Skip to content
Product
PlatformPricingDimensionsIntegrationsSecurityCompare
Solutions · by use case
Customer, guest and fan experienceProduct feedback and prioritizationRevenue, churn and retentionAgenciesContact center performanceAI agent and chatbot analyticsMulti-location performanceReviews and competitive benchmarking
Solutions · by industry
Hotels and hospitalityRestaurants and food serviceFitness and leisureRetail and e-commerceSports and live entertainmentInsuranceBanking and financial servicesSoftware and SaaSAutomotiveConsumer products and CPGContact centers and BPOsPublic sector and nonprofitReal estateGaming and betting
Learn
Artifacts GalleryCustomer storiesResearchBlogThought leadershipLessons Log in Start now
Blog

How CPG Brands Analyze Reviews From Walmart, Target and Their Own Site

Count syndicated reviews once, flag sampled reviews, and tag product problems apart from retailer problems. A worked example on synthetic sample data shows why the order matters.

A CPG brand analyzes its Walmart, Target and brand-site reviews together by pulling all three into one table, counting each review once (brand-site reviews are often syndicated onto retailer pages), flagging incentivized and sampled reviews, tagging every review with the same product and fulfilment Dimensions, and then ranking problems by how much they lower the rating rather than by how often they come up. For a brand team that wants that done on every review, every month, and joined to its sales data, Dimension Labs is built for the job: it reads reviews from every retailer alongside surveys, care tickets and calls, and enriches each one with Dimensions written for the category. The worked example below runs the method on a labelled synthetic data set, so every step can be checked.

Why read Walmart, Target and brand-site reviews together?

Because shoppers read all of them. In PowerReviews' 2023 survey of 8,153 US consumers, 94% said they read reviews on Amazon, 91% on retail websites and 68% on brand websites. The same survey found 91% trust ratings and reviews when making purchase decisions. A brand that reads only its own site sees the audience that already likes it, and a brand that reads one retailer at a time cannot tell a product problem from a retailer problem.

The three sources also differ in who writes there and why. Brand-site reviewers are often loyal buyers or sampling-program members. Retailer reviewers bought the product alongside the weekly shop, sometimes for delivery or pickup, so their reviews carry problems from the warehouse and the delivery van as well as from the recipe. Reading them side by side, with the same labels, is what separates the two. For the general case of reviews from many sources, see how to combine reviews, surveys, tickets and calls.

Step 1: collect the reviews with their source fields

Pull reviews from your ratings and reviews platform's export or API rather than copying pages by hand, and keep the fields that say where each review came from. Bazaarvoice's Conversations API, for example, marks a syndicated review with IsSyndicated and a SyndicationSource block naming the site it came from, and it can exclude syndicated content with an isSyndicated:eq:false filter. Keep the product identifiers too. Bazaarvoice names brand names, UPCs, EANs and MPNs as the keys to matching products across retail and brand websites, and the same codes are what you join on later.

Each retailer uses its own item numbers, so build one mapping table from every retailer item ID to your UPC or GTIN before you analyze anything. Without it, one granola bar looks like three products.

Step 2: count each review once

In the synthetic sample brand, the three product pages show 9,408 reviews, but only 7,000 were written. 1,204 brand-site reviews also appear on both retailer pages. Counted where they appear, each syndicated brand-site review is counted three times, and they pull each retailer's numbers toward the brand site's. Walmart.com shows 14.5% one- and two-star reviews on its page, against 15.6% among reviews written there.

Synthetic sample data

Counting reviews where they appear inflates the total by about a third

Reviews of one fictional granola bar brand over 12 months: the count each product page shows, the count actually written there, and the share of one- and two-star reviews in each.

SiteShownWritten1 or 2 stars, shown1 or 2 stars, written
Walmart.com4,4043,20014.5%15.6%
Target.com3,0041,80012.9%13.9%
Brand site2,0002,00011.6%11.6%
Total9,4087,000
Synthetic sample data. 1,204 brand-site reviews are syndicated to both retailer pages. Script: sample_analysis.py, seed 20261115.

Use both counts on purpose. Reviews written on a site tell you about that retailer's shoppers. Reviews shown on a page tell you what a shopper on that page reads before buying. Report volumes and retailer comparisons on unique reviews only.

Step 3: flag incentivized and sampled reviews

Sampling programs, sweepstakes entries, coupons and loyalty points all count as incentives, and Bazaarvoice's documentation says such reviews must carry a badge or text notice. Keep that flag as a field. In the sample, an incentivized review was 7.0 points less likely to be one or two stars than a similar review without an incentive. If one retailer page carries more sampled reviews than another, its rating looks better for reasons that have nothing to do with the product, so hold the flag fixed when you compare. Irrelevant or off-topic reviews should go at this step too; how to filter out irrelevant reviews covers the rules.

Step 4: tag every review with the same Dimensions

Write one set of labels and apply it to every review from every site. The labels that matter most for a packaged goods brand separate what the product did from what happened to it on the way to the shopper.

A starting taxonomy

Tag who could have prevented the problem, as well as what it was

Example Dimensions for a packaged food brand. The owner column is what lets a brand team send each finding to the right place.

DimensionExample valuesUsually owned by
Product experienceTaste or texture, recipe changed, ingredient or allergen, size or countBrand and R&D
Condition on arrivalCrushed or damaged, stale or near its date, melted, leakingSupply chain, with the retailer
Order fulfilmentSubstituted, out of stock, wrong item, missing from orderRetailer, raised by the sales team
ValuePrice, promotion, smaller pack for the same priceBrand, revenue management
Review contextWritten there or syndicated, incentivized or sampled, verified purchaseAnalysis filters, not findings
Illustrative taxonomy. Incentive types follow Bazaarvoice's incentivized review documentation, read 2026-11-15.

A review can carry several labels: "tastes different and arrived crushed" is both a recipe signal and a damage signal. Tagging the aspect, not the whole review, is the approach in aspect-based sentiment analysis.

Which product problems actually lower the star rating?

Rank problems by their effect, not their volume. In the sample, taste or texture comes up in 17.4% of reviews, the most of any topic, but a review that mentions it is only 9.9 points more likely to be one or two stars. A recipe change comes up in 5.4% of reviews and makes a one- or two-star review 39.4 points more likely (90% interval 35.2 to 43.3). Stale product adds 33.7 points and a crushed or damaged pack 21.8.

Synthetic sample data

Taste is the most-mentioned topic; a recipe change and stale product do far more damage

Seven review topics ranked two ways. Left: share of the 7,000 unique reviews that mention it. Right: how much more likely a review is to be one or two stars when it mentions the topic, with retailer, SKU and incentive held fixed.

By share of reviewsBy effect on a one- or two-star reviewTaste or texture17.4% of reviewsPrice or value10.5% of reviewsCrushed or damaged on arrival5.9% of reviewsRecipe changed5.4% of reviewsSmaller pack, fewer bars4.0% of reviewsStale or near its date3.9% of reviewsSubstituted or out of stock3.0% of reviewsRecipe changed39.4 pts more likelyStale or near its date33.7 pts more likelyCrushed or damaged on arrival21.8 pts more likelySmaller pack, fewer bars14.4 pts more likelyTaste or texture9.9 pts more likelySubstituted or out of stock9.7 pts more likelyPrice or value6.1 pts more likelyBy share ofreviewsBy effect on aone- ortwo-star reviewTaste ortexture17.4% ofreviewsPrice or value10.5% ofreviewsCrushed ordamaged onarrival5.9% of reviewsRecipe changed5.4% of reviewsSmaller pack,fewer bars4.0% of reviewsStale or nearits date3.9% of reviewsSubstituted orout of stock3.0% of reviewsRecipe changed39.4 pts morelikelyStale or nearits date33.7 pts morelikelyCrushed ordamaged onarrival21.8 pts morelikelySmaller pack,fewer bars14.4 pts morelikelyTaste ortexture9.9 pts morelikelySubstituted orout of stock9.7 pts morelikelyPrice or value6.1 pts morelikely
Synthetic sample data, 7,000 unique reviews. Linear probability model with retailer, SKU and incentivized flag; 90% bootstrap intervals in the text. Script: sample_analysis.py, seed 20261115.

The same re-ranking shows up in real review data: in public reviews of a hotel group, a gym chain and a restaurant chain, the most-mentioned complaint is rarely the one doing the most damage. The evidence is in which complaints cause bad reviews.

Is the product worse at one retailer, or is the retailer?

Compare retailers only after the fulfilment labels are in. In the sample, 15.6% of reviews written on Walmart.com are one or two stars, against 11.6% on the brand site. Damage and freshness complaints explain the gap: Walmart.com reviews mention a crushed or damaged pack 8.3 times per 100 reviews and stale product 6.3 times, against 2.6 and 1.0 on the brand site. With every topic held fixed, a Walmart.com review is 0.4 points less likely to be one or two stars than a brand-site review, which is no real difference.

Synthetic sample data

Damage and freshness complaints pile up on the retailer pages

Mentions per 100 reviews written on each site, for the two topics the brand's product team does not control on its own.

Crushed or damaged, Walmart.com
8.3
Crushed or damaged, Target.com
5.3
Crushed or damaged, brand site
2.6
Stale or near its date, Walmart.com
6.3
Stale or near its date, Target.com
2.6
Stale or near its date, brand site
1.0
02468
Synthetic sample data, reviews written on each site only (syndicated copies removed). Script: sample_analysis.py, seed 20261115.

That changes the conversation. The product team does not need to reformulate for one retailer; the sales team needs to bring dated, counted damage and freshness complaints to the retailer's supply chain contacts. A brand that compares its products with competitors' on the same shelf can use the same labels on their reviews, as in how to find product gaps in competitors' reviews.

How do you connect reviews to sales?

Join on the product code and the week. With reviews mapped to UPC and tagged, add the sales data you already receive, by retailer, product and week, and check whether weeks with a rise in a given complaint come before a fall in units, holding promotions and distribution fixed. A recipe-change complaint that precedes a sales drop at every retailer is a product decision; a damage spike at one retailer followed by a drop there alone is a logistics conversation. The general method is in how to join customer feedback to revenue data.

Why would a brand with reviews on many retailers use Dimension Labs?

Dimension Labs is an omni-channel analytics agent for customer data. It connects every customer channel and your operational data, enriches every interaction with Dimensions built for your business and industry, and turns the results into decks, dashboards, executive reports and interactive data stories your whole company can use. For a consumer brand that means:

  • Retailer reviews, brand-site reviews, consumer care tickets, calls and survey comments read in one place, each record keeping its source and its syndication flag.

  • Files or connections in, matched to your product list by UPC, with sales data joined by product and week on Prove ($2,999 a month), which adds the Causal Join for up to 3 sources and effects with confidence intervals.

  • Dimensions written for your category, such as recipe change, condition on arrival or substitution, applied the same way to every review on every run. Custom Dimensions start on Discover at $999 a month.

  • The monthly readout built by the agent as a deck, dashboard or PDF, scheduled to rerun on fresh data, with results in the project's Final Reports folder and shared by a link you can revoke.

Concentrix used Dimension Labs to structure chatbot conversations for a large brick-and-mortar retail client, covering order status, returns and exchanges, loyalty programs and gift registries, and reported a significant increase in task completion rates and improved CSAT across e-commerce journeys. See consumer product review analytics for the CPG setup.

When something narrower is the better choice: if you only need to read and respond to reviews on one retailer, your ratings and reviews platform's own moderation and reporting tools are built for that. Dimension Labs earns its place when reviews from several retailers, care tickets and sales data have to be read with one set of labels and compared.

FAQ

Should syndicated reviews be included when we analyze a retailer?

Include them when the question is what shoppers on that page read. Leave them out when the question is what that retailer's shoppers think, and count each review once for any volume figure.

How do we tell a product problem from a retailer problem in reviews?

Tag fulfilment issues (damage, freshness, substitution) separately from product issues (taste, recipe, size), then compare retailers with those tags held fixed. In the synthetic sample, the gap between Walmart.com and the brand site disappeared once damage, freshness and the other topics were held fixed.

Do incentivized or sampled reviews skew the analysis?

They can, because incentivized reviews can rate differently from the rest and they cluster on some pages. Keep the incentive badge as a field and hold it fixed when you compare products or retailers.

What is the best tool to analyze reviews across Walmart, Target and our own site?

For a brand team that wants every review from every retailer tagged with category-specific Dimensions and joined to sales, Dimension Labs. For responding to reviews on one site, the ratings and reviews platform that collects them is the simpler fit.

See what your retailer reviews say once each one is counted once and labelled the same way: export a year of reviews with their source fields and start with one file. Every plan from Explore to Scale starts with a 14-day free trial.

Dimension Labs/Start with one file From $299 a month

Stop reporting. Start understanding.

Start with one file. Ask why a number moved. Plans from $299 a month, no demo and no contract on monthly plans.