FAQ

That’s what the brand voice input is for — and the output is scored on brand adherence every run, so you can see how well it matched. The quality of your result tracks the quality of your input: see Teaching enrichment your brand voice.

You don’t need a formal brand book. Answer the questions on the brand voice page — personality, tone, do’s and don’ts, vocabulary — and paste excerpts of copy that already sounds right. That’s enough to start, and you can refine after your first run.

No — and it’s checked. Every generated field is reviewed against your source data, and unsupported claims are the most heavily penalized issue in quality scoring. Content the engine can’t ground in your data gets flagged, not shipped.

Yes. Accept everything, accept everything above your quality threshold, or go product by product. See Reviewing and accepting results.

Yes. Enriched content is applied as a raw data override, so you adjust it with the tools you already use: a transformed data override for a single product, or a transformer for a pattern across many. See Editing enriched content.

Yes, and the scores tell you where to focus — you can re-run based on how the output scored, targeting products that fell below your threshold. Change something first (brand voice, config, or source data), since re-running the same setup produces similar results. See Re-running enrichments.

Yes — list restricted claims in your brand voice input and they become hard rules. See the do’s and don’ts section.

A primary ID needs to be set on the field execution order screen before enrichments can be run. New databases don’t have one set, so this is a required first step on every new database — set the primary ID there, then start your run.

No. Enrichment applies its output as a raw data override, so your imported source data is never replaced and you can always compare the two. See Enrichment, transformers, and exports.

Either — you choose per batch. Product level generates one set of content that all variants share; variant level generates content for each variant. Variant level sends far more items, so it uses more of your allotment. See Product level or variant level.

Not today. Filling structured attribute values is attribution, which is not part of data enrichment today but is on the roadmap. Enrichment generates copy, and it won’t infer facts your source data doesn’t contain. See What enrichment does (and doesn’t do).

Branded copy, SEO and discovery metadata, AEO content, and per-channel marketing fields — you choose which in your config. Every field and its purpose is listed in Fields in an enrichment run.

A real product name plus factual substance — materials, features, use cases. The fuller and more accurate your source data, the better the output. See What makes good source data.

Yes, untouched. Enrichment writes to new, dedicated fields rather than changing existing ones, so no transformer’s inputs change and your first enrichment alters nothing downstream. You map enriched fields into your exports when you’re ready. See Enrichment, transformers, and exports.

Include them in a new batch, reusing a config that’s already proven on similar products. Most teams run a standing pass when new collections land. See Re-running enrichments.

Answer pending — do not publish this page without it.

Google, Meta, TikTok, Amazon, eBay, and Walmart — plus storefront, SEO, and AI-answer-engine content from the same run. See Where enriched fields go.

Changing feed content does affect how channels serve your products — which is exactly why rollout is under your control, per field and per channel, and why we recommend a proper A/B test instead of a big-bang switch.

Yes — enriched fields only go live where your export mappings send them. Test on Google before touching Meta, or vice versa.

Baseline before you switch, roll out to a segment with a holdout, and measure CTR, conversion, and ROAS per group. The full method: Proving enrichment works.

Because tagging is required and doesn’t happen automatically. Analytics identifies enriched products from UTM parameters on your feed product URLs (for the Traffic report) and custom labels in Google Merchant Center (for the Paid Media report). See Measuring impact in Analytics.

Three factors — accuracy, consistency, and brand adherence — backed by specific, per-field issues you can inspect. See Understanding your quality scores.

Yes — usage counts enrichments sent, not accepted. Tune your config on small runs before running at volume. See Usage, billing, and limits.

Enrichment stops — there are no overage charges. Contact your account team to extend your allotment.

Yes, as often as you like — each rerun counts toward usage.