|
| Incremental attribution, incremental conversions | +46% | 37 studies Meta optimization guide via PPC Land (opens in a new tab)Jul to Oct 2024Reported | Jul to Oct 2024 | Yes |
Every change
4 changes logged this week, highest impact first.
| Change | Impact | Source | When |
|---|---|---|---|
| The permissions audit stops being a settings check and becomes a standing weekly line across four places, including the API call your tooling makes. Old: audit the Advantage+ toggles and screenshot delivered ads. New: assume every new feature arrives on by default, and check all four screens every week | High | PPC Land (opens in a new tab) and Adweek (opens in a new tab), four dated reports Oct 15 to Nov 6, two of them Meta-sourcedReported | Oct 15 to Nov 6, 2026 |
| Product data joins buying, creative and measurement as a standing area. The catalog is a media input, not plumbing: Meta reads it across all Sales campaigns through product showcase, product browsing, availability signals and product sets. New rules: weekly feed diagnostics with zero errors through peak, product IDs matching the IDs the Pixel sends, and margin-tier labels so prospecting points at products worth selling | High | Meta developer documentation on catalog behavior, plus the v27.0 showcase defaultReported | Nov 4, 2026 |
| Peak benchmarks rebuilt on a 1,362-advertiser dataset with a disclosed method, replacing a thinner vendor figure. The instruction does not change: judge peak on cost per order and blended MER, not CPM. The evidence behind it does, because two independent 2025 samples now agree on direction | Medium | Confect study, published Sep 25, 2026, alongside the earlier Billy Grace studyReported | 2025 data |
| API version hygiene becomes a tooling question, not a date question. Current version is v27.0, released Nov 4. Marketing API version auto-upgrade shipped Jul 29, 2026, so a tool can move versions without being asked, which is how a default-on feature reaches an account nobody touched. New rule: confirm each integration's version and whether auto-upgrade is on, before every peak period | Medium | Meta's own Graph API changelog, read Nov 16, 2026Confirmed | Nov 4, 2026 |
Standing rules
How we read Meta now, newest first. Meta has run as a weekly channel read since July 2026. This is the first public issue, so rules and retired claims carry the backlog rather than only this week's changes. See the full log across every issue.
| Rule | Area | Change | Logged | Supersedes |
|---|---|---|---|---|
| Audit the full permissions surface weekly, all four places: Advantage+ Creative toggles, account-level Advertising Settings, Commerce Manager > Settings > Offer merchandising, and the API degrees_of_freedom_spec on ads your tooling creates. Assume new features arrive on by default | Governance | replaces | Audit enhancement settings per ad and screenshot delivered ads weekly | |
| Weekly catalog diagnostics with zero errors through peak, match rate 90% or better, and product IDs that match the IDs the Pixel and Conversions API send | Product data | new | Nothing | |
| Label the catalog by margin tier and point prospecting at a high and mid margin product set, so Meta does not optimize toward the cheapest, worst-margin SKU | Product data | new | Nothing | |
| Confirm every integration's API version, and whether Marketing API auto-upgrade is on, before each peak period | Governance | new | Nothing | |
| One offer per ad during promotional periods, and make the offer legible in the asset rather than leaving it to the landing page | Creative | refines | One offer per ad during promotional periods | |
| Cap daily budget edits at +20%, or follow Meta's in-product learning-phase prompt where it gives one. No significant edits on peak days | Buying | refines | Cap daily budget edits at +20% | |
| Size structure with weekly budget divided by cost per purchase divided by 50 to get the maximum ad sets, recalculated on current cost per purchase rather than last quarter's | Buying | confirms | Nothing | |
| Start from an incrementality factor of 1.0 for click-only prospecting and calibrate per account with a holdout, geo test or Conversion Lift | Measurement | new | Meta ROAS overstates incremental ROAS by 1.5 to 3 times | |
| Before any Meta measurement change, baseline Meta-reported purchases against store orders over 28 days, then track the ratio. Never scale on reported gains alone | Measurement | new | Nothing | |
| Rebuild cost per acquisition and MER targets on peak-period costs before peak, and judge peak on cost per order and blended MER rather than CPM | Costs | new | Nothing |
| Rule | Area | Change | Logged | Supersedes |
|---|---|---|---|---|
| Three to five genuinely distinct concepts per product category, built 9:16 with sound and captions, with the hook and offer in the first three seconds. AI variants are fuel inside a concept, never a new concept | Creative | new | Nothing |
Retired claims
Figures and claims we stopped repeating, and why.
| Claim | Why we retired it | Retired |
|---|---|---|
| Andromeda 2.0 shipped an October 2026 update | A dozen agency blogs published it in six weeks. No Meta announcement, engineering post or changelog entry supports it | |
| Any single peak CPM multiplier quoted as a 2026 forecast | Both available samples are 2025. Say CPM up 36 to 45 percent in 2025 samples, as a direction, not a forecast | |
| Marketing API v24.0 retires Nov 10, 2026 | It ended Oct 6, 2026, per Meta's own changelog. The current version is v27.0 | |
| Andromeda and GEM drove 15.7% more conversions | The figure traces to Meta's Q2 2026 earnings and covers GEM plus the user-understanding models on Facebook, platform-wide. It is not a GEM-only or Andromeda figure, and not something an individual account should expect | |
| The Pixel change recovers up to 15% more purchases | Unsourced agency claim | |
| Advantage+ lowered the conversion threshold from 50 to 25 per week | No primary source exists. Traced back to a 2024 practitioner post asking the question |
What to watch next week
- Product showcase in the wildWhether any advertiser reports catalog products turning up in a non-catalog ad, and whether Meta publishes a help article or states a default.
- 2026 peak cost dataThe first real Black Friday CPM and cost per order reads from trade press or a named dataset. Everything we have is 2025.
- Pixel on-device measurementThe default landed Nov 12 and nobody has published what it did. Also open: whether recovered conversions feed bidding or carry purchase value.
- The paused termsA new effective date, a sign the AI language is being rewritten rather than reinstated, and the UK competition regulator's process running through January 2027.
- Six-second ThruPlayOne practitioner reports a new ThruPlay duration option, 6 seconds alongside the usual 15, on Awareness and Engagement campaigns. No Meta announcement. If it is real, cost per ThruPlay stops being comparable between ad sets. Unverified.
- Product-level metrics for catalog campaignsOne practitioner reports a Top products panel with seven product metrics, which would finally show which products the system favors. One report of sitelinks counted as products. No Meta documentation. Unverified.
- The post-peak windowWhether the auction actually softens in late December, and by how much.
Terms used
- Learning phase
- The stretch right after you create or significantly change an ad set, when Meta has too little recent conversion data to deliver efficiently and is still exploring with your budget.
- Product showcase
- A feature that places items from your product catalog next to your ad creative, so an ad briefed as one product can serve as a product grid.
- Product browsing
- An in-app page Meta builds from your catalog, shown after someone taps your ad, instead of sending them to your own site.
- Detected offers
- A Commerce Manager setting that lets Meta read discounts off your website and display them in your ads.
- MER
- Marketing efficiency ratio: total revenue divided by total ad spend across all channels, over a stated window. A revenue readout, not a margin one.
- Incrementality
- The sales that would not have happened without the ad, measured by holding a group back rather than by what the platform reports.
- Event Match Quality
- Meta score from 1 to 10 for how well the customer details you send match real Meta accounts. Meta documents 6 or better as the target.
- ThruPlay
- A video view counted once someone watches to a set duration, usually 15 seconds, or to the end if the video is shorter.
- Conversions API
- A server-to-server connection that sends purchase and other events to Meta directly from your store, alongside the browser Pixel.
Where we read it
The outlets we watch for this channel. A figure's status follows its best source.
- Meta for Business news (opens in a new tab)Tier 1, the platform itself
- Meta investor relations (opens in a new tab)Tier 1, the platform itself
- PPC Land (opens in a new tab)Tier 2, trade press
- Search Engine Land (opens in a new tab)Tier 2, trade press
- Social Media Today (opens in a new tab)Tier 2, trade press
- Barry HottTier 3, practitioners and vendors
- Bram Van der HallenTier 3, practitioners and vendors
- Common Thread CollectiveTier 3, practitioners and vendors
- Foxwell DigitalTier 3, practitioners and vendors
- Jon Loomer (opens in a new tab)Tier 3, practitioners and vendors
- Triple Whale (opens in a new tab)Tier 3, practitioners and vendors
