Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Meta Ads › Catalog & Commerce

Catalog Ads

How Advantage+ catalog ads work: product sets, dynamic retargeting with pixel events, catalog prospecting, creative templates, feed quality, and reports.
Guideintermediate13 min readUpdated

What a catalog ad is

Meta's documentation calls the format Advantage+ catalog ads, formerly dynamic ads. The mechanism is simple to describe. You maintain a catalog in Commerce Manager, which is a structured list of items with an id, a title, a description, an image, a price, availability, a link and a set of optional attributes. You create a campaign that references the catalog, or a product set drawn from it, and a template that says how a product should be laid out in the ad. At delivery, Meta picks the items for each person and renders the template with that person's items. Nobody designs the ad for a specific product; the system does, for every product, continuously.

Three components therefore have to be right. The catalog must be complete and accurate, because its fields are the ad's copy and imagery. The events sent by the pixel and Conversions API must carry item identifiers that match the catalog, because that is how a person is connected to a product. And the product set and audience choice must express the business logic, because the system optimises within whatever boundary you give it. The conversions API guide covers how the server-side events are set up; this page assumes events are flowing and concentrates on what the catalog format does with them.

Catalogs are not limited to retail products. Meta documents catalog types for travel (hotels, flights, destinations), property, vehicles and media, each with its own schema, and the same dynamic mechanism applies. This guide uses the retail case because it is the most common, and the logic carries across. Apparel catalogues add their own demands, such as size and colour variants and consent-dependent retargeting, which the fashion and apparel marketing guide sets out.

Product sets: segmenting the catalog

A product set is a filter on the catalog: a saved rule that selects items by attribute. Meta documents filters on fields such as category, brand, price, availability, product type, condition and the custom label fields, which are free-text attributes you fill in the feed for exactly this purpose. An ad set references one product set, and the delivery system chooses items only from within it.

Product sets are where business logic enters the format. Common patterns:

  • Margin tiers. A custom label that marks high-margin items lets you run a prospecting ad set that only shows products worth acquiring a customer with, and a separate ad set for the rest with a stricter return target.
  • Stock depth. A set that excludes items with low stock stops the ad promoting something that will be gone before the click lands.
  • Price bands. Separate sets for entry-level and premium items let you match the creative template and the message to the price point, and keep average order value (AOV) visible per set.
  • Seasonal and promotional sets. A set built on a sale label or a season attribute carries the promotion without touching the rest of the catalog.
  • Exclusions. A set that removes discontinued lines, gift cards, or items that cannot be shipped to a given market.

Sets are rules, not snapshots, so items enter and leave as the feed updates. Keep the number of sets manageable; each one attached to its own ad set splits the budget and the learning. A handful of sets that map to real differences in margin, stock or message is usually enough, and the full catalog is the right default for retargeting, where the person's behaviour already chooses the products.

Dynamic retargeting: events, windows, intent levels

Retargeting is the format's original use and still the most reliable. Meta documents a set of retargeting options that select people by what they did with catalog items, as recorded in the pixel and app events: viewed or added to basket but did not purchase, added to basket but did not purchase, upsell (viewed or bought items from one set, now shown items from another), cross-sell (bought from one set, now shown a different set), and a custom combination where you define the include and exclude events and windows yourself.

Retargeting optionWho is includedWhat they are shownTypical use
Viewed or added to basket but not purchasedPeople with a ViewContent or AddToCart event in the window and no PurchaseThe items they interacted with, and related items from the setThe default reminder rung for most retail accounts
Added to basket but not purchasedPeople with an AddToCart event in the window and no PurchaseThe items they addedThe hottest rung; short window, reassurance message
UpsellPeople who viewed or bought items from one product setItems from a different, usually higher-value, setMoving a browser of entry-level items towards premium ones
Cross-sellPeople who purchased from one product setItems from a complementary setAccessories, consumables and companion products after a purchase
Custom combinationPeople matching include and exclude events and windows you defineItems from the chosen setAnything the standard options cannot express, such as excluding a specific event

The events that make this work are the standard commerce events, ViewContent, AddToCart and Purchase, each carrying content identifiers and a content type that tell Meta which catalog items were involved. Where the identifiers on the event do not match the item ids in the catalog, the person is still in the audience but the system cannot tell which products they looked at, and it falls back to choosing items by other signals. Matching is therefore the first thing to check when retargeting shows the wrong products, and Commerce Manager's diagnostics show it directly.

Each retargeting option takes a window in days, and the window is an intent setting. A short window after a basket add is the hottest audience the account has; a longer window after a product view is a reminder audience whose conversion rate fades as it ages. The useful structure is a ladder with separate ad sets by intent level, each with its own window and its own template message: basket abandoners within days, product viewers over a few weeks, and a cross-sell set for recent purchasers. Each rung excludes purchasers within its window, which the format's own options handle, and the custom audiences guide explains how the underlying Remarketing audiences are built and exclusions are layered where the standard options are not enough. Where the shop also sends cart reminder emails, the comparison of abandoned cart emails and remarketing ads shows how to split basket abandoners between the two channels and suppress one against the other. If buyers keep seeing catalog ads for the item they just bought, the page on retargeting ads reaching recent buyers checks each rung's purchaser exclusion, the purchase event's content IDs, the windows and the cross-sell set.

Prospecting with the catalog

Meta documents a broad audience option for catalog ads, in which the delivery system shows products to people who have not interacted with your business, choosing the items for each person from the signals Meta holds about them. The documentation is clear that this relies on the catalog's attributes and on the system's learning from your pixel and app data; it works best when retargeting has already been running, because that history teaches the system which of your products convert for which kinds of people.

Prospecting with the catalog is a different proposition from prospecting with a designed ad. The ad is a product grid or a single product chosen for the person, so the creative that persuades is the product imagery and price rather than a story about the brand. That suits businesses with wide catalogs and visually self-explanatory products, and suits less well a single hero product with a complex pitch, where a designed ad will outperform. The product set matters more in prospecting than in retargeting: with no browsing history to draw on, the system is choosing from the whole set, so a set restricted to proven sellers or high-margin items gives it a better starting point than the full catalog with its long tail. Under Advantage+ audience, which applies here as to any ad set, the audience inputs are suggestions and the exclusions are controls, so exclude existing customers and recent site visitors to keep the prospecting ad set spending on strangers.

Judge catalog prospecting on new-customer cost per acquisition (CPA) and first-order margin rather than on the blended return, and give it the budget and time the learning phase needs before deciding; the audience is large and the learning is slower than in retargeting.

Creative: templates, overlays, intro cards, video

The template is the part of a catalog ad you design. Meta documents several formats: a single image or video that shows one product, a carousel where each card is a product, and a collection that opens into a full-screen Instant Experience showcasing a product set. Within each, the product image, name, price and description are pulled from the catalog, and you control the frame around them.

Overlays add catalog information on top of the product image: price, strikethrough price, a percentage discount, free delivery, or a custom label. Meta documents overlay options with position, shape and colour settings. They are the quickest way to make a price point or a promotion visible in the feed, and they only work if the underlying fields (sale price, shipping) are populated in the catalog.

Frames and templates wrap the product image with a branded border or background, which helps a catalog ad look like the brand rather than a generic grid. Meta documents template options for applying a frame across all products in a set.

Intro cards place a static image or video card at the start of a carousel, before the dynamic product cards. They are used to carry a brand or offer message that the product cards cannot, and to give the carousel a first frame that is designed rather than assembled. When the carousel should tell a curated story rather than show whatever the catalog selects, the guide to carousel ads covers building one by hand, card by card.

Video in catalog ads comes in two forms Meta documents: a video field in the catalog so that individual products can carry their own clip, and dynamic formats that assemble product imagery into motion. Both depend on the catalog carrying the assets; the template cannot invent them.

Dynamic formats and creative is Meta's option to let the system choose, per person, between a carousel, a collection or a single product, and to vary elements such as the description, based on what it predicts will perform. It is on by default in some flows and is the catalog equivalent of Advantage+ creative. Where it is on, judge the ad set as a whole rather than trying to attribute results to a format.

Illustrative example — not a real result

A homeware retailer rebuilds catalog ads around product sets

A homeware retailer with around 6,500 items had one catalog campaign showing the full catalog to a 14-day viewers audience and one broad prospecting ad set on the same full catalog, with a combined ROAS of about 2.8 against a break-even of 2.4. Diagnostics showed 18% of ViewContent events carried ids that did not match the catalog, traced to a variant-id versus parent-id inconsistency on one product template. After the fix, the account added three custom labels (margin tier, stock depth, bestseller flag) and built four product sets. Prospecting was limited to the roughly 900 bestsellers with a margin tier of high and stock above 20 units; retargeting kept the full catalog but excluded a low-stock set. A carousel template with a strikethrough price overlay and a free-delivery intro card replaced the plain grid. Over the following two months the prospecting ad set's ROAS moved from 1.9 to 3.1 and its average order value rose from £54 to £71, retargeting held at about 5.2, and the share of spend going to items that then sold out before the click fell from a noticeable fraction to almost nothing.

Feed quality and the catalog diagnostics

The catalog is the creative, so feed quality is creative quality. Meta documents several ways to supply it: a file upload, a scheduled fetch from a hosted feed URL, a connection from a supported commerce platform, the pixel's microdata on product pages, and the API. Scheduled fetches and platform connections keep prices and availability current, which matters because an ad that shows a sold-out item or a stale price wastes the click and, in the case of pricing, can breach Meta's advertising standards.

Commerce Manager's diagnostics tab lists issues at item and catalog level: required fields missing, values in the wrong format, images that fail the size or content requirements, links that return errors, and items rejected for policy reasons. It also reports the event matching described earlier, so that items which appear in pixel events but not in the catalog, or the reverse, are visible. The first job after any catalog change is to read this tab; the second is to read it on a schedule, because feeds break quietly when a platform update renames a field. For a catalog being connected for the first time, the Meta catalog setup checklist adds the event matching, product set and ownership checks to tick before the first catalog campaign.

Beyond passing diagnostics, the fields that most affect performance are the ones the template renders and the ones the system uses to choose products: clear product images on a plain background for the main image, with lifestyle alternatives in the additional image fields; titles that lead with the product and the attribute a buyer searches by rather than with the brand; accurate category and product type values; sale price alongside price when a promotion runs; and the custom labels that make product sets possible. Businesses that already maintain a Google Shopping feed will recognise most of this, and the Google Ads product feeds guide covers the shared discipline of feed hygiene in more depth; the attribute names and accepted values differ, so validate against each platform separately.

Reading product-level reports

Ads Manager reports catalog campaigns like any other at the campaign, ad set and ad level, and adds a breakdown by product id that shows impressions, clicks, spend and the catalog's conversion events per item. That breakdown is the view that makes the format manageable: it shows which items absorb spend without selling, which sell but at a poor return, and which carry the ad set. It is also where product sets prove themselves, because a set built on a margin label can be compared with the full catalog on return on ad spend (ROAS) rather than on revenue alone.

Read the product breakdown with the catalog open beside it. An item with high spend and no purchases is usually a feed problem (wrong image, broken link, stale price, out of stock after the click) before it is a demand problem. An item with purchases but a low return may still be worth keeping if it brings first-time customers who return. Items that never serve at all are often missing a required attribute, or sit outside every product set, or have been rejected in diagnostics.

The format's return is easy to misread because retargeting takes credit for purchases that would have happened anyway. Judge retargeting rungs by incremental tests or at least by comparing short-window abandoner results against longer-window viewer results, and judge prospecting on new-customer cost and margin. The ROAS calculator converts spend and revenue per product set into the comparable figure, and the commerce hub places catalog ads among the other commerce formats. The author's Meta Ads expertise page describes how catalog structures are maintained on retail accounts over time.

When particular items never appear in delivery, the diagnosis page for catalog products not showing in ads follows each item from diagnostics through product sets, content IDs, stock and feed refresh.

How catalog prospecting and dynamic retargeting fit into a complete shop account is set out in the Meta Ads setup for online sales, alongside purchase events and attribution.

When the catalogue's job is online orders, the paid social channel page for online sales sets out when catalogue ads should lead, when hero-product creative should, and how holdouts judge both.

Common mistakes

  • Launching before the events match the catalog. Mismatched ids produce the wrong products or fallback products; check diagnostics first.
  • One product set for everything. The full catalog is right for retargeting and wrong for prospecting; build sets on margin, stock and bestsellers.
  • Stale feeds. A file uploaded once drifts from reality within days; schedule a fetch or connect the platform.
  • Empty optional fields. Overlays, sets and good titles all depend on fields the feed omits; populate sale price, shipping and custom labels.
  • Judging prospecting on blended return. Retargeting inflates the blend; read new-customer cost and margin per set.
  • Ignoring the product breakdown. The ad set total hides the handful of items spending without selling.
  • Promoting items that sell out before the click. Exclude low-stock items with a product set rule.

Frequently asked questions

Why do my catalog ads show the wrong products or no products at all?

The usual cause is a mismatch between the item identifiers in your pixel and Conversions API events and the identifiers in the catalog. Dynamic retargeting picks products from the content identifiers on the ViewContent, AddToCart and Purchase events a person triggered, so if those do not match the catalog's item ids, or the content type is wrong, Meta cannot connect the person to a product. Commerce Manager's diagnostics show the event-to-catalog match, and fixing it there usually fixes the ads.

Can catalog ads reach people who have never visited my site?

Yes. Meta documents a broad audience option for catalog ads, usually called prospecting, where the delivery system chooses products for people who have shown no prior interest in your business based on signals it holds about them. It depends on the catalog carrying rich, accurate attributes and on enough purchase data for the system to learn which products appeal to whom, so it tends to work better once retargeting has been running for a while.

Do I need a different catalog for Meta and Google Shopping?

Not necessarily. Both platforms accept a product feed with a largely overlapping set of attributes, and many businesses generate one source feed and adapt it to each platform's specification. The field names, accepted values and required attributes differ enough that the feed should be validated against each platform separately, and the diagnostics on each side read independently.

Terms used on this page

Tools to try

Where this page is used

Part of these learning paths

Also related

Sources

  1. Meta Business Help Centre — Meta Business Help Centre

Related