Reviews Aggregation, 8 sources, one widget

Why aggregation beats single-source, and where it stops

The case for showing more reviews is well documented. The Spiegel Research Center at Northwestern found that purchase likelihood for a product with five reviews was 270% greater than for a product with none, that the effect was larger for higher-priced products, and that purchase likelihood tended to peak at ratings between 4.0 and 4.7 rather than at a perfect 5. Their work also found that identifying reviewers with a verified buyer badge improved the odds of purchase by 15%.

Aggregation helps because different sources answer different shopper questions. A Trustpilot or Google score answers "is this company reliable, will my order arrive?" Product reviews answer "does this jacket run small?" Photo and video reviews answer "what does it look like on someone like me?" A shopper who sees all three on the PDP has fewer reasons to leave the page to check. The piece on Trustpilot, Feefo and Google reviews compared covers what each platform is good for.

Where aggregation stops is the number. A company-level rating and a product-level rating measure different things, and averaging them produces a figure that describes nothing. It also produces a figure you cannot legitimately mark up, which chapter three covers.

The source map

Before you design the display, map what you have. The table below covers the common sources for a Shopify brand and how each can be used. The Product Ratings column reflects Google Merchant Center's published list of supported aggregators, which includes Judge.me, Yotpo, Okendo, Loox, Stamped, Trustpilot, Reviews.io, Feefo, Bazaarvoice and PowerReviews among others.

In Idukki, the sources that are live in the product today are Google Reviews, Trustpilot and TripAdvisor imports, a Judge.me connection that imports product reviews with their photos and video and tags them to the matching products, and first-party rated uploads from customers. All of them land in one reviews inbox that can be filtered by status, star rating, verified buyer and source. The reviews page shows how they display, and Google reviews as shoppable content covers the display patterns that work for company-level sources.

JSON-LD that survives Google's guidelines

Google's review snippet guidelines contain four rules that decide what an aggregated display may mark up. Reviews and ratings you mark up must be readily available to users on the marked-up page. AggregateRating needs a ratingValue and at least one of ratingCount or reviewCount. For LocalBusiness and Organization markup, reviews the business controls about itself are ineligible for stars, including when they arrive via an embedded third-party widget such as Google Business reviews. And, in plain words: don't aggregate reviews or ratings from other websites.

Taken together, the safe pattern is narrow. Each PDP carries Product markup whose AggregateRating is computed only from first-party reviews of that product, shown on that page. Company-level scores from Google or Trustpilot appear visually, labelled with their source, and stay out of the markup. If your display tool emits JSON-LD automatically, check what it counts: a product aggregate that quietly includes imported third-party reviews is the pattern the guideline rules out. The wider schema reference is the review schema and rich results handbook.

The feed side, including the 50-review floor and the monthly refresh, is covered step by step in the Google Shopping stars playbook.

Moderation and sentiment per source

Aggregating sources multiplies the moderation surface. The rules below separate what you should filter from what you must not.

• Filter spam and abuse. Profanity, hate, personal data, links to unrelated sites and obvious bot text. Google's Product Ratings policies also exclude reviews primarily generated by automated programs or AI, and low-quality or off-topic reviews.
• Filter conflicts of interest. Reviews from staff, agencies or anyone with a vested interest. The FTC rule prohibits insider reviews that do not clearly disclose the material connection.
• Do not filter by sentiment. Hiding one- and two-star reviews because they are negative is review suppression. Use sentiment to route reviews (a one-star review with a delivery complaint goes to support) and to sort, never to decide what the public sees.
• Label the source. A shopper should be able to tell a Trustpilot company review from a product review written by a verified buyer. Source labels also make the display easier to defend if anyone asks.
• Keep it consistent across sources. One moderation policy applied to every import, written down, so the decision on any single review can be explained later.

Idukki's pipeline scores sentiment and applies profanity filtering on import, and hiding a review in the inbox also removes it from the public rating aggregate. That is the right behaviour for spam and abuse, and exactly why hiding should never be used for ordinary negative reviews. The broader operating model is in UGC moderation best practices.

Post-purchase triggers

Imported third-party reviews make the page more convincing, but the only reviews you can mark up are your own product reviews. Growing that first-party library is the long-term job, and it runs on one well-timed request to every buyer.

On Shopify, Idukki's review request email can be triggered from the order or from its fulfilment status (in transit, out for delivery or delivered) with a delay in days, and links to a hosted form that accepts photos and video. It is off by default and requires both the merchant's opt-in and Idukki's approval before any shopper is emailed. For the message itself, see how to ask customers for reviews.

Comparing the cost

Vendor pricing changes too often to print, so rather than a price table, here is the worksheet. Fill in each line for the platforms you are considering, over the same period and order volume.

A cost comparison that ignores the tools you retire will always favour the single-source app. The build-versus-buy model in the cost of UGC, build versus buy works through the same logic for the wider UGC stack.