What is core-size automation in Meta catalog ads?

Core-size automation is a merchandising rule: keep a style eligible while the sizes that actually convert are in stock, and pull it back when that run dies, even if leftover tails remain. Pausing when the parent hits zero is too late. Pausing when any size dies is too blunt.

If you only remember one line: parent out of stock is a warehouse event. Core-size failure is a media event. They happen on different days, and the media event is the expensive one.

The three rules people mash together

Operators, agencies, and software all say "we pause when it goes out of stock." They rarely mean the same thing.

Catalogue availability is the floor under all three. If Navy / M is unavailable, dynamic product ads should stop shopping that line item. That is the job covered in the variant-level DPA guide. Core-size is the next question: should the hoodie itself stay in "bestsellers in stock" when M and L are gone and XS / XXL still have 40 units?

Why parent cover lies in fashion

A hoodie with 90 units looks healthy. Split it: Sand/M and Sand/L are zero. Those two variants drove most paid conversions last month. The leftover 90 is XS, XXL, and olive. Days of cover on the parent is a meeting metric. Days of cover on the sizes ads actually sell is the desk metric.

Beauty kits and one-SKU gadgets can survive parent thinking. Apparel cannot. Footwear is worse. See the vertical runbook in footwear catalog ads.

If the model in the ad wears a size you cannot pick, pack, and ship this week, you are not brand building. You are buying refunds.

How you actually name a core

Do not copy a textbook. Pull 90 days of orders. Split paid vs organic if you can. Rank size options by units, then by revenue. A useful starting cut:

• Core: sizes that make roughly 70-85% of units in that category.
• Secondary: sizes you still want on site, but you will not prospect them alone.
• Tail: sizes you keep for completeness and wholesale, not for cold traffic.

Worked mix for a UK womenswear hoodie: M 32%, L 28%, 12/S 16%, XL 11%, 8/XS 7%, XXL 6%. Core is S-L (or 8-14 if you sell numbered). If M and L die, the style is commercially broken for paid even if the parent is "in stock".

Kids, petite, plus, and shoe sizes have different cores. Trainers and Chelsea boots do not share a run. Recalculate when you change cut or factory. A core from last year's oversized drop will lie about this year's slim fit.

Where the rule lives (four layers, none of them magic)

You cannot ask Meta for "core-size automation" as a campaign objective. You encode the rule.

1. Variant availability. Dead tails flip out of stock. This is Shopify inventory plus the Meta sales channel or feed. It stops dynamic ads shopping a missing UK 16. It does not decide if the parent stays in a bestseller set.
2. Paid buffer on the core only. Hide the last 2-5 units of converting sizes from paid, even if the site can still sell them. Tails can use a thinner buffer. Write this down. Buffer stock for paid is the sibling.
3. A label the set can see. Metafield or tag such as core_cover=healthy vs broken. Product sets that say "in stock" at parent level will keep serving broken styles. Sets that require the label will not.
4. Human exceptions. Static UGC that features a dead core size. Launch colourways you will restock on Tuesday. Pre-order stories. Those are approve/reject, not feed rules. Ralph's posture here is correlation and pending review, not a silent rewrite.

The mechanical how-to for stripping one size without killing the parent is this operator guide. The combination of stock plus wasted spend is inventory and performance exclusions.

How to pause Meta ads for out-of-stock size variants, screen by screen

The policy above tells you when a style should leave the set. This is the click path for the narrower job people search for: stopping delivery on the single size that just died, without touching the parent or the ad object. Meta has no button that pauses one variant, so you are suppressing at item level and letting the ad keep serving its siblings.

1. Confirm the grain first. Shopify admin, Products, open the style, and check each variant carries its own SKU. Then Commerce Manager, Catalogue, Items, and search that SKU. If the search returns one row per style rather than one per size, your feed is publishing parents and nothing below will work. Fix the mapping before anything else.
2. Let inventory drive availability. With the Facebook and Instagram sales channel, Settings, Apps and sales channels, the channel pushes quantity changes on its own cadence rather than instantly. For a feed setup, check Commerce Manager, Data Sources, your source, Scheduled uploads and read the interval you are actually on. Do not take anyone's word for the number, including mine: cadences change by channel version, feed host and store size, and the only figure that matters is the one your own stopwatch returns in step five.
3. Verify the item flipped. Back in Items, open the SKU and read the Availability field. It should read out of stock, not in stock with quantity 0. Item diagnostics in the same screen will flag rejections that quietly keep an old value alive.
4. Stop the parent inheriting. In Catalogue, Product sets, edit the set your ad set uses and add Availability is in stock as a filter. Sets built on a collection name alone will keep the parent card in rotation with the dead size on the model.
5. Leave the ad alone. Ads Manager, Automated Rules cannot act on an item. Rules see campaign, ad set and ad only, so a rule written for this job will pause 40 living sizes to kill one.

Run the check on a Wednesday, not a Friday. Sell one unit down to zero on a quiet SKU, start a stopwatch, and record the minutes between the Shopify change and the Items screen updating. That number is your exposure window, and it is the input for the paid buffer rather than a reason to write more rules.

Where a desk helps is the bit this procedure cannot do by hand every night: Ralph puts variant stock next to Meta and Google spend so the lag and the money are on the same screen, and stages the set change or per-campaign exclude for you to approve rather than writing it silently. Writers of that kind default off until you turn them on.

What agencies actually do (and what to ask)

Search also asks: which agencies run catalog campaigns that stop when a product goes out of stock? Honest version: plenty will say yes. The useful interview question is grain.

• "When the product hits zero" usually means parent quantity in a Monday deck.
• "We watch bestsellers" usually means three heroes, not 400 variants.
• "The feed handles it" is correct for SKU suppression and silent on core-size.

A retainer that cannot answer "what happens Friday night when UK 10 dies and XL is fat" is selling hours, not a stock-aware catalog. Cost bands and what you are actually buying sit in Shopify retainer cost and Ralph vs agency. Software vs people is a bottleneck question, not a morality play: agency vs software vs buyer.

Google Shopping, same policy, different pipe

Merchant Center will keep a parent looking shoppable if any variant is in stock. That is the same lie. Supplemental feeds can overlay availability and custom labels without overwriting the primary feed. Details: GMC supplemental feeds and Shopping when the shelf is empty. Encode core_cover once. Use it on both paid shelves.

A Friday night worked example

Sand hoodie. Core S/M/L. 18:40, last M sells on site. L has 1 unit (below your paid buffer of 2). S has 7. XL has 22. Catalogue should mark M unavailable within your measured lag. L should already be withheld from paid if the buffer is real.

If your set is "all hoodies with parent inventory > 0", Sand stays in, dynamic ads lean into S and XL, and the UGC still shows the model in M. If your set requires core_cover=healthy, Sand drops from prospecting. Remarketing can still use remaining sizes if you want. Static featuring M gets staged for pause.

Time the lag with the catalogue lag test. If M is still available an hour later, fix the bridge before you write more rules.

What this is not

• Not a bid strategy.
• Not auto-pausing an ad set when one size sells out. The ad object can keep running on siblings.
• Not a reason to delete tail sizes from Shopify. Customers and wholesale still need them.
• Not WMS truth. If Shopify locations are a fiction, every rule above is show.

Ralph's honest seat: variant stock next to Meta and Google demand, then a package you approve. Docs: stock, product sets. Wider frame: stock-aware advertising. It is also not a fix for a store with a genuine conversion problem, which is better diagnosed with the funnel arithmetic in why your Shopify store isn't getting sales. At the point of a single order you are nowhere near needing this kind of rule, and the earlier stage is covered in first Shopify sale next steps.

Do this week

• Export 90 days of units by size for your top 20 paid styles. Circle the core.
• Check those cores in Meta Commerce Manager against Shopify locations. Count mismatches.
• Write the rule in one sentence: "Style stays in prospecting sets only if [these sizes] are above [buffer]."
• Pick one broken hero and apply it by hand. Then decide if a label/set is worth the plumbing.

If that sentence already dies in Slack every busy week, you do not need another dashboard tile. You need one desk that treats the size run as a media input. Why that desk exists: Ralph for fashion, why Ralph for catalog ads. Cover math: days of cover. When stock is already thin: ads when stock is thin. Sets rot: product set rot. Static still lying: dead-size creative.

Encode a core-size policy for Meta catalog ads

1. Name the core from orders. Pull 90 days of paid and organic units by size option. Mark the sizes that make 70-85% of units as core for that category.
2. Map variants to catalogue IDs. Confirm Shopify variant ID or SKU equals Meta retailer_id. Parent IDs cannot carry a core-size rule.
3. Keep availability honest. Dead tails go unavailable. Core sizes get a paid buffer so ads stop before the last unit.
4. Tag styles when the core is broken. Use a metafield or custom label such as core_cover=healthy|broken. Rebuild in-stock product sets off that label, not parent quantity.
5. Stage the human exceptions. When a hero's core dies, pause static that features those sizes and review prospecting. Do not wait for parent stock to hit zero.

If a sold-out size is the problem: Ralph reads every size against every campaign overnight, finds the ones still paying for an empty shelf, and stages the exclusion with a revert for restock day. You say yes. See Ralph · How Ralph handles out-of-stock sizes · Docs