Enrichment

Enrichment turns scoring insights into action. For each product, Lumio generates content designed to make AI shopping agents confidently recommend the product. Enrichment is included on the Professional plan and above.

What gets generated

Five content types per product, in a single pass:

Optimized titles

Titles get restructured into the format AI agents prefer: [Brand] + [Product Type] + [Key Attribute] + [Variant]. This is structure for matching, not SEO keyword stuffing.

Before: “The Explorer Boot” After: “Canopy Explorer Waterproof Leather Hiking Boot — Men’s, Brown, Wide Fit”

Attribute-dense descriptions

Descriptions are rewritten to include the specific attributes AI agents look for: materials, dimensions, compatibility, use cases, differentiators. Factual and attribute-rich, not marketing copy. The tone follows the workspace voice rules and brand profile.

Q&A pairs

5–8 question/answer pairs per product matching the conversational queries real shoppers ask AI assistants — sizing, materials, care, compatibility, use cases, comparisons.

These are critical for AI discovery because they match how people actually shop with AI (“Will this boot fit wide feet?” “Is it good for rocky trails?”).

Usage scenarios

4–6 “best for…” or “ideal if…” scenarios that bridge the gap between what a product is and what someone needs.

Category attributes

5–10 vertical-specific attributes that are missing from current data. Apparel: fabric, fit, inseam. Electronics: battery life, connectivity, compatibility. Beauty: skin type, key ingredients, application method.

How enrichment uses brand context

Every enrichment draws on four inputs from the workspace:

1. Brand adjectives — Shape the tone and word choice
2. Customer persona — Content is written to resonate with the target buyer
3. Product vertical — Category-specific attributes and Q&A topics are prioritized
4. Voice rules — Active rules are injected directly into the prompt as constraints

Scoring also matters: enrichment reads each product’s score breakdown to target the weakest dimensions first.

Workflow

Prerequisite

Products must be scored before they can be enriched. Bulk enrichment with unscored products returns a clear error.

Bulk enrichment

From Enrichment, click Enrich catalog to process every product. Bulk enrichment runs through Anthropic’s Message Batches API for cost-efficient processing — products are batched at 500 per request. The job runs in the background and an email arrives when it finishes.

If a job is interrupted, the next run resumes from where it left off.

Single-product enrichment

From the Scores or Enrichment view, click Re-enrich on any product to regenerate content for just that product. This call is synchronous — results appear immediately. Useful for reviewing output quality before running a full catalog pass.

Feedback loop

Enrichments can be rated thumbs up/down, commented on, or edited inline. Two ways feedback flows back:

• Inline edits apply directly. No regeneration, no token cost.
• Apply feedback triggers a job that re-enriches only the products with thumbs-down ratings, using the feedback as context. Inline-edited products are skipped — the edit is the answer.

Persistent per-product memory (approved/rejected/edited content per type) carries forward into future enrichments so the system learns from past corrections.

How usage is measured

Enrichment is included on Professional and Elite plans, billed per workspace by product limit rather than per credit. Token costs are absorbed into the plan price. Products beyond the plan limit are billed at the per-product overage rate. See Plans & pricing.

Delivering enrichments

Once products are enriched, push the content to where it matters:

• Shopify — Push to Shopify writes enrichments to product metafields. The Lumio theme extension renders them as JSON-LD on product pages, merged with live Shopify data (price, availability) at render time. Push progress streams over Server-Sent Events.
• Google Merchant Center — Coming soon.
• CSV / TSV export — Coming soon.