A Feed Builder feed writes a Data Builder field’s stored rows with the Use Value as Feed transform. Each run reads the stored rows instead of building them.
Set up the feed#
- Set up and fill the Data Builder field. See Set up a field.
- Create a Feed Builder feed.
- On the Source tab, set the Products setting and conditions you want. The field stores every active product; the feed decides which ones go out.
- If the field stores a value for each variant, turn on Include product variants as separate feed rows.
- On the Feed Fields tab:
- Set Output Format and Delimiter to match the field.
- Map exactly one field: the Data Builder field’s code as the source, any output name (the ready-made files use
row), and the transform Use Value as Feed.
- On the Header tab, enter the column line for CSV, or the opening XML for XML. See below.
Rules for Use Value as Feed#
- It only works when it’s the only field mapped. With other fields, the feed treats it like a normal field.
- Products or variants with no stored value are skipped.
- With variants on, each variant’s stored row is one line. If none of a product’s variants has a stored value, the product’s own row is written.
CSV feeds#
The feed can’t see the columns inside the stored rows, so it doesn’t write a column line. Put it in the Header text, in the same order as the field’s mapping, ending with {{newline}}:
id,title,description,link,image_link,price{{newline}}
For a tab-delimited feed, separate the names with {{tab}}:
id{{tab}}title{{tab}}description{{tab}}link{{tab}}image_link{{tab}}price{{newline}}
If you change the field’s mapping, change this line to match.
XML feeds#
Data Builder wraps each stored XML row in <record>, not the <item> element Google expects. For Google’s or Microsoft’s XML format, use Feed Builder alone, or the tab-delimited format with Data Builder. The ready-made Data Builder files for Google and Microsoft use the tab-delimited format.
JSON feeds#
Each stored row is one JSON object, and the feed writes them as an array.
Keeping the feed current#
The field updates its stored rows a few minutes after a change, and the feed must run after that. The ready-made files give the feed’s task a 300-second trigger delay so it runs after the field’s Queued task.
After a whole-store refresh of the field, run the feed (Process Now under Marketing > Feeds) once the field’s Queued task has stopped.
Example#
The ready-made ChatGPT Data Builder file sets up:
- a Data Builder field
chatgpt_product_feed_data: CSV, comma, a value for each variant, with the ChatGPT fields - two tasks for the field: Queued Products (enabled, triggered) and All Products (disabled)
- a Feed Builder feed
chatgpt_product_feed: Available products with a price above 0, variants on, one field mapped as Use Value as Feed, the column line in the Header text - a nightly task for the feed, triggered after changes with a 300-second delay