Performance Max Strategy for BFCM: Asset group structures, feed optimizations, and audience signals.
· PPC

Most brands prepare Performance Max (PMax) for Black Friday by raising budgets, lowering ROAS targets and trusting the algorithm to spend the extra money well. That hands the most expensive weekend of the year to a system working from whatever inputs it was given in October.
PMax picks placements, bids and audiences on its own. Your control sits upstream, in 3 inputs that decide what it does with a budget several times higher than normal:
- Asset group structure decides which products and messages travel together.
- The product feed decides what PMax can show and how well it matches queries.
- Audience signals and search themes tell PMax where to start looking before it has data of its own.
We reviewed PMax asset group data from 6 ecommerce accounts we manage to see how the system behaves when spend concentrates. Below is how we set up each input for Black Friday (Nov 27, 2026) and Cyber Monday (Nov 30, 2026), plus the controls worth adding. Structural changes need to land by early November so Smart Bidding can settle before peak week.
Asset group structure: add BFCM groups to campaigns with history
The safest BFCM structure keeps your existing PMax campaigns intact and adds holiday asset groups inside them. Google's retail guidance recommends this setup because it protects evergreen products while the new groups borrow the campaign's conversion history (Google). A new campaign launched in November enters peak week with no data, which is the worst position to bid from.
We build the BFCM layer with 4 components:
- One asset group per product theme. Group products that share a buyer and a message, such as a hero collection, a gift set range or a category on deep discount. Give each group its own landing page, headlines and images written for that set only.
- Listing groups that match the theme. Point each BFCM asset group at its products using custom labels or product type, so it never inherits the whole catalog. Offer messaging served next to full-price products confuses the shopper as much as the algorithm.
- Evergreen groups that stay live. Products outside the promotion keep serving, which also gives you a clean baseline to compare against after the event.
- Asset group scheduling. Set BFCM groups to start when the promotion goes live and end when it closes, so expired offers never serve on Cyber Tuesday.
Google also recommends uploading holiday asset groups at least 2 weeks ahead so they clear review (Google). With Black Friday on Nov 27, your upload deadline lands on Nov 13.
A separate campaign earns its place in 2 situations. One is a product set that needs its own budget or ROAS target, such as holiday-only merchandise or a high-margin range you want to protect. Another is a product set that would otherwise absorb the whole budget. Outside those cases, every extra campaign thins the data in the rest.
BFCM creative inside asset groups
Every BFCM asset group needs full coverage across text, images and video, because missing asset types limit the inventory a group can serve on (Google). Ad Strength is the fastest way to spot gaps, so fix anything below Good before launch.
The creative rules we follow for promo groups:
- Put the offer in the text assets. Headlines and descriptions should carry the discount, the dates and any free shipping threshold, since PMax uses these assets on Search and Shopping placements.
- Let product imagery carry the visual. Lifestyle and product shots do the heavy lifting, with a single offer badge or price callout layered in sparingly.
- Supply your own video. If an asset group has no video, Google may generate one from your images, text and feed (Google). A short cut-down of your best Meta or YouTube creative almost always beats the generated version.
- Review text customization. Automatically generated headlines can pull older or full-price messaging from your site. If the generated copy conflicts with the promotion, adjust it or turn off text customization on BFCM groups.
- Add a promotion asset at campaign level. It shows the offer as an annotation on Search inventory without rewriting every headline. Promotion assets also carry their own start and end dates.
Save a copy of every BFCM asset you upload. Asset reporting is far easier to read in the post-event readout when you know which assets were promotional.
Feed optimizations that matter most during BFCM
For a retail PMax campaign, the product feed does more work than any asset group. It decides which products are eligible, which queries they match and which price and offer annotations shoppers see. Sale weekend is when feeds break most often, because prices change across the whole catalog at once.
Here are the feed changes we prioritize, in order of impact:
- Schedule sale pricing. Use sale_price with sale_price_effective_date so discounted prices switch on and off on time, without a manual feed push mid-event. Price mismatches between the feed and the landing page are a common BFCM disapproval.
- Turn on automatic item updates. Merchant Center uses them to correct price and availability from your site when the feed lags, which makes them a safety net while prices move.
- Submit Merchant promotions early. Promotions add a deal annotation to Shopping placements after passing a separate review. Google also supports price drop, shipping and returns annotations that help an offer stand out next to competitors (Google).
- Rewrite titles on hero products. Lead with the product type, followed by the attributes people search for, such as size, material or color. A vague title on a top seller costs impressions at the moment demand peaks.
- Add BFCM custom labels. Tag products by margin band, stock depth, promo status and hero status. These labels power the listing group segmentation in the next section.
- Use a supplemental feed for temporary changes. Event-only titles, labels and promo flags are easier to add and remove through a supplemental feed than through the primary source.
Finish by checking inventory sync, including how often the feed is fetched or whether it runs through the Content API or a platform app. Sold-out products need to drop out fast enough that PMax never spends on stock you cannot ship.
Listing groups: steer spend toward the products you want to sell
Left alone, PMax spends where it already sees conversions, so a small share of the catalog usually takes most of the budget. During BFCM that can be the wrong set of products, either because margins are thin at the sale price or because stock will not last the weekend. Listing groups built on custom labels let you correct that.
The segmentation we use most often:
Segment | Built from | What we do with it |
|---|---|---|
Hero products | Custom label for top sellers with deep stock | Own BFCM asset group with the strongest creative and offer |
High-margin range | Custom label by margin band | Separate campaign if it needs its own ROAS target, otherwise its own asset group |
Low-impression products | Products with little or no spend in the last 30 days | Own asset group or campaign so they get a chance to serve |
Low stock or low margin at sale price | Custom label set before the event | Excluded from BFCM groups, or excluded entirely for the event window |
Before launch, confirm each BFCM asset group contains only the products its creative talks about. Then look for products accidentally excluded by an old listing group filter, since a product left out of every group cannot serve anywhere in the campaign.
Audience signals and search themes: where PMax starts looking
Audience signals show PMax who has converted before, which gives it a starting point. They do not restrict delivery, so PMax will serve outside the signal whenever it predicts a conversion there (PPC Land). Signals matter most in the first days of a new asset group, before it builds conversion data of its own.
For BFCM asset groups, we build signals from the strongest source down:
- Customer Match lists of past purchasers. Google names consented first-party purchaser data the most valuable type of audience signal (Google). Split the list where you can, for example past BFCM buyers or customers of the category the asset group sells.
- Recent site visitors and cart abandoners. These capture shoppers already in market who are likely waiting for the sale.
- Custom segments built from search terms. Use the queries your best customers search before buying, including deal-intent phrasing for the category.
- Interests and in-market segments. These add little that PMax would not find on its own, so we only use them when first-party lists are thin.
Refresh Customer Match uploads in early November, since a list last updated in summer leaves out everyone who bought since.
Search themes act more like targeting than audience signals do. Google says they behave like phrase or broad match keywords in Search campaigns, with up to 50 allowed per asset group (PPC Land). Add them to BFCM asset groups for the product and deal queries you expect to spike, such as your category plus "Black Friday" or "sale." An identical exact match keyword in a Search campaign takes priority over a matching search theme, so themes fill gaps around your Search coverage rather than replacing it.
If you also run Meta, feed the same first-party lists into both platforms. A past-purchaser list that lives in only one platform is a signal the other platform never sees.
What our accounts show
We reviewed PMax asset group performance across 6 ecommerce accounts we manage over the 90 days to Oct 28, 2026. Each of the 3 patterns below shapes how we set up BFCM.
Branded PMax flatters the blended number. Half the sample, 3 of 6 accounts, run a branded PMax campaign alongside non-brand ones. In all 3, the branded campaign returned far more per dollar than the non-brand campaigns.
Account | Branded PMax ROAS | Non-brand PMax ROAS | Ratio |
|---|---|---|---|
Account A | 11.6 | 1.5 | 7.7x |
Account B | 11.2 | 1.5 | 7.5x |
Account C | 5.7 | 2.3 | 2.5x |
When brand and non-brand traffic share a PMax campaign, the blended ROAS hides how the non-brand side is performing. That is the case for brand exclusions or a separate branded campaign before BFCM spend ramps.
One asset group usually takes about half the budget. Across the 7 PMax campaigns in the sample running more than 1 active asset group, the top group took a median 51% of campaign spend, with a range of 28% to 63%. PMax does not split budget evenly, so having an asset group is no guarantee a product set gets meaningful spend.
Because spend concentrates in groups PMax already favors, a BFCM group switched on the morning of Black Friday competes against established groups the system trusts. Either launch BFCM groups early enough to build some history, or move the offer creative into the proven group for the event window. Keep the dedicated group for products that need their own message.
Seasonal libraries are already standard practice. Most accounts in the sample keep a library of paused seasonal asset groups inside evergreen campaigns. A further 2 accounts run a dedicated campaign for low-impression products. Both patterns match the structure we recommend above.
All data anonymized. The window covers the 90 days to Oct 28, 2026. With a sample this small, treat these as patterns to check in your own account rather than benchmarks.
Budgets, bidding and the controls worth using
Structure, feed and signals set what PMax can do, while budget and bid settings decide how hard it pushes during the event. A few controls keep that push pointed at incremental sales.
Seasonality adjustments. Google positions these for short events where conversion rates shift by more than 30% over 1 to 7 days, which fits Black Friday weekend (Google). They rely on conversion history, so they only work on existing campaigns. For longer promotions, Google recommends adjusting ROAS or CPA targets manually.
Budget headroom. Raise budgets ahead of the peak days rather than during them. Check that no campaign is capped on its busiest day, because a PMax campaign that hits its budget by midafternoon on Black Friday leaves its strongest hours unbought.
New customer acquisition goal. Make this decision deliberately before the event. New Customer Value mode bids higher for new customers while still converting returning ones, whereas New Customer Only excludes returning buyers entirely (Google). Brands using BFCM to build their customer file should choose the first option. For a brand focused on margin from an existing base, neither may fit.
Brand exclusions. Without them, PMax can absorb branded searches your Search campaign would have captured anyway, which inflates its reported ROAS. Apply them if you want a clean read on PMax's incremental contribution.
Negative keywords. PMax now supports up to 10,000 campaign-level negative keywords (Google). Review search term insights in early November, then exclude competitor terms, irrelevant queries and anything your Search campaigns should own.
The A/B test toggle. Google has started offering a setting that saves asset group edits as an A/B test (PPC Land). PPC Land also reports an auto-apply option switched on by default in some accounts. Check this setting before you upload BFCM assets, because a split test running on a promo asset group over the event weekend splits the traffic your new creative receives.
During and after the event
Once BFCM traffic arrives, your job shifts from building to protecting. Structural edits during peak week cost more than they save, so limit changes to budgets, feed fixes and anything visibly broken.
Check these daily through the event:
- Budget pacing by campaign. Confirm no PMax campaign caps out before the hours that convert best in your account.
- Merchant Center diagnostics. Look for new price mismatch or availability disapprovals, especially on hero products. Fix them in the feed rather than the campaign.
- Stock on hero products. Exclude or relabel products close to selling out so PMax stops spending on them.
- Asset group spend split. Compare each BFCM group's share of spend against your plan. A group taking almost nothing is your signal to move its offer into a stronger group.
After Cyber Monday, wind down in sequence:
- End the seasonality adjustment on schedule.
- Let the scheduled BFCM asset groups pause.
- Remove expired promotions from Merchant Center.
- Return budgets and ROAS targets to normal levels over a few days rather than all at once.
Run the readout within 2 weeks while the detail is fresh. Compare each BFCM asset group against the evergreen groups in the same campaign, check how new customer volume moved if you used the acquisition goal, then note which feed fixes made a measurable difference. That readout becomes the starting point for next year's plan.
PMax BFCM checklist
When | Task |
|---|---|
Early November | Add custom labels for margin, stock, promo and hero status |
Early November | Build BFCM asset groups inside existing campaigns with matching listing groups |
Early November | Refresh Customer Match lists and add search themes for deal queries |
Early November | Review search term insights and add negative keywords |
Early November | Decide on brand exclusions and the new customer acquisition goal |
Before uploading assets | Check whether asset group edits are being saved as an A/B test |
By Nov 13 | Upload BFCM assets and asset groups with scheduling set, so they clear review |
By Nov 13 | Submit Merchant promotions and schedule sale_price with effective dates |
Week of Nov 23 | Raise budgets and set the seasonality adjustment for the event window |
Nov 27 to Nov 30 | Watch pacing, disapprovals, stock and asset group spend daily |
Early December | Wind down adjustments, budgets and promotions, then run the readout |
PMax will spend whatever budget you give it during BFCM. Whether that spend buys incremental sales depends on work finished before Nov 13, when your holiday assets need to enter review. Build BFCM asset groups inside campaigns with history, using listing groups that protect margin and stock. Schedule sale pricing so the feed never lags the site, then refresh Customer Match lists before you raise a single budget.
Once the event starts, protect what you built and save the analysis for the readout. The brands that win BFCM on PMax walk into the weekend with inputs that were ready 2 weeks earlier.