When to Launch BFCM Campaigns So They Exit Learning Before the Auction Peaks

· Tutorials/Guides

When to Launch BFCM Campaigns So They Exit Learning Before the Auction Peaks

Every November, we see brands launch Black Friday campaigns into the most expensive auction of the year while the algorithm is still working out who should see them. A campaign that goes live on Nov 16 spends its learning phase inside peak week, paying record CPMs for unstable delivery.

Black Friday lands on Nov 27, 2026, with Cyber Monday following on Nov 30. The deadline that matters arrives weeks before either date. New Google campaigns belong in market between Oct 19 and Oct 26. Meta campaigns and ad sets follow between Oct 26 and Nov 2. From Nov 13 onward, the account structure stays locked.

Those dates give each campaign 1 to 2 learning cycles plus a buffer before mid-November CPM inflation sets in. We run Google, YouTube and Meta for 8-figure ecommerce brands, so we plan BFCM backward from the learning phase instead of forward from the promo date.

This guide covers the learning-phase math behind each deadline, the budget ramp that quietly resets stable ad sets, a clean way to load offer creative, and a working calendar you can drop into your plan this week.

The 3 dates that matter

Work backward from Nov 27 and three deadlines fall out of the learning-phase math.

  • Oct 19 to Oct 26 (Google). New campaigns, Performance Max, and any bid strategy or target change go live. Google gets the earlier window because Smart Bidding and PMax need a longer runway to settle.

  • Oct 26 to Nov 2 (Meta). New campaigns and ad sets go live, including a dedicated BFCM ad set seeded with teaser creative.

  • Nov 13 (structure lock). From here, the only changes are budget increases you scheduled in advance and the Google seasonality adjustment.

Launching inside these windows gives every campaign 1 to 2 learning cycles before mid-November CPMs start climbing. A launch that misses them pays peak prices while the algorithm is still guessing.

Meta: the learning phase runs on volume

A Meta ad set exits learning after roughly 50 optimization events within 7 days of its last significant edit. Meta treats these as significant edits, and each one restarts the count:

  • targeting changes

  • a new optimization event or bid strategy

  • adding a new ad

  • large budget changes

  • pausing for 7 days or longer

An ad set launched on Nov 2 that hits 50 events by Nov 9 gets 2 full weeks of stable delivery before peak week. Launch the same ad set on Nov 16 and it learns inside the most expensive auction of the year, with unstable delivery on top of peak CPMs.

Volume decides the outcome more than the calendar does. At 50 events a week, each ad set needs about 7 conversions a day. If an ad set can't reach that at pre-peak budgets, launching it earlier won't fix the problem.

Do not split budget across more ad sets than your conversion volume can feed. Consolidate until each ad set clears 7 conversions a day, or optimize for a higher-volume event such as add to cart.

Google: Smart Bidding needs a longer runway

Smart Bidding usually needs 1 to 2 weeks to settle after a new campaign launch or a bid strategy change. Google also recommends giving Performance Max several weeks of data before you judge it, which is why the Google window closes a week ahead of Meta's.

Set your tROAS and tCPA targets during the Oct 19 to Oct 26 window, then leave them alone. A target change in late November sends the bid strategy back into learning right as conversion rates spike.

Handle the event itself with a seasonality adjustment. Schedule one covering roughly Nov 26 to Dec 1 and enter the conversion rate lift you expect for that window. Google built the feature for short conversion rate spikes of a few days, so Smart Bidding raises bids for BFCM traffic without relearning.

End the adjustment on Dec 1. If it runs into post-peak traffic, Smart Bidding keeps bidding for conversion rates the account no longer delivers, and that spend comes straight out of contribution margin.

The budget ramp is the hidden trap

Data Druid, our account intelligence platform, tracks spend and performance across the client accounts we manage. For BFCM 2025, we measured each account's worst 30 consecutive days of spend and its single peak day against a normal August day. The benchmark covers 12 accounts with enough history to qualify, with all data anonymized.

Multiple of a normal August day

25th percentile

Median

75th percentile

Worst 30 consecutive days

1.07x

1.79x

2.37x

Peak single day

1.84x

3.78x

9.93x

The median account ran close to double its August daily spend for a full month, with a peak day near 4x. The spread between quartiles is wide, so plan from your own August baseline and last year's curve.

Jumping from August budget straight to 3x on Black Friday is the kind of large edit that knocks a stable ad set back into learning. Step budgets up through the first half of November instead, so the account reaches peak levels while delivery is still stable.

Meta's budget scheduling lets you preset increases for specific dates. Before you rely on it, confirm in Meta's current Help Center documentation whether scheduled increases count as significant edits.

Getting BFCM creative in without resetting learning

Adding an ad counts as a significant edit on Meta. Drop your offer creative into your best-performing ad set on Nov 26 and you reset its learning at the worst possible moment.

There are 2 cleaner ways to handle it:

  • Seed a dedicated BFCM ad set. Launch it in the Oct 26 to Nov 2 window running early-access or teaser creative. When the offer goes live, swap the offer creative into this ad set so any relearning stays contained there, away from your evergreen winners.

  • Load the offer creative early and plan the relearn. Upload it paused, then activate it during a lower-stakes week in early November. You accept a single relearn on your schedule instead of one forced on you during peak week.

We default to the dedicated ad set for most accounts. It protects the ad sets carrying your baseline revenue, and its early-access phase gives you a read on which offer angles convert before the auction gets expensive.

Use the early launch to log a CPMr baseline

CPMr (Cost Per Mille Reached) is the cost to reach 1,000 unique people, calculated as CPM x frequency. Phil Kiel, our Director of Paid Social, popularized it internally as a leading indicator because it climbs weeks before CTR, CPA or ROAS show trouble.

Launching in late October gives you a clean pre-peak CPMr baseline. Log it weekly from Nov 2 so you can diagnose any climb in November by splitting it into its 2 inputs:

  • CPM rising with frequency flat is seasonal auction pressure. Every advertiser is bidding into the same inventory, so ride it and let your scheduled budget increases absorb it.

  • Frequency rising with CPM flat means your new campaigns are cycling through the same audience. Fix it with tighter exclusions and campaign consolidation.

The second pattern is the one to catch in early November, while the structure is still open to changes. After the Nov 13 lock, the fix itself becomes a significant edit.

Your BFCM 2026 working calendar

Copy these 5 phases into your plan and assign an owner to each.

  1. Oct 19 to Oct 26. Google new campaigns, PMax, and any bid strategy or target changes go live.

  2. Oct 26 to Nov 2. Meta new campaigns and ad sets go live, with the BFCM ad set seeded on teaser creative.

  3. Nov 2 to Nov 13. Confirm every ad set has exited learning, step budgets up, and log weekly CPMr.

  4. Nov 13 onward. Structure stays locked, with only pre-scheduled budget moves and the Google seasonality adjustment allowed.

  5. Nov 23 to Dec 1. Peak window, with Black Friday on Nov 27 and Cyber Monday on Nov 30.

BFCM performance gets decided in October. Google campaigns need to be live by Oct 26, Meta campaigns by Nov 2, and the account structure locked by Nov 13. Every structural change after that point trades peak-week stability for a relearn you pay for at the year's highest CPMs.

Pull up your account this week and count the ad sets you plan to launch for BFCM. Check each one against the 7-conversions-a-day bar, consolidate the ones that fall short, and put your Google launch dates on the calendar before Oct 19.


Back to all articles