Create a Data Feed for Content Search

Your shop search does more than find products; customers also expect inspiration and advice (blog posts, topic pages, videos, datasheets). FactFinder NG can index and rank this content alongside products. This guide shows how to structure a reliable, lean data feed for Content Search: which fields are essential, which are optional, and how to keep results relevant.

Why do I need a data feed?

The data feed is the source FactFinder NG indexes and compares to each search term. If a field is not in the feed (or not indexed), it cannot be found. Provide content data in a structured CSV so NG can import and update it reliably.

Prerequisites:

• Maintain content in a separate NG channel from products

• Define which columns are searchable, filterable, sortable in the channel schema

• Automate feed generation and import on a schedule

How should the CSV file for Content Search be structured?

A content feed differs from a product feed. Keep it lean: do not index the full article body. Long bodies add noise and hurt precision. Instead, index structured metadata and short teasers that capture intent.

Relevant fields for your content data feed

Must‑haves

• Title: Human‑readable title shown in results. Mark as searchable and boost modestly.

• URL: Destination URL for the content. Without it, results cannot be clicked.

• Image URL: Thumbnail shown in suggest/results.

• Teaser/Short description: One or two sentences that summarize the content; great for recall without adding noise.

• Categories/Headings: Taxonomy values used for filters (facets). Mark as filterable.

• SEO title / SEO description: Often contain the most important keywords; index both but give lower weight than Title.

• Tags/Keywords: Synonyms, colloquial terms. Curate from analytics to capture how customers search.

Nice‑to‑haves

• Headings/Subheadings: Additional query terms without the verbosity of full body text.

• Rating: For display and as a ranking tie‑breaker.

• Publication date: For freshness sorting and filters.

• Content type: e.g., blog, video, datasheet, recipe. Useful for filters and badges.

Example content data feed excerpt

Suggested CSV columns (example):

title	url	image_url	teaser	category	seo_title	seo_description	tags	rating	publish_date	content_type
How to choose the right drill	https://shop.example.com/magazine/choose-drill	https://cdn.example.com/img/drill-guide.jpg	Step‑by‑step guide to drill types and use cases	DIY>Tools	Drill buying guide	Learn which drill fits your needs	drill;power tool;DIY	4.6	2025-04-10	blog
Store opening hours	https://shop.example.com/stores	https://cdn.example.com/img/stores.jpg	Find hours, locations and services	Stores	Store hours and locations	Visit your nearest store today	opening times;store locator		2025-01-05	page

Other best practices

• Use a consistent field separator (semicolon recommended) and quote every field.

• Ensure the separator does not occur unescaped in field values.

• Encode as UTF‑8.

• Provide as plain CSV or ZIP to save bandwidth.

• Keep products and content in separate channels with separate feeds. Include the channel name in the filename, e.g., ng_export_de_content.csv.

• One record per line; avoid line breaks inside fields.

• Ensure each row has the same number of fields.

Indexing tips:

• Mark only meaningful text fields as searchable (e.g., title, teaser, headings, seo_*, tags)

• Mark taxonomy fields as filterable/faceted (e.g., category, content_type)

• Use recency and rating as secondary ranking signals, not primary boosts

Conclusion: Better data = better Content Search.

Even the best search can only retrieve what you index. Strong, lean metadata keeps results relevant and maintainable. Questions? Contact us: https://www.fact-finder.de/kontakt.html

For examples and ideas, see our blog: https://www.fact-finder.com/blog/ecommerce-content-search/