Product feed optimisation for Meta, the fields that matter

Feed optimisation has a bad name because most advice about it is a list of forty fields with no weighting. In practice six fields decide whether your catalog ads work, and the rest are housekeeping.

Here they are in the order they are worth your afternoon.

The six that matter

id

Stable, unique, variant level. The id is how everything joins: the pixel event, the supplementary feed, the product set, the reporting.

The failure is an id that changes. If your platform regenerates ids on a re-sync, every product looks new to Meta. Learning resets, and retargeting breaks because the pixel sends an id the catalog no longer has.

Check this once: export the catalog today, export it in a month, and compare ids for products that did not change.

availability

in stock, out of stock, available for order, preorder. Meta will not serve an out of stock product, which is the behaviour you want, and it is the field most often stale.

If your feed runs once a day, you can advertise a sold out product for up to 24 hours. Every one of those impressions is paid for and lands on a page with no buy button. See out of stock products in catalog ads.

image_link

The single most important creative decision in the account, and it sits in a spreadsheet column. Everything a shopper sees on the card except the price and the name comes from here.

Rules and sizes are in Shopify product images for Meta catalog ads.

title

Feed titles come out of your store written for search. In an ad they are truncated around 25 characters on most placements. A title that front loads the category and the colour beats one that leads with the brand.

Do not fix this in your store. Override it in a supplementary feed so the storefront title keeps doing its job for search.

price and sale_price

price is the full price. sale_price is what it costs today. Meta compares the two to decide whether there is a price drop, and the built in price drop label appears only when sale_price is lower.

Use sale_price_effective_date for a window, so a promotion starts and ends without anyone remembering to edit a file.

custom_label_0 to custom_label_4

Five free text fields that do nothing on the card and everything to your campaign structure. They are how you cut the catalog into product sets.

Useful things to put in them:

• Margin band, such as high, mid, low.
• Best seller rank band.
• Season or drop.
• Stock depth, such as deep or thin.
• Whether the product is currently discounted.

Spending on the top margin band rather than the whole catalog is a structure change, not a creative change. It is often the fastest profit improvement available.

What you can safely ignore

None of these are wrong to fill. They are just not where the money is.

The order to work in

1. Fix ids if they are unstable. Nothing else is reliable until they are.
2. Get availability updating at least every few hours.
3. Fix the images, including the variant level mismatches.
4. Override titles for the products you actually advertise.
5. Put real values in two custom labels and split the campaign by them.
6. Only then worry about the long tail of optional fields.

How to tell if the feed is the problem

Open Commerce Manager and look at the catalog's item count against your store's product count. A gap is the first clue. Then look at the data source run history for errors and warnings. Read the warnings, because they are usually the products that are not serving.

Three numbers to watch weekly:

• Items in the catalog against products in the store.
• Items with errors.
• Items out of stock.

If the third number is high and your store disagrees, the feed schedule is the problem.

You can see your feed data drawn onto a card, with the price and the rating on it. Paste your store URL into the preview.