Catalog & Dynamic Product Ads

Dynamic ads run themselves. The feed, the audience, and the exclusions behind them still need someone watching.

Catalog ads remove much of the manual work of building a thousand ads. The trouble starts when a stale feed shows sold-out products, a missing exclusion retargets a recent buyer, or a pixel mismatch breaks the personalization. We verify the feed, tracking, and exclusions before relying on the automation. For ecommerce or marketplace businesses running (or about to run) catalog and dynamic product ads on paid social.

A catalog advertising setup with scheduled checks for feed health, audience exclusions, and product-set logic before automated delivery is trusted.

A paid-social strategist reviewing catalog feed health, exclusion rules, and product-set performance side by side

Some of the 500+ brands we've worked with

See all references
  • BMW
  • BNP Paribas Cardif
  • Obilet
  • QNB Finansfaktoring
  • Bluemint
  • Koleksiyon Mobilya
  • Elle

How the work runs

Automation starts with feed health

AI runs continuous feed and match-rate checks. The catalog owner and strategist decide catalog scope, exclusion policy, and when an automated bid strategy can be trusted. AI does not change feed or exclusion rules.

How we hold ourselves to it

  • A catalog ad is only as good as the feed and pixel behind it
  • Recent buyers stay out of retargeting
  • Each product set carries a targeting or merchandising decision
  • AI flags feed and match-rate problems. A human approves catalog scope

  1. Audit the feed and the tracking behind it

    Check the product feed for accuracy, freshness, and required fields, and confirm pixel or Conversions API events are firing and matching catalog item IDs.

    Feed and tracking health record covering accuracy, freshness, required fields and pixel or Conversions API match against catalog item IDs.

    AI assist
    Runs an automated feed-quality and event-match check and flags specific broken items or fields.
    Human gate
    Catalog owner confirms which flagged issues get fixed before launch.
    Owners
    Catalog owner + media specialist

  2. Build product sets on purpose

    Group products into sets (best sellers, category, sale items) that carry a real targeting or merchandising decision, and document the rationale behind each one.

    Product sets with the targeting or merchandising decision behind each one documented.

    AI assist
    Proposes a product-set structure from catalog metadata and sales data.
    Human gate
    Ecommerce or catalog owner approves the product-set logic.
    Owners
    Catalog owner + merchandising lead

  3. Set exclusion rules for recent buyers and edge cases

    Decide who gets excluded from seeing ads for a product they just bought, and how out-of-stock or discontinued items get pulled from rotation automatically.

    Exclusion rules covering recent buyers and the automatic removal of out-of-stock or discontinued items.

    AI assist
    Drafts the exclusion rule set and flags product sets with no exclusion attached.
    Human gate
    Strategist approves every exclusion rule before it's built.
    Owners
    Strategist + catalog owner

  4. Build and validate the campaign, paused

    Configure the catalog campaign (audience, product sets, exclusions, creative template), while paused, and verify it against the feed and rule set before it can spend.

    Paused-build QA record verifying audience, product sets, exclusions and creative template against the rule set.

    AI assist
    Compares the paused build against the approved feed, sets, and exclusions, and flags mismatches.
    Human gate
    A second reviewer signs off on the paused build before launch.
    Owners
    Media specialist + second reviewer

  5. Let it run, and keep watching the feed

    Monitor feed health and match rate continuously alongside performance, since a personalization engine can look fine on the surface while quietly serving stale data.

    Continuous feed-health and match-rate monitoring recorded alongside delivery.

    AI assist
    Flags feed errors, dropping match rate, or a sudden shift in which products are serving.
    Human gate
    An authorized media owner approves any change to product sets, exclusions, or budget.
    Owners
    Media lead + catalog owner

  6. Review catalog health and campaign performance together

    Check whether feed quality, exclusions, and product-set logic still hold, and whether the campaign's results reflect real catalog performance or a data problem.

    Joint review separating a real catalog result from a data problem, with the decision recorded.

    AI assist
    Drafts a combined feed-health and performance review from the monitoring data.
    Human gate
    Catalog owner and strategist approve any change to scope or rules.
    Owners
    Catalog owner + strategist

What lands with your team

A record behind every automated campaign

Before anyone trusts a catalog campaign's numbers, there's a record showing the feed and tracking behind it were actually checked.

  • Feed and tracking health record

    Feed accuracy, freshness, and pixel or Conversions API match rate, checked against catalog item IDs.

    Accepted when

    Feed accuracy, freshness and match rate are checked against catalog item IDs, not assumed from a green dashboard.

    Cadence: Set schedule, per change

  • Product-set and exclusion register

    Every product set's logic and the exclusion rules attached to it, including recent-buyer and out-of-stock handling.

    Accepted when

    Every set states its logic and the exclusions attached to it, including recent-buyer and out-of-stock handling.

    Cadence: Updated on strategy change

  • Paused-build QA record

    Evidence the live campaign matches the approved feed, sets, and exclusions before spend started.

    Accepted when

    The live campaign was verified against the approved feed, sets and exclusions before the first impression was bought.

    Cadence: Per launch or rebuild

  • Catalog and campaign decision log

    What changed in the feed, sets, or exclusions, why, and how performance responded.

    Accepted when

    Each change to the feed, sets or exclusions is recorded with its reason and how delivery responded.

    Cadence: Monthly, or after incidents

Before the work starts

When this work is the right next step

The automation is only as honest as the feed and the exclusions behind it. Catalog ads can look effortless from the outside. Upload a feed, connect a pixel, and let the system personalize. Yet a stale price, a missing exclusion, or a broken event match rarely causes an obvious failure. The campaign keeps showing the wrong product or retargets someone who already checked out.

A good fit when

  • Your catalog has too many products for manual ad builds, but the feed still needs a repeatable way to keep prices and stock current.
  • Pixel or Conversions API events are firing, yet nobody can tell whether their item IDs still match the products in the catalog.
  • Feed errors keep reaching campaigns without a clear response, so sold-out items, old prices, or broken links can continue serving after the source data changes.
  • Stale prices, sold-out items, or broken links remain unchecked in the feed.
  • Recent buyers keep seeing the exact product they purchased.
  • Catalog item IDs do not reliably match pixel or Conversions API events.
  • Product sets still reflect an older catalog.

Better handled as other work when

  • A few manual ads already cover the catalog, so product-set automation would add upkeep without solving a real scale problem.
  • No working pixel or event tracking connects product views and purchases to catalog item IDs, so personalization would run on missing signals.
  • Product data changes faster than the feed can be trusted, and stale prices or discontinued items would keep reaching live ads.

People who own a single channel

Paid Search, Paid Social, CRO, and Programmatic each run under a named owner at Zeo. The consultants below are matched to the channel this page is about, so you can see who you'd actually work with.

Tools we use

Tools behind this work

  • DataFeedWatch

    keeps the feed itself accurate in real time, which is what the automation is actually trusting

Bring the feed behind the ads

Find out whether the automation can trust its inputs

Start with the feed, match rate, recent-buyer exclusions, and product sets. We'll identify where weak data may be shaping delivery and performance.
Check your catalog setup

The questions we hear most

We audit the feed against the platform's needs, including accuracy, freshness, and required fields. The underlying feed infrastructure usually sits with an ecommerce platform or developer. If building or maintaining it falls outside scope, we'll name that explicitly. You still get a record of what needs fixing before catalog ads rely on it.