How to Connect Your Shopify Product Feed to Video Ad Creative

Open Shopify Admin and review every product you intend to advertise: title, featured image, price, compare-at price, inventory, and variant options. Hide or unpublish discontinued SKUs, draft products, and gift cards so they cannot enter the feed. A catalog with gaps is cheaper to fix here than three weeks into a catalog campaign that is already spending.

Sort first by missing images and empty prices, then by variants. A product with five colourways and only two populated images will show broken creative to three out of five interested shoppers, and Meta and TikTok both silently deprioritise items that fail catalogue diagnostics rather than throwing an obvious Ads Manager error. Fix variant-level images and titles so each option can stand alone in a dynamic ad — 'Blue / Small' is not a title a stranger can understand in a 1:1 placement. Do this pass before you install any sales channel, or the first sync will freeze the broken state into Commerce Manager.

Decide now whether the feed should include the whole catalog or a curated subset. Stores under a few hundred SKUs usually do better feeding everything and letting the platform rotate, while stores with thousands of low-margin, made-to-order, or discontinued items should exclude them with a Shopify collection or a product tag before the sales channel syncs. Every irrelevant product in the feed is a product that can win an impression it should never have had, and catalog campaigns will not warn you that the winner was a gift card.

Install the official Facebook and Instagram sales channel (or Meta's current Shopify app), connect the ad account and Commerce Manager catalog, and sync the collection you just curated. Wait for the first full sync to finish before you build a campaign — launching against a half-synced catalog is how out-of-stock items leak into Advantage+. Confirm the catalog ID in Commerce Manager matches the one the ad account can actually use.

After the first sync, open Commerce Manager diagnostics and read the error list item by item. Missing GTIN or brand fields, price mismatches between the feed and the live product page, and images under the minimum resolution are the three issues that quietly shrink eligible inventory. For a small catalog, review every flagged SKU by hand rather than trusting the automated pass; feed delays of a day or two are common after a price change or restock, and stale data can stay live longer than you would expect.

Confirm the pixel and Conversions API are attached to the same ad account as the catalog. A catalog without purchase events can still serve ads, but Advantage+ shopping and catalog sales campaigns learn slowly without recent purchase signal — roughly 50 attributed purchases a week is the floor that makes automated delivery stable rather than noisy. If you are below that, the feed connection is still worth doing, but run a simpler conversion campaign first so the pixel has something to learn from.

Repeat the connection for TikTok Shop or TikTok catalog ads using the TikTok sales channel or catalog manager, mapping the same ad-eligible collection. Keep one source of truth in Shopify; do not maintain a second spreadsheet feed unless a channel cannot read Shopify directly. Mirror the same product tags and collections so both catalogs advertise the same eligible set.

The failure mode to avoid is two catalogs that drift. If Meta syncs from Shopify hourly and TikTok is still on a CSV you uploaded last month, you will produce video ads from live URLs that one channel can fulfil and the other cannot. Use native Shopify channels wherever they exist, set them to a regular sync (hourly is plenty for most stores), and treat a manual CSV as a last resort you re-export after every bulk price edit, not as a set-and-forget file. Write the sync source for each channel in your campaign doc so a new teammate does not 'help' by uploading another CSV.

Restricted categories — supplements, cosmetics, and anything with a claim review — can sit in pending or rejected status on one platform while they are live on another. Check each listing's compliance status individually after the first import. A batch sync can silently leave a handful of SKUs unpublished, and you will not notice until a video is live and the product card or catalog ad has nothing valid to attach to. Build a three-day buffer before a launch in those categories rather than generating video the same afternoon you hit sync.

In Commerce Manager (and the TikTok equivalent), build product sets for the cuts you actually advertise: bestsellers, a new-launch collection, a margin-safe subset, and a retargeting set of recently viewed items. Do not run the entire catalog as one undifferentiated set if half of it should never win spend. Name sets after the job, not after the date you built them.

Product sets are how you keep catalog campaigns honest. A 'bestsellers' set built on a Shopify tag or a sales-velocity rule lets Advantage+ and catalog ads concentrate on items that already convert, while a 'new launch' set isolates products that still need prospecting creative. Without sets, the algorithm will happily spend on whatever has a clean image and a low price, which is often not the SKU you can afford to acquire a customer on. Revisit set rules after any collection edit so a newly tagged draft cannot slip into bestsellers.

Handle variants as a creative decision, not only a feed decision. If colourways look meaningfully different, each variant needs its own image — and later its own video — because a shopper who loves the ad but not the swatch shown will bounce on the product page. If variants are the same product in different sizes with identical photography, advertise the parent and let the page handle size selection. Mixing those two cases in one set is how catalog ads look randomly wrong, and no amount of hook testing will fix a swatch mismatch.

Pick 5–10 products from the ad-eligible collection — bestsellers by units, plus one or two well-reviewed items with weak traffic. For each one, paste the Shopify product URL into Klip Kanvas so the title, price, images and review language become a creative brief, then generate distinct hook variants rather than one video per SKU. Skip the long tail until those heroes have a read.

Bestsellers and hidden-gem products need different scripts even though the workflow is the same. A bestseller already has demand, so the video can lean on social proof and reorder language. A hidden gem with strong reviews and low traffic usually has a discovery problem, so the script should lead with the specific benefit reviewers keep repeating — that is the angle nobody has shown this audience yet. Do not write both as generic product demos and expect the catalog to sort it out; the feed can match SKUs, not angles.

Pull briefs from the live URL, not from a spreadsheet of last season's copy. Shopify titles, compare-at prices and on-page claims change during sales, and a video generated from stale notes will disagree with the product page the ad clicks through to. That mismatch is both a conversion leak and, on some platforms, a policy risk. If a product has no usable reviews yet, write three benefit lines by hand and keep claims inside what the listing already says, then replace those lines once reviews exist.

Produce at least one AI UGC video per hero product, plus a second hook on your top two or three SKUs. Before you export, check that the spoken or on-screen product name, variant, and any price or offer line match the Shopify page as it is today — not as it was when you built the brief. A mismatch here will outlive the campaign that launched it, so catch it at export.

Price and stock drift is the Shopify-specific failure. A video cut during a site-wide promotion will keep quoting that offer after the promotion ends unless you catch it at export. Build a habit of opening the live product URL in another tab immediately before launch, not only at brief time. If you sell variable products, decide whether the video sells the family or one hero variant; selling a single bestselling colour converts better when variants look different, because the catalog ad can otherwise show a mismatch between the talking-head demo and the tile image.

Export 9:16, 1:1 and 16:9 from the same master so catalog placements in Reels, feed and in-stream do not fall back to a cropped still. Name files with SKU, hook and ratio — for example serum-hookA-9x16 — so when you read reports two weeks later you can tell which product-and-hook pair actually produced the purchase, not just which campaign did. Put that same string in the Ads Manager ad name or the feed-to-creative link breaks in reporting two weeks later, when you actually need the pair on the readout.

Upload the videos as primary creative in a catalog sales or Advantage+ shopping campaign fed by the same catalog you synced, with a couple of product stills as backup. Let the platform match product to shopper from the feed; do not build a separate ad set per SKU on a first launch unless you have fewer than roughly 10 products getting regular traffic.

Catalog-based campaigns need a handful of active products with recent purchase volume to learn. If you only have two or three consistent sellers, a full-catalog campaign will underperform a focused single-product campaign built around your best video. As a rough threshold, stores with fewer than about 10 products getting regular traffic should start with those hero videos in a conversion campaign, then move to catalog structure once inventory and order data can actually feed the algorithm. Do not treat 'we connected the feed' as the same thing as 'the feed should run the campaign.'

Load a minimum of three ad variations, ideally six or more with genuinely different hooks — not four crops of the same opening line. Advantage+ tests creative across a wider audience than a manual campaign, so a single angle gets shown to people it does not fit. Mix two or three UGC videos, one or two statics, and catalog dynamic creative if the SKU range is large enough that a shopper might need to see their specific variant. Leave the campaign alone for the first three to four days so it can gather the roughly 50 optimisation events learning wants.

Leave the Shopify channel on a scheduled sync and put a weekly 15-minute catalog check on the calendar: diagnostics in Commerce Manager, a spot-check of three live ads against their product pages, and a pass for any SKU you unpublished or discounted that week. Feeds break silently after bulk edits and app updates, which is why the check has to be calendar, not vibes.

The two events most likely to corrupt a feed are a bulk price edit and a sales-channel or pixel-app update. After either one, re-open diagnostics and fire a test purchase so you know both the catalog and the purchase event still match. A growing gap between platform-reported purchases and Shopify orders — larger than the normal 10–15% attribution mismatch — is usually a tracking or feed break, not a creative problem, and it is cheaper to catch on a Tuesday checklist than after a month of scaled spend.

When you add a new product, run the same path on that SKU only: tag it ad-eligible, wait for sync, generate a brief from the live URL, export ratios, and add the video to the existing catalog campaign rather than spinning up a new campaign per launch. The feed connection is the foundation; every future product should plug into it instead of repeating setup from scratch. If a launch SKU is still pending in Commerce Manager, do not 'temporarily' point the video at a sibling product just to save a day.