Catalogue Retargeting and Advantage+ Catalogue Ads

The number we ask for before quoting is catalogue match rate: the share of your purchase and view-content events whose content_id resolves to an item in the catalogue. Below about ninety percent, dynamic ads degrade quietly. They keep spending, they keep reporting a return, they simply stop showing the item the person looked at. The cause is usually banal — the site fires a variant SKU while the feed publishes parent IDs. Roughly half the accounts we audit have this, and it has generally been running for years. Feed engineering: Scheduled fetch every four hours for a stable catalogue; the Catalog Batch API where price or stock moves faster than that. Out-of-stock items are the quiet tax, so we hold the served out-of-stock rate under two percent and alert when it breaks — an ad for a lounge suite with a fourteen-week lead time is an advertisement for your competitor. Titles and descriptions written for the ad unit rather than lifted from the product page. Image fields checked against what the crop does at one-to-one. Custom labels carrying the fields the campaign actually needs: contribution margin band, freight band, clearance flag. Product sets built from margin, not from the menu: Your category tree exists for shoppers. A product set exists to control spend. We build sets from margin bands and sell-through, so the campaign cannot quietly put the entire budget behind the lowest-margin item in the range for the sole reason that it converts most easily. Exclusions and the replacement cycle: Excluding purchasers for thirty days is the default and it is wrong for anything durable. A mattress buyer is not back in market for years, so the post-purchase job is a different catalogue entirely — bases, protectors, bedding. We build cross-sell sets from actual co-purchase pairs in your order table rather than from the platform's generic cross-sell logic, which knows nothing about your range and will happily show a second sofa to someone who just bought one. The holdout, which is the point of the engagement: Retargeting reports the best return in every account and deserves the least trust, because much of it buys people who were already coming back. We run a four-week holdout, user-level where the audience supports it and matched geographies where it does not, and report incremental revenue beside attributed revenue. In the accounts where we have run it, a meaningful share of retargeting-attributed revenue survives the test and a meaningful share does not. Which is which is the finding you are paying for, and it moves the budget in one direction or the other every time. Broad versus bound: Advantage+ catalogue ads served to a broad audience usually carry more incremental volume than the remarketing-only setup, and most of the budget belongs there. We frequently finish this engagement recommending your retargeting spend goes down while catalogue prospecting goes up. If that conclusion is unacceptable in advance, do not engage us — the test will produce it anyway and you will have paid for an argument. Not included: Creative production, product photography, site merchandising, and Google Shopping feeds. The same source feed usually serves both, and we will flag what our changes break on the Google side, but fixing it there is not our work. Who this is not for: Sites under roughly sixty thousand sessions a month, where the audience pools cannot sustain distinct sets and one broad campaign does the entire job. We will say that in the audit and decline the build rather than sell you segmentation you cannot fill. Catalogues whose structure changes seasonally without version control. And anyone wanting a thirty-day abandoned cart window running permanently — we cap it, and we will show you the decay curve that explains why.