First, measure reviews per delivered order
Count the right thing. Divide new reviews by orders delivered in the same period, not by orders placed, since an undelivered order cannot produce a review. Then split the count by where each review came from: the email or text request, an on-site prompt, the insert card or a link the customer found alone. The reviews and social proof guide lists these sources and what each needs; tag each request and the card's link so the source travels with every review.
For the email channel, read the request as a small funnel: messages sent, opened, clicked through to the form, forms completed. A gap between clicks and completed forms points to effort; few sends points to a missing or broken trigger; requests opened weeks before anyone could have used the product point to timing. For the card, compare reviews arriving through its tagged link or code with the orders that carried it. Finally, check how many submitted reviews were published. A store that collects reviews and then hides or loses them has a different problem from one that never collects them, and the table separates the two. Track the per-order rate as a standing KPIA metric tied directly to an objective. so that a future drop is noticed in weeks rather than seasons.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Requests are opened but rarely acted on; replies mention the item has not arrived | The request's trigger event and the wait before sending | Requests sent before the product has arrived or been used | Start the clock from delivery and wait for first use, per product |
| Many clicks through to the form, few completed reviews | The steps between the email click and a submitted review | A request that takes too much effort: a login, a long form or no direct link | A one-tap rating from the message into a short, pre-filled form |
| Few or no review requests sent at all | Whether a post-purchase flow exists and which orders enter it | No post-purchase flow carrying the request at the right moment | Build the request into the post-purchase flow on the delivery event |
| The insert card's link or code is almost never used | What else the card asks for alongside the review | An insert card asking for too many things at once | Give each card one job and time any review ask for after first use |
| Reviews were removed, the page shows only top ratings, or shoppers question them | Moderation rules, incentive wording and who receives requests | Incentives or filtering that break review rules, so reviews are removed or distrusted | Ask every buyer, keep incentives unconditional and disclosed, publish all genuine reviews |
| The review tool holds reviews, yet product pages and search results look empty | Whether reviews render on the page and match the rating markup | Reviews collected but not shown on the page or in markup, so products look empty | Display reviews beside the price and mark up only what the page shows |
Fix, verify and prevent each cause
Asked before there was anything to say
A request sent at dispatch, or timed from the order date, reaches the customer while the parcel is still travelling. The reviews guide explains that the request lands best after use, a few days for clothing and longer for anything that needs installing, and the post-purchase flow tutorial starts the clock for the request from the delivery event, holding the message when no delivery event arrives.
Fix: move the trigger to delivery and set the wait per product group from how long first use takes.
Verify: the share of opened requests that end in a submitted review rises, and reviews describe use rather than packaging.
Prevent: review the waits whenever a new product type launches, since one setting for the whole catalogue drifts out of step with what customers buy.
A request that asks too much
Each extra step between the message and a submitted review loses people: a login page, a form with many required fields, or a link to a generic page where the customer must find the product. The reviews guide asks for a request that names the product, shows its image, makes the first click record the rating and opens a form that is already half complete, and its source table notes that an on-site prompt needs a login or a verified-purchase link.
Fix: send a direct link tied to the order that records the star rating on the first tap, then asks one question that invites detail.
Verify: completions as a share of form visits rise.
Prevent: test the request on a phone after every change to the review tool, from the message to the thank-you screen.
No flow carrying the request
Some stores intend to ask and never built the message, or built it so that only some orders enter. The post-purchase tutorial places the review request inside a flow triggered by real order progress, decides with whoever owns compliance whether it counts as a service or a marketing message, and skips it when the customer has already reviewed, returned the item or has an open complaint.
Fix: add the request as a step of the post-purchase flow, with one reminder and then a stop.
Verify: requests sent per delivered order approach the share of customers the flow can lawfully reach.
Prevent: give the flow a named owner and a review date, as the tutorial asks, so a platform migration cannot switch it off unnoticed.
A card with five jobs
An insert reaches every buyer, which makes it tempting to load with everything. The packaging and order inserts guide warns that a card carrying a discount, a social handle, a review request, a newsletter offer and a referral link gives the customer five decisions and no reason to make any of them. It also notes that the card arrives before use, so a card telling the customer when to review usually works better than one asking on opening.
Fix: give the card one job; where that job is the review, ask for an honest review after first use, with a tagged link.
Verify: reviews arriving through the card's link rise against orders carrying it.
Prevent: choose each card's job per order type and compare cards through their tagged links, as the inserts guide describes.
Reviews lost to the rules
Reviews can be lost after collection: those gathered with a reward tied to sentiment may have to come down, and shoppers discount a page where every rating is perfect. The reviews guide summarises the FTC's business guidance, set out in its Endorsement Guides questions and answers: an incentive may be offered for reviewing but not for a positive review, must not depend on what is written, and must be disclosed; asking only likely-happy customers or publishing by rating misrepresents the picture as surely as inventing reviews. The inserts guide adds that the CMA has made clear that faking reviews or suppressing genuine negative ones is unacceptable. This page is educational, not legal advice; check the rules in each market.
Fix: send the request to every buyer, remove any reward tied to sentiment, label incentivised reviews, and limit moderation to abuse, spam and personal data under a published policy.
Verify: the published share of submitted reviews rises and the page shows the critical reviews alongside the rest.
Prevent: have the review wording, incentives and moderation policy approved together whenever any one of them changes.
Collected but invisible
If the review tool holds reviews and the product page still looks empty, the display failed. The reviews guide puts the star summary and count beside the price and gives customer photos, a form of UGCContent about a brand made by customers or creators rather than by the brand itself., a gallery near the product images, because reviews hidden under a tab change few decisions. The product structured data tutorial adds ratings to the markup only where real reviews are shown, with the same count and score the page displays, following Google's Product structured data documentation.
Fix: render reviews in the page itself and generate the rating markup from the same data.
Verify: a sample of product pages shows reviews without a click, and Google's Rich Results Test reads the same figures.
Prevent: add the rating markup to the template checks that run after every theme or review tool update.
A prevention routine for the review programme
Treat reviews per delivered order as a figure the store reads every month, split by source, beside the published share of submissions. Before any change to the review tool, the post-purchase flow or the insert card, walk the whole path from delivery to a published review on a phone. When new products launch, set their wait before the request from how long first use takes, and add them to the flow's product branches.
Keep the rules part of the routine rather than a one-off check. Whoever owns the programme should be able to show that every buyer is asked, that no reward depends on sentiment, that incentivised reviews carry a label and that moderation removes only what the published policy names. Read the reviews themselves, too: the reviews guide treats their language and recurring complaints as material for product copy, and a programme that collects reviews and never reads them wastes half their value. The effect worth tracking downstream is conversionShare of visits or clicks that convert. on product pages that gained reviews. The scope stays with products sold on the store: reviews of the business on Google belong to the local reviews and reputation management guide, and other store symptoms sit in the e-commerce troubleshooting section.
Skrimble Board Games moves its request after the first game night
Skrimble Board Games, an invented online board game shop, delivers about 1,200 orders a month and gains 14 new reviews, a rate of 1.2 per 100 delivered orders. Split by source, 11 come from the email request and 3 from the insert card. The email goes out at dispatch, when the average order is still 3 days from the door, and its link opens a review page that requires an account: 9% of recipients click, and 13% of those finish. The insert card carries a discount for the next order, a social handle, a newsletter sign-up and the review ask together. The shop moves the email to 10 days after the delivery event, swaps the link for a one-tap star rating that opens a pre-filled form, and reduces the card to one line: "Played it yet? Tell other players what you honestly thought," with a tagged link. It also adds the rating summary beside the price, where it had sat under a tab. Two months later, clicks reach 17%, completions 58%, and new reviews reach 61 a month, 5.1 per 100 delivered orders, with 9 arriving through the card.
Common mistakes
Counting reviews against orders placed rather than delivered. Fixing the email when the reviews are lost in moderation. Adding a reward for good reviews to lift the count. Leaving reviews under a tab and adding rating markup the page does not show.
