Multi-source reviews without breaking Google markup

Google expects one authoritative aggregate per product page (for the review type you claim), with consistent ratingValue, reviewCount, and reviews that belong to that SKU or group definition. When two apps inject schema:

• Counts double (site reviews + syndicated reviews counted twice)
• itemReviewed points at different @id values (product URL vs SKU vs ProductGroup)
• Video review widgets add a second Review array in a shadow DOM the theme does not know about

The page looks fine to shoppers. Rich result testing tools show red errors. AI search crawlers may pick the wrong snippet entirely.

If you are mid-migration, read our 30-day parallel run for reviews and UGC replatforming before you toggle two apps live on production.

Source-of-truth rules (pick one, document it)

Before touching Liquid or tag managers, write a one-page policy:

1. 1Which system owns on-site text reviews for schema purposes?
2. 2Which system owns syndicated/off-site counts—if any—and are they eligible for on-page schema under Google’s guidelines?
3. 3Who disables the loser app’s JSON-LD output?

Common patterns:

Budget comparisons for video-capable apps are in our Judge.me vs Loox vs Okendo buyer’s guide—but the markup owner is a strategy call, not a feature checkbox.

Parallel run: validate markup weekly, not at launch

During parallel operation, run the same four URLs through Rich Results Test every Friday:

• Best-selling SKU with syndicated reviews
• Long-tail SKU with only on-site reviews
• Bundle or ProductGroup parent
• Variant-heavy SKU with query-param URLs

Log: reviewCount, ratingValue, presence of duplicate @type": "AggregateRating", and whether review arrays reference verified purchasers.

Disable schema emission in the non-owner app using vendor settings first; if the vendor hides JSON-LD behind a paid tier, use theme guards that strip duplicate scripts (document the strip in runbooks so the next agency does not “fix” it back).

Idukki sits adjacent to this stack—we surface UGC and shoppable proof, not star aggregates—so our docs on gallery embeds and theme placement assume your review app already owns honest rating schema.

UGC galleries and reviews: do not merge counts in JSON-LD

Teams sometimes want Instagram posts beside reviews and a single “social proof score.” That belongs in merchandising, not structured data. Mixing UGC likes into reviewCount is how you lose trust with regulators and search quality teams.

Keep:

• Reviews → review app schema
• UGC → visual proof modules, optional VideoObject where appropriate, separate from star aggregates

Our compliance checklist for FTC and fake reviews covers disclosure on UGC; it does not authorize inventing review counts from posts.

Headless and composable storefronts

If 2027 includes a Hydrogen or custom front end, stop generating JSON-LD in two layers (edge worker + legacy theme snippet). Centralize product schema in the product API response with a field structuredDataOwner: bazaarvoice | okendo | internal.

Public caching makes this harder: a stale aggregate at the CDN edge is worse than no stars. Follow etag patterns we outlined for headless UGC in headless UGC and public JSON etags—the same discipline applies to review payloads.

AI search and citation hygiene

Large language models and AI-overviews often quote on-page text plus structured fields. When two aggregates disagree, citations drift (“4.2 stars” vs “4.8 stars” for the same URL).

Align visible on-page copy with schema:

• If syndicated quotes appear in UI, label source clearly
• If schema excludes syndicated counts, UI should not imply they are included in the big star header

Our PDP citation checklist for AI search commerce treats rating consistency as a first-class field.

Operational checklist before you turn off the old app

• Search Console → Enhancements → Review snippets: note open issues
• Export review count by SKU from old and new systems; reconcile >5% deltas
• Paid search extensions using star extensions match on-page aggregate
• Email templates pulling dynamic stars updated to new API
• Remove old app embed scripts that injected hidden JSON-LD

For broader stack cleanup after peak, pair this with the post-holiday Shopify stack audit.

When Idukki enters the picture

We make Idukki to map customer posts and shoppable clips to SKUs—not to replace your review aggregator. In multi-source projects, Idukki widgets should sit below or beside the review block, sharing SKU keys but not schema types.

Rights for UGC remain separate from review submission flows; see help center guidance on permissions. Benchmarks and case study formats—not fabricated lift—live on resources.

Bottom line

Multi-source reviews work when one system owns markup, syndication is explicit in UI, and parallel runs are boring weekly rituals. Everything else is a rich-result lottery—and a compliance footgun. Fix the owner, strip the duplicates, and keep UGC out of your star counts.