THEPROOFFILE

A note before you start

Almost everything a DTC brand says about itself is something the customer can’t check before paying. Made in Vermont. Clinically tested. Handmade. Loved by 40,000 customers. Shoppers know this, so they discount the words and look for proof, and the proof they find first is usually your reviews.

Those two numbers are the guide in miniature. The first says review scores aren’t a clean measure of what customers think: one early vote, assigned by a coin flip, moved where ratings ended up. The second says the old shortcuts, buying a few reviews, holding back the bad ones, rounding a claim up, now carry a price per instance.

So this guide does two jobs. It shows, from the research, how reviews and claims shape belief. And it shows how to build the record that lets you say true things with confidence: a proof file with one row for every claim on your site and in your email.

A claim without a file behind it is a bet that nobody asks.

It builds on The Honest Test, on the product page, and The Whole Machine, on the email flows the review ask lives in.

How to read it

Start with The Proof Audit, or with the map just below, which shows every kind of claim and what proof it needs. Or follow a path:

• Founder: Every Claim on One Page, Claims Nobody Can Check, The Proof File, The Rules, Dated.
• Retention or email lead: The First Review, The Ask, Show the Bad Ones.
• Brand or copy lead: The Proof File, Origin, Handmade and Effort, Grading the Persuasion Canon.
• Analyst: part two, then For Your Analyst.

Three tools run in the page. Nothing you type leaves your browser.

What’s proven and what isn’t

• Filed: regulator orders, court documents and filings, with the date.
• Published: peer-reviewed research, books by the researchers themselves, and official law and regulator guidance, with the date it was true.
• Reported: what a named company, research center or outlet published. Review platforms publish much of the data in this field about the thing they sell, so the text says “vendor data” each time.
• Derived: my arithmetic, with the inputs shown.

Examples that open with Say or Picture use made-up round numbers. Every source is listed in Appendix C. Where the law comes up, it’s dated and it isn’t legal advice: have counsel review anything you’ll publish.

The further down a claim sits, the less the customer can check it, and the more your reviews, your file and regulators have to say about it.

PublishedDates are when each rule or guide took effect or was last revised; chapter 11 has the details and sources. Economists have sorted claims this way since Phillip Nelson (1970) and Michael Darby and Edi Karni (1973): search qualities you can check before buying, experience qualities you learn by using, and credence qualities you may never be able to verify.

Most DTC brands put their strongest words in the bottom four rows, because that’s where the differentiation lives. Those are also the rows where customers depend most on your word, and where a regulator is most likely to ask for your file.

Your best claims are the ones customers can’t check. That’s why they need the best proof.

Do this

• Pick your top product page and your last campaign email. Mark every claim in them with its row in this table. Count how many sit in the bottom four rows. That count is the size of your first proof file.

The audit isn’t about how many reviews you have or how high the stars are. It’s about whether the things customers read about you are true, whether you can show it, and whether your review system reports what customers think or what you’d like them to think.

Open your top product pages, last five campaign emails, post-purchase flow and review app settings. Score each check 0 to 2: 0 if it failed or nobody can answer it, 1 if partly true, 2 if clean.

If nobody can find the proof in a day, you don’t have proof. You have a memory.

The twelve checks

1. Every claim is listed · 5 minLook at: Your top three product pages, home page and last five campaign emails.
   Good: Every factual claim in them appears in one list, with where it runs.
   Cost if wrong: Claims multiply by copy-paste, and nobody knows which pages to fix when one turns out wrong.
   Read next: The Proof File
2. Each claim has dated evidence and an owner · 4 minLook at: Pick three claims at random and ask for the evidence.
   Good: A document for each, dated within the last year, with a named person who checks it.
   Cost if wrong: The FTC expects a reasonable basis before a claim runs, not after someone asks.
   Read next: The Proof File
3. The hardest claims have the right kind of proof · 4 minLook at: Your health, environmental and origin claims.
   Good: Health claims rest on studies, green claims on certifications or data, origin claims on a bill of materials.
   Cost if wrong: These are the claims customers can least check and regulators most often do.
   Read next: Claims Nobody Can Check
4. Reviews sit near the buy button, with count and spread · 3 minLook at: Your top product page on a phone.
   Good: Star average, review count and the distribution by star are visible without hunting, and shoppers can sort to the lowest.
   Cost if wrong: Reviews are the proof shoppers look for first. Hidden ones do no work.
   Read next: Reviews Move Sales
5. New products reach five reviews fast · 4 minLook at: Products launched in the last six months, and days from launch to the fifth review.
   Good: Under 30 days for your main products, with a plan for the rest.
   Cost if wrong: The first five reviews carry most of the conversion effect, and the first one steers the rest.
   Read next: The First Review
6. Small counts aren’t dressed up as big wins · 3 minLook at: Your “top rated” collection or sort order.
   Good: It ranks by a score that accounts for count, not the raw average.
   Cost if wrong: A 5.0 with three reviews outranks a 4.7 with 900, then falls apart in public.
   Read next: The First Review
7. Rating is tracked by month and by product age · 3 minLook at: Whether anyone reports average rating of new reviews by month.
   Good: A monthly line for each top product, read against its age and review count.
   Cost if wrong: You mistake the normal downward drift for a quality problem, or miss a real one.
   Read next: Why Ratings Drift Down
8. The ask is timed to use, not to shipping · 4 minLook at: The trigger and delay on your review request flow.
   Good: It fires from delivery plus the time it takes to use the product, and differs by category where use differs.
   Cost if wrong: Early asks produce reviews about the box and the courier instead of the product.
   Read next: The Ask
9. No gating, no sentiment-linked rewards · 4 minLook at: Your review app’s settings and any incentive copy.
   Good: Everyone gets the same ask. Any reward is the same whatever the rating, and disclosed on the review.
   Cost if wrong: Buying reviews conditioned on sentiment is banned outright under the FTC’s 2024 rule, and gating is flagged in its guidance.
   Read next: The Ask
10. Every genuine review is published and answered · 4 minLook at: Held, hidden or rejected reviews in the last 90 days, and replies to 1- and 2-star reviews.
    Good: Only reviews that break a published policy are held, each with a logged reason, and negatives get a reply within two business days.
    Cost if wrong: Holding back low ratings cost Fashion Nova $4.2 million.
    Read next: Show the Bad Ones
11. Origin, handmade and process claims are documented · 4 minLook at: Every “made in,” “handmade,” “small batch” and “family” claim.
    Good: Each maps to a bill of materials or a production record you’d show a stranger.
    Cost if wrong: Unqualified Made in USA labels are covered by an FTC rule with civil penalties.
    Read next: Origin, Handmade and Effort
12. Someone owns the rules · 3 minLook at: Who last checked your review and claim practices against current law.
    Good: A named person, a date this year, and a note on the FTC review rule, the EU green claims rules applying from September 27, 2026, and the UK rules if you sell there.
    Cost if wrong: The rules changed three times in two years, and ignorance isn’t a defense.
    Read next: The Rules, Dated

Score as you go; your band appears when all twelve are in.

Read your score

Checks 2, 3, 9, 10 and 11 carry legal exposure, not just lost sales. If any of those five scored 0, treat it as this week’s work whatever your band says.

A shopper can see a hoodie’s color. They can’t see whether it will pill after ten washes, and they’ll never see where the cotton was grown. Every claim falls into one of those three bins.

Three kinds of claim

Economists named the bins fifty years ago. Phillip Nelson (1970) split product qualities into search qualities, which you can check before buying, and experience qualities, which you only learn by using. Michael Darby and Edi Karni (1973) added credence qualities, which a buyer may never be able to verify even after use: whether a supplement contains what it says, whether a candle was poured by hand, whether a mailer is recyclable in your town Published.

The theory predicts that buyers trust search claims most, because lying about them is pointless. In 1990 Gary Ford, Darlene Smith and John Swasy tested that directly. Consumers were more skeptical of experience claims than search claims, and more skeptical of subjective claims (“luxurious”) than objective ones (“100% cotton”). The researchers didn’t find the extra step they expected for credence claims over experience claims Published. Read that as: once a shopper can’t check a claim before paying, the discount is already applied. It doesn’t matter much to them whether they’ll find out next week or never.

The discount starts the moment a claim can’t be checked on the page.

What shoppers do instead

They look for someone else who has checked. For most DTC products that means reviews: other buyers with no reason to flatter you. PowerReviews, which sells review software, reports that 95% of shoppers consult reviews before buying Reported (vendor data). The other checkers are certifiers, publications, and your own guarantee, which is a claim you pay for if it’s false. A generous return policy says “we’re confident” in a way copy can’t; the returns guide covers what it costs.

That’s why reviews and claims belong in one book. A claim your reviews contradict (“runs small,” says every third review of your “true to size” jeans) costs you twice: the sale, and belief in your next claim.

The regulator applies the same logic

The Federal Trade Commission’s 1984 policy statement on substantiation says advertisers need a reasonable basis for a claim before it runs. If an ad says or implies a level of support, like “tests show” or “doctors recommend,” the advertiser must have at least that level Published. For health claims, the FTC’s December 2022 guidance sets the bar at “competent and reliable scientific evidence” and says randomized controlled trials are generally what experts would require. Testimonials, however sincere, are not a substitute Published.

Notice the overlap. The claims customers discount most are the ones regulators scrutinize most. Both are reacting to the same fact: only you know whether it’s true.

Move claims up the table

You can’t make a credence claim checkable, but you can make it more specific, name who checked it, and show the document. Each step moves it closer to a search claim.

The figures are illustrations and the brackets are placeholders. Never fill them with a number, certifier or study you don’t have on file.

Do this

• Take the claims you marked on the map and rewrite the three vaguest ones as specific, checkable statements. If you can’t, because you don’t know the number or who checked it, that’s a claim to pull until you do.

Brands rarely lie on purpose. They drift. A founder says “all natural” in an interview, a copywriter repeats it, the formula changes, and three years later nobody knows where the claim came from. The proof file stops the drift.

The row

1. The claim, word for wordExactly as published. “Made in USA” and “Made in the USA with imported fabric” are different claims with different rules.
2. Where it runsEvery place: pages, footer, email and SMS, ads, packaging, briefs, macros.
3. KindIts row on the map: fact, performance, social proof, endorsement, origin, process, health, environmental.
4. EvidenceA link to the document itself: test report, certificate, bill of materials, study, review export.
5. Source and dateWho produced the evidence, when, and whether they’re independent of you.
6. What the evidence supportsThe widest wording the evidence allows. This is the column that catches drift.
7. Owner and review dateOne person, and when they’ll check it next. Health, origin and green claims every six months or when a supplier or formula changes; the rest yearly.

A spreadsheet is enough. Appendix B has the columns ready to paste.

The column that matters most is the one that says how far the evidence goes.

A worked example

Say a candle brand’s footer reads “Hand-poured in the USA,” it sends 12 campaigns a month to 150,000 subscribers, and its product pages get 200,000 views a month. That footer claim is delivered about 1.8 million times a month in email alone, before counting the site. Now say the wax blend is imported and the pouring moved to a contract manufacturer last spring, which uses a filling machine for the 16-ounce size. The claim was true when written. For at least one product it no longer is, in two ways: origin and process.

At the review date, the “what the evidence supports” column would read “poured in the USA from imported wax; 8-ounce size poured by hand.” Still a good claim, just not the one in the footer.

Where claims hide

• Subject lines, product titles and alt text, written fast and never reviewed.
• Review widgets: “verified buyer” and “4.8 average” are claims about how you checked and counted.
• Influencer briefs and support macros, where your claims go out in other people’s words.
• AI-drafted copy, which invents plausible specifics. No number it writes ships without a row.

Score a claim

The scorer weighs four things: how little the buyer can check, how thin your evidence is, how far the wording outruns it, and whether a specific rule covers the term. It’s my weighting, not a legal test; evidence counts most because you control it.

With the example’s numbers, the “Made in USA” footer scores 73, in the band to pull or qualify now, and the first fix is the evidence: a supplier’s email is not a bill of materials. At two million views a month it would be seen about 24 million times in a year. Score your own top ten claims and fix them in score order.

Do this

• Start the file this week with your top ten claims, one row each, and score them. Anything at 65 or above comes down or gets qualified before the next campaign. Then add every new claim before it ships: no row, no publish.

Products with more reviews sell more, but that could be sales causing reviews. Researchers have found three ways around the problem, and all three say reviews cause sales.

Same books, two stores

In 2006 Judith Chevalier and Dina Mayzlin compared the same books on Amazon and Barnes & Noble’s website. Anything about a book itself, its author, its marketing, its cover, is the same on both sites, so a change in reviews on one site but not the other isolates the effect of the reviews. An improvement in a book’s reviews raised its relative sales on that site. And the effect was lopsided: a one-star review hurt more than a five-star review helped Published.

Reviews were also overwhelmingly positive: 67% of reviews on Barnes & Noble were five stars, and 53% on Amazon Published. Hold that thought for chapter 5.

Different designs, same answer

• Restaurants. Michael Luca used the way Yelp rounds ratings to the nearest half star: restaurants either side of a rounding line are nearly identical but show different stars. A one-star increase led to 5% to 9% more revenue for independent restaurants, and no measurable change for chains Reported (Harvard Business School working paper).
• Amazon products. Sherry He, Brett Hollenbeck and Davide Proserpio tracked products that bought fake reviews in Facebook groups and used a temporary change in Amazon’s policy to show that rating manipulation had a large causal effect on sales Published. That’s why the market for fake reviews exists. The same paper found what happens next, in chapter 6.

A one-star review costs more than a five-star review earns. Plan your review system around the bad ones.

The first five

Northwestern’s Spiegel Research Center analyzed data from PowerReviews, a review platform, across many retailers. A product with five reviews had a purchase likelihood 270% greater than the same kind of product with none. The benefit of each extra review fell off quickly after the first five, and showing reviews lifted conversion more for higher-priced products Reported (vendor data, analyzed by a university research center).

Treat the exact numbers with care. This is observational data, and products that collect reviews may differ from products that don’t. But it fits the peer-reviewed work: Ewa Maslowska, Edward Malthouse and Vijay Viswanathan found, using data from two retailers, that the rating’s effect on purchase was strongest when a product had many reviews, when the shopper read them, and when the product was expensive Published. For a DTC brand that means the review gap that costs most is the one on new and expensive products.

So coverage beats average: the reviews worth most are the first five on products that have none. Show stars, count and a link near the price and the button; The Honest Test covers the product page.

Do this

• Pull revenue by product with each product’s review count (Appendix A). Mark every product under five reviews. The revenue on that list is the size of your review gap.

Your average rating looks like a survey result. It isn’t. A survey asks everyone; reviews come from whoever chose to write.

A few people write most of it

Jakob Nielsen called it participation inequality in 2006: in most online communities, about 90% of users only read, 9% contribute occasionally, and 1% contribute most of the content. His example from Amazon: one reviewer alone had written 12,423 book reviews Reported. Clay Shirky’s Here Comes Everybody (2008) makes the same point at book length: contributions follow a steep power law, and the average contributor doesn’t exist.

The J

Plot most products’ reviews by star and you get a J: a tall bar at five stars, a smaller bump at one, and little in between. Nan Hu, Paul Pavlou and Jennifer Zhang found that 78% of book ratings, 73% of DVD ratings and 72% of video ratings on Amazon were four stars or higher Published.

Then they ran the test that settles it. In a controlled experiment, everyone rated the same music CD. Their ratings formed an ordinary hump in the middle. The same CD’s ratings on Amazon formed a J Published. Same product, different shape, because on Amazon nobody was required to rate it.

Two filters explain it. Purchasing bias: people who expect to dislike a product don’t buy it, so never review it. Under-reporting bias: the delighted and the furious “brag or moan”; the mildly satisfied stay quiet. So the average is a poor proxy for quality; look at the spread and the two peaks too Published.

The middle of your customer base is the part that doesn’t write.

Why the middle goes quiet

Leif Brandes, David Godes and Dina Mayzlin (2022) added a third filter: attrition. Reviewers with moderate experiences are more likely to stop reviewing over time than those with extreme ones. In a large field experiment with an online travel platform, plain review-request emails, with no incentive, made the reviews that came in less extreme Published. Asking everyone brings back the middle. That’s the strongest research case for a systematic ask, and chapter 8 is about how.

Some reviewers never bought

Eric Anderson and Duncan Simester found that about 5% of reviews at a large private-label retailer came from customers with no record of buying the product, and those reviews were significantly more negative. Many came from the firm’s most loyal customers Published. Some are gift recipients, not fakes. But it’s a reason to label verified buyers. Spiegel’s analysis of PowerReviews data found a verified-buyer badge raised the odds of purchase by 15% Reported (vendor data).

Do this

• Export your reviews and chart the distribution by star for your top five products. Note the share at three stars. If it’s under 10%, your middle isn’t writing, and your ask isn’t reaching everyone. Show the distribution on the page, and label verified buyers.

If people rated only what they saw, the order of reviews wouldn’t matter. It does. People read what came before and adjust, and the first rating sets the direction.

The experiment

Lev Muchnik, Sinan Aral and Sean Taylor ran a randomized experiment on a social news site like Reddit, published in Science in 2013. Over five months, 101,281 comments were randomly assigned at birth to get one up-vote, one down-vote, or nothing. The votes had nothing to do with the content. The researchers then watched 308,515 ratings from real users Published.

• One fake up-vote made the next viewer 32% more likely to up-vote too, and the comments ended with ratings 25% higher than the untouched ones.
• One fake down-vote was corrected. Users pushed back, and those comments ended no different from the control group.

Positive herding compounds; negative herding gets corrected, at least on that site. In Matthew Salganik, Peter Dodds and Duncan Watts’s MusicLab experiment with 14,341 participants, showing what others downloaded made hits bigger and success less predictable. The best songs rarely did badly and the worst rarely did well, “but any other result was possible” Published.

Your first reviews don’t just describe the product. They steer every review that follows.

Why early buyers skew it

The first reviews come from whoever buys first, and they aren’t typical. Xinxin Li and Lorin Hitt showed that early buyers’ particular tastes shape later buyers’ decisions and can mislead them Published. Your loyalists produce fans; cold paid traffic produces mismatches. Neither is your month-six customer.

Don’t buy the start

Fake early reviews work briefly, which is why people buy them. He, Hollenbeck and Proserpio found the effect was short-lived: after products stopped buying fake reviews, their ratings fell and their share of one-star reviews rose significantly, especially for young products Published. In 2019 the FTC brought its first case challenging fake paid reviews on an independent retail website: Cure Encapsulations had paid a website to post Amazon reviews of a weight-loss supplement and asked it to keep the product at five stars. The judgment was $12.8 million, suspended on payment of $50,000 Filed. Since October 2024, fake reviews carry civil penalties per violation (chapter 11).

Launch for the first review

1. Seed with the right customersSend launch stock to customers who match the product’s core buyer. A free product for an honest review is allowed in the US if the reward doesn’t depend on what they say and the review discloses it Published. Google bans incentives for any review on Google Maps Reported.
2. Ask every launch buyer, on timeThe first 100 orders get the ask at the day chapter 8 finds, with one reminder.
3. Answer the first negative within a dayIn public and specific, with what you’re changing. Readers correct unfair negatives when the facts are visible.
4. Hold “top rated” labelsUntil the count can carry them, which is what the tool below checks.

How much a small count can carry

A 4.9 average from 12 reviews looks better than a 4.6 from 840. It isn’t, yet. One more one-star review would pull the first to 4.6, while the second would barely move. The fix is to rank by a score that shrinks small counts toward your store’s typical rating and then asks how low the true average could plausibly be. It’s a simplified version of the Bayesian approach the statistician Evan Miller describes for star ratings Reported.

With the defaults, product A’s 4.9 adjusts to 4.67 with a lowest plausible score of 4.29, while B’s 4.6 adjusts to 4.60 with a floor of 4.54. B is the one to feature. A needs about 27 more reviews at 4.9 to pass it, and one more one-star review would show it at 4.6 Derived. Use the same score to sort your “top rated” collection.

Do this

• Write a launch plan for your next product that ends at five reviews, not at launch day: who gets seed stock, when the ask goes, who answers the first negative. And re-sort your “top rated” collection by the lowest plausible score.

A product launches at 4.8 and a year later sits at 4.5. Often nothing went wrong with the product. The later buyers are different people, reading a longer and messier set of reviews, and they rate differently.

The pattern

David Godes and José Silva (2012) separated two clocks: how long a product has been on sale, and how many reviews came before a given one. Once they controlled for calendar time, the time pattern actually rose. But the sequence pattern fell: each successive rating tended to be lower than the one before, controlling for the product and the reviewer Published.

Their explanation: as reviews pile up, it gets harder to tell which apply to you, so more people buy something that doesn’t suit them, and rate it lower Published. Two other findings point the same way. Early buyers have unusual tastes, so the first reviews can mislead later buyers (Li and Hitt). And frequent reviewers tend to differentiate themselves from the crowd, pulling ratings down over time, while occasional reviewers follow it (Wendy Moe and David Schweidel) Published.

Drift is mostly mismatch. The fix is helping the wrong buyer notice before they buy.

Compare like with like

• Compare at similar counts. A 4.8 at 30 reviews and a 4.5 at 3,000 aren’t on the same scale; use the adjusted scores from chapter 6.
• Track new reviews by month, not the running average, which hides changes for months.
• Look for breaks, not slopes. A slow slide is drift. A step down after a supplier, batch or sizing change is a product problem.

Slow the slide

• Say who it isn’t for. “Runs small.” “Not for oily skin.” Every mismatch you prevent is a low review and often a return you don’t get.
• Let shoppers filter reviews by people like them: size, skin type, breed. Godes and Silva’s mechanism is readers failing to find the relevant reviews.
• Keep asking everyone. Spiegel’s analysis of PowerReviews data found ratings from verified buyers who were emailed after purchase stayed more consistent over time, while ratings from anonymous, self-motivated reviewers declined Reported (vendor data). The systematic ask is a brake on drift.

Do this

• Chart the monthly average of new reviews for your top five products for the last two years (Appendix A), with supplier, formula and packaging changes marked. Treat any step down of 0.3 stars or more that lines up with a change as a product issue, and everything else as drift.

PowerReviews, a review platform, reports that 70% of its clients’ reviews come from emails sent after purchase Reported (vendor data). So the review request is not a small flow. It’s the machine that writes most of your social proof, and its settings shape the result.

Asking changes who answers

• It brings back the middle. In Brandes, Godes and Mayzlin’s field experiment, plain request emails with no incentive made incoming reviews less extreme Published.
• It raises volume and can crowd out the rest. When hotels joined TripAdvisor’s review solicitation program, review volume rose 34.3% and ratings rose 0.151 stars, but reviews got 16.9% shorter and unprompted reviews fell 15.5% (Baojun Gao and colleagues, 2025) Published. Solicited reviews partly replace organic ones rather than adding to them.
• Money buys count; norms buy length. In a field experiment with a large online clothing retailer in China, Gordon Burtch and colleagues found a financial incentive produced more reviews but not longer ones. Telling customers how many peers had written reviews produced longer ones. The two together did best Published.

So: ask everyone, once, with one reminder; use a true social-norm line (“2,314 customers have reviewed this”); and expect a reward to buy more short reviews, not better ones.

Time it to use

A review is only useful if the writer has used the product. Asks timed to fulfillment arrive before the product and produce reviews about the courier, or nothing. The right trigger is delivery plus the time to form a view: days for a T-shirt, weeks for skincare or supplements.

Say a brand ships 3,000 orders a month, delivery takes four days, and most customers have used the product enough to judge it about two weeks after it arrives. Say 8% of customers who have used it answer an ask, 3% of those who haven’t still reply with something about delivery, and interest halves every 30 days after the right moment. The tool shows what the send day does.

With those defaults, an ask on day 7 produces about 51 product reviews a month plus 71 about delivery, so 58% of what comes in says nothing about the product. Moving it to day 18 produces about 240 product reviews a month, and a new product selling 150 a month reaches five reviews in about two weeks instead of two months Derived. Your own rates will differ; the shape usually won’t.

Trigger the ask from delivery plus use, never from the shipping label.

If results take a month to show, as with skincare, send a first-impressions ask at two weeks and a results ask at five, labeled on the review. Text only customers who agreed to texts. Appendix B has both versions.

What you may not do

• No gating. Don’t ask only the customers you expect to be happy, or route unhappy ones to support instead of the review form. The FTC’s 2022 guidance: “Don’t ask for reviews only from people you think will leave positive ones” Published.
• No rewards tied to sentiment. Since October 21, 2024, the FTC’s rule bans giving anything of value conditioned, expressly or by implication, on a review expressing a particular sentiment Published. “Leave a five-star review for 15% off” is exactly that.
• Disclose any reward on the review itself. A free product or a discount for any honest review is a material connection under the FTC’s Endorsement Guides, and the UK requires incentivized reviews to say so Published.
• Check each platform’s own rules. Google Maps bans incentives for any review and bans selectively asking happy customers Reported. Marketplaces set their own rules too.

Do this

• Change your review flow’s trigger from fulfilled to delivered, set the delay by category with the tool, and remove any star-rating filter or sentiment-linked reward from the flow this week.

Shoppers go looking for the bad reviews. PowerReviews data cited by the Spiegel Research Center says 82% of shoppers specifically seek out negative reviews Reported (vendor data). If they can’t find any, they don’t conclude the product is perfect. They conclude something is being hidden.

Perfect looks fake

In the Spiegel Research Center’s analysis of PowerReviews data, purchase likelihood peaked for products rated between 4.0 and 4.7 stars, then fell as ratings approached 5.0 Reported (vendor data, analyzed by a university research center). The researchers’ reading: negative reviews “establish credibility and authenticity.” It’s observational, so treat the range as a guide.

There’s experimental support. Danit Ein-Gar, Baba Shiv and Zakary Tormala (2012) found in four studies that adding a small dose of negative information to an otherwise positive description made people like a product more, which they called the blemishing effect. The conditions matter: it worked when the negative came after the positive and when people weren’t reading closely Published. A product page fits both. Lead with the summary and the praise; put a real critical review one tap away.

A 4.6 with visible one-stars is more believable than a 5.0 with none.

The case: Fashion Nova

In January 2022, the FTC announced that the fast-fashion retailer Fashion Nova would pay $4.2 million to settle charges that it blocked negative reviews. From late 2015 through November 2019, according to the FTC, the company used a third-party review interface that automatically posted four- and five-star reviews and held lower-rated ones for approval that never came. Hundreds of thousands of lower-rated reviews were never published. It was the FTC’s first case about a company concealing negative reviews. The same day, the FTC sent letters to ten companies that provide review-management services and published guidance on collecting and showing reviews Filed.

The lesson is in the mechanism. Nobody had to delete a review; a moderation setting did the work. Most review apps have one. Find yours.

What you may remove, and how

You can hold back some reviews, but the rules are narrow and must apply to positive and negative reviews alike:

• US. The FTC’s 2024 rule bans implying the reviews you show are most or all of those submitted when you’ve suppressed negative ones, with exceptions for content such as defamation, trade secrets and clearly false reviews Published. Its 2022 guidance adds: don’t subject negative reviews to more scrutiny Published.
• UK. The Competition and Markets Authority’s April 2025 guidance says publishers must not suppress genuine negative or positive reviews or selectively promote either Published.
• EU. Since May 28, 2022, traders showing reviews must tell consumers whether and how they check that reviews come from real customers. In a January 2022 sweep, the European Commission and national authorities found 104 of 223 major sites didn’t explain how reviews were collected and processed Published.

Write a moderation policy, publish it next to the reviews, apply it to every star level, and log every removal with the reason. Appendix B has a starting draft.

Answer them

A reply to a one-star review is written for the next hundred shoppers. Thank them, state the fact without arguing, say what you changed, and offer a person to talk to, within two business days. Never offer anything for changing or removing the review.

Do this

• Open your review app’s moderation settings today. Turn off any auto-hold by star rating, publish every genuine review in the queue, and add a “most helpful critical review” next to the most helpful positive one on your top product pages.

Customers can’t check where a product was made or whose hands made it, yet they pay for those facts. Three lines of research explain why; one FTC rule sets the cost of getting them wrong.

Authenticity is contagious

George Newman and Ravi Dhar (2014) found that people see products made in a brand’s original factory as carrying the “essence” of the brand, a belief in contagion, and so as more authentic and more valuable than identical products made elsewhere. People more sensitive to contagion showed the effect more strongly, and priming the idea of contagion strengthened it Published. Their studies used everyday branded goods, including Levi’s jeans Reported (Yale School of Management).

The founding workshop and the town on the label carry that essence. That’s real value, and it’s why moving production without changing the story is a problem.

Handmade contains love

Christoph Fuchs, Martin Schreier and Stijn van Osselaer (2015) found across four studies that “handmade” makes products more attractive, largely because people feel handmade products symbolically contain the maker’s love. In one study, people asked what they would pay for a bar of soap for their mother offered $6.56 when it was described as handmade and $5.63 when machine-made Published, about 17% more Derived. The effect held for gifts to loved ones but not to distant recipients, and was stronger when the goal was to show love rather than to get the best-performing product Published. So “handmade” earns most on gifts, and on the occasions people buy them.

People value visible effort

Ryan Buell and Michael Norton (2011) called it the labor illusion. In five experiments with travel and dating websites, people who were shown the work being done on their behalf valued the service more, and could even prefer a longer wait to an instant result, with identical results Published. In later field experiments in food service, Buell, Tami Kim and Chia-Jung Tsay found that when cooks and customers could see each other, customer-reported quality rose 22.2% and service got 19.2% faster Published.

Show the work you really do: sourcing, batch testing, hand-finishing, fit sessions. Shown effort is valued; it just has to be real.

The claims customers value most are about things they can’t see. Keep the records they’d want to see.

The rule on Made in USA

The FTC’s Made in USA Labeling Rule took effect on August 13, 2021. It covers labels and mail-order promotional material, including material sent electronically. An unqualified “Made in USA” claim requires that final assembly or processing happen in the US, that all significant processing happen in the US, and that all or virtually all ingredients or components be made and sourced in the US Published. Violations carry civil penalties. A qualified claim, like “Made in USA from imported leather,” is the honest alternative when the inputs come from abroad.

Enforcement is active. From 2021 to 2024 the FTC brought 11 Made in USA actions with $15.75 million in judgments Filed. In July 2025 it sent warning letters to four companies, and to Amazon and Walmart about third-party sellers’ US-origin claims Filed.

The case: Williams-Sonoma

In 2020, Williams-Sonoma settled FTC charges that it had falsely advertised several product lines as all or virtually all made in the USA, and agreed to an order requiring truthful origin claims. In April 2024 it agreed to pay a $3.175 million civil penalty for violating that order, which the FTC called the largest ever in a Made in USA case. According to the FTC, mattress pads sold under its Pottery Barn Teen brand were marketed as “Crafted in America from domestic and imported materials” but were made in China, and six other products advertised as made in the USA were also deceptive Filed.

A company with a legal department, under an order naming the exact issue, still let the claim run on products it didn’t fit. Copy outlives the facts. That’s what the proof file is for.

Do this

• For every origin, handmade and process claim, build the bill of materials: each component, its country of origin, where each step happens and whether it’s done by hand. Qualify or remove any claim the bill doesn’t support, starting with product titles and feeds, where they spread furthest.

Review and claim rules used to say only: don’t deceive. Now they name specific practices and set penalties per violation. This is a dated map as of September 2026, not legal advice; have counsel review your practices against it.

United States

PublishedFTC, Federal Register and eCFR texts listed in Appendix C. Green Guides status: no revision in the Federal Register, and none listed by the BWD Strategic tracker, checked September 2026.

The review rule has started to bite. On December 22, 2025, the FTC sent warning letters to ten companies about possible violations, noting penalties of up to $53,088 per violation Filed. That is the 2025 inflation-adjusted cap; the Federal Register showed no 2026 adjustment as of September 2026 Published. Before the rule, in October 2021, it had already sent a Notice of Penalty Offenses on endorsements to more than 700 companies, which makes it easier to seek civil penalties from recipients who later use those practices Filed.

European Union

• Reviews, since May 28, 2022. Under the changes made by Directive (EU) 2019/2161, a trader that shows consumer reviews must say whether and how it ensures they come from customers who used or bought the product Published.
• Green claims, from September 27, 2026. Directive (EU) 2024/825, on empowering consumers for the green transition, entered into force on March 26, 2024. Member states had until March 27, 2026 to write it into national law, and it applies from September 27, 2026. It adds to the EU’s list of practices banned in all circumstances: generic environmental claims like “eco-friendly” or “green” without recognized excellent environmental performance; sustainability labels not based on a certification scheme or set by public authorities; claims that a product has a neutral, reduced or positive climate impact because of offsetting; and environmental claims about a whole product when they apply to only part of it Published.

If you sell into the EU, act on the green claims rules now: “carbon neutral shipping” backed by offsets is exactly what the new list bans. A separate EU proposal on substantiating green claims is a different law; ask counsel where it stands.

United Kingdom

The Digital Markets, Competition and Consumers Act 2024 lists practices that are unfair in all circumstances, including submitting or commissioning fake reviews, concealing that a review was incentivized, and publishing reviews or review information in a misleading way. Those provisions came into force on April 6, 2025 Published. The Competition and Markets Authority’s guidance, published two days earlier, says businesses that publish reviews must take “reasonable and proportionate steps” to prevent and remove banned content, may use incentives only if the review says so and still reflects a real experience, and must not suppress genuine negative or positive reviews Published. Penalties under the Act can reach £300,000 or 10% of turnover, whichever is higher, and the CMA can now impose fines itself instead of going to court Published.

The question used to be whether a practice was deceptive. Now it’s often whether it’s on a list.

The case: Keurig Canada

Green claims get enforced too. In January 2022, Keurig Canada agreed to pay a C$3 million penalty to settle the Canadian Competition Bureau’s concerns about its K-Cup pods. The Bureau found that recyclability claims were false or misleading, because municipal recycling programs outside British Columbia and Quebec didn’t accept the pods, and that the instructions for preparing them for recycling didn’t match what many municipalities required. Keurig also donated C$800,000 to an environmental charity, paid C$85,000 in costs, changed its packaging claims and published corrective notices, including to its subscribers by email Filed.

The claim was one word on the box. The evidence it needed was a map of where the pod could be recycled. That’s the “what the evidence supports” column, filled in by a regulator.

Do this

• Name one owner for these rules and put a quarterly check on the calendar. First job: list every environmental claim you make to EU customers and compare it with the four banned practices above, which apply from September 27, 2026.

Marketing runs on famous findings, and some didn’t survive a second look. Before one goes into your playbook, ask: was it tested in the field, and has anyone besides the original team found it again?

PublishedAll references in Appendix C. The verdicts are my reading of the evidence I could verify for this guide, in September 2026.

Trust findings that were randomized, run in the field, and found twice.

Build on what held up: coverage, the first five reviews, the systematic ask. Don’t cite mixed findings to justify a decision; test the tactic if your traffic allows (The Honest Test). Use one-paper findings only where they’re cheap and true anyway, like showing a critical review or the real work behind the product, and don’t promise a lift.

Do this

• List the persuasion tactics in your current site and email playbook and write the finding each one relies on. Anything resting on a “mixed” or “didn’t hold” finding goes on the test list or comes out.

Review programs get checked when something goes wrong. The scorecard makes the check routine, so problems show up as a number turning the wrong way instead of a crisis.

The first two and the last three are compliance numbers: a slip there goes straight to the owner named in chapter 11. Read the middle seven together: if reviews per 100 orders jump while the three-star share falls, check that nobody switched on a filter or a reward.

A review program you only check in a crisis is a crisis you scheduled.

Appendix A has the queries. The one field you’ll usually add is a flag for reviews about delivery.

Do this

• Build the scorecard once by hand for last month, even with gaps. Put it on the monthly business review agenda, and give each number an owner before next month’s version.

The order matters. The claims carry the legal exposure, so they come first. The review settings come next, because a filter running today is a problem today. Then the ask, which takes weeks to show results. Then the routine that keeps it all true.

1. Week one: the claimsMark every claim on your top pages and last five emails against the map (front). Start the proof file with your top ten claims and score them (chapter 3); pull or qualify anything at 65 or above. Build the bill of materials for origin and handmade claims (chapter 10).
2. Week two: the review settingsTurn off any auto-hold by star rating and publish the queue (chapter 9). Remove gating and sentiment-linked rewards (chapter 8). Publish the moderation policy, the star distribution and the verified badge. Reply to unanswered one- and two-star reviews from the last 90 days.
3. Week three: the askMove the trigger to delivery plus time to judge, by category, with one reminder. Write the next launch plan to end at five reviews (chapter 6). Re-sort “top rated” by the lowest plausible score.
4. Week four: the recordBuild the first scorecard (chapter 13) and the monthly new-review chart for your top products (chapter 7). Name the owner for the rules and book the quarterly check (chapter 11). Re-score the audit.

At day thirty you’ll have fewer claims, each backed by a document someone can find, and reviews that are slightly less perfect and more believable.

Prove what you say, publish what they say, and ask when they know.

Do this

• Put the four weeks on the calendar today, with an owner for each line, and open your review app’s moderation settings before you close this page.

Whoever owns proof, a new hire, an agency or you, needs six things on day one. Without them, the first month goes on hunting for documents.

1. Admin access to the review appWith the moderation settings, the request flow’s trigger and delay, and a full export of every review, including held and rejected ones.
2. Every place claims liveSite, email templates and flows, SMS, ads, product feeds, packaging files, influencer briefs and support macros.
3. The supplier documentsCertificates, test reports, country-of-origin statements and bills of materials, with dates.
4. Orders with delivery datesSo the ask can be timed from delivery and review rates can be counted per delivered order.
5. A list of changesEvery supplier, formula, packaging and production change of the last two years, with dates. It explains most steps in the rating chart.
6. A named legal contactSomeone who can answer “can we say this?” within a week, and who knows which countries you sell into.

Do this

• Ask for all six in one message today. Whatever comes back slowest is where the brand knows least about what it’s saying.

• Authenticity: What Consumers Really WantJames H. Gilmore and B. Joseph Pine II, 2007Why people pay for the real thing, and why a brand has to be what it says it is before it can say it. Read it next to Newman and Dhar’s research on contagion.
• Yes! 50 Scientifically Proven Ways to Be PersuasiveNoah J. Goldstein, Steve J. Martin and Robert B. Cialdini, 2008A tour of social-proof research by people who ran it. Enjoy it, then check each finding against the grades in chapter 12 before you build on it.
• The Wisdom of CrowdsJames Surowiecki, 2004Crowds judge well when their members judge independently. Reviews that follow the first review aren’t independent, which is chapter 6 in one sentence.
• Here Comes EverybodyClay Shirky, 2008How a few people make most of the contributions online. The reason your average rating isn’t your customers’ average.
• BlockbustersAnita Elberse, 2013Why attention concentrates on a few hits. Early social proof is part of how a product becomes one.
• “The Effect of Word of Mouth on Sales: Online Book Reviews”Judith Chevalier and Dina Mayzlin, Journal of Marketing Research, 2006The paper that showed reviews move sales, and that one-star reviews move them more than five-star ones.
• “Social Influence Bias: A Randomized Experiment”Lev Muchnik, Sinan Aral and Sean J. Taylor, Science, 2013The cleanest evidence that early ratings steer later ones. A short paper that should change how you launch.

Andrew Lauchner runs Growth Legend, embedding inside consumer brands to own lifecycle, email and SMS, and revenue operations. He is the author of The Second Order, on turning first-time buyers into second-time buyers, and The Whole Machine, on the fundamentals of DTC growth, along with a series of field guides for DTC operators at andrewlauchner.com.

As Senior Director of Growth and Retention Marketing at Gallery Furniture, he rebuilt the customer journey and the sales playbooks together. He has worked on growth and retention at Binance and 3Commas, and has been Head of Growth and Retention at Greatness Wins and at Nexus Agriscience.

Andrew answers every note from operators working on this, including those looking for someone to own it. Write to andrew@growthlegend.com or message him on LinkedIn.

The formulas

Review coverage by revenue

-- revenue in the last 90 days by product, with published review count
-- reviews: review_id, product_id, order_id, rating, created_at, status
SELECT p.product_id, p.title,
       SUM(ol.price * ol.quantity)                     AS revenue_90d,
       COALESCE(r.reviews, 0)                          AS reviews,
       CASE WHEN COALESCE(r.reviews, 0) < 5 THEN 1 ELSE 0 END AS under_five
FROM order_lines ol
JOIN orders o   ON o.order_id = ol.order_id
JOIN products p ON p.product_id = ol.product_id
LEFT JOIN (SELECT product_id, COUNT(*) AS reviews
           FROM reviews WHERE status = 'published'
           GROUP BY product_id) r ON r.product_id = p.product_id
WHERE o.created_at >= CURRENT_DATE - INTERVAL '90 days'
GROUP BY p.product_id, p.title, r.reviews
ORDER BY revenue_90d DESC;

Sum revenue where under_five = 1 and divide by the total for the scorecard’s “revenue under five reviews.”

Review rate per delivered order

-- reviews per 100 delivered orders, by delivery month, split by topic
-- reviews.about_delivery: true when the review only discusses shipping or packaging
SELECT DATE_TRUNC('month', f.delivered_at)                          AS month,
       COUNT(DISTINCT f.order_id)                                   AS delivered,
       100.0 * COUNT(r.review_id) FILTER (WHERE NOT r.about_delivery)
             / COUNT(DISTINCT f.order_id)                           AS product_reviews_per_100,
       100.0 * COUNT(r.review_id) FILTER (WHERE r.about_delivery)
             / COUNT(DISTINCT f.order_id)                           AS delivery_reviews_per_100
FROM fulfillments f
LEFT JOIN reviews r ON r.order_id = f.order_id
WHERE f.delivered_at >= CURRENT_DATE - INTERVAL '12 months'
GROUP BY 1 ORDER BY 1;

Count reviews against the month the order was delivered, not the month the review arrived, or a change in ask timing will look like a change in response. Leave the latest two months out of any comparison until their reviews have had time to come in.

Drift: new-review average by month and by sequence

-- monthly average of new reviews, and average by position in the sequence
WITH seq AS (
  SELECT r.*, ROW_NUMBER() OVER (PARTITION BY product_id ORDER BY created_at) AS n
  FROM reviews r WHERE status = 'published'
)
SELECT product_id,
       DATE_TRUNC('month', created_at) AS month,
       COUNT(*)                        AS new_reviews,
       ROUND(AVG(rating), 2)           AS avg_new,
       ROUND(AVG(rating) FILTER (WHERE n <= 10), 2)          AS avg_first_10,
       ROUND(AVG(rating) FILTER (WHERE n BETWEEN 11 AND 100), 2) AS avg_11_to_100
FROM seq
GROUP BY 1, 2 ORDER BY 1, 2;

Chart avg_new by month with supplier, formula and packaging changes marked. A step of 0.3 stars or more at a change date is a product question; a slow slide is drift.

Days to the fifth review

-- for products published in the last 6 months
WITH ranked AS (
  SELECT product_id, created_at,
         ROW_NUMBER() OVER (PARTITION BY product_id ORDER BY created_at) AS n
  FROM reviews WHERE status = 'published'
)
SELECT p.product_id, p.title, p.published_at::date AS launched,
       (r.created_at::date - p.published_at::date)  AS days_to_fifth
FROM products p
LEFT JOIN ranked r ON r.product_id = p.product_id AND r.n = 5
WHERE p.published_at >= CURRENT_DATE - INTERVAL '6 months'
ORDER BY days_to_fifth NULLS FIRST;

The syntax is Postgres. In BigQuery, replace FILTER (WHERE ...) with COUNTIF or AVG(IF(..., rating, NULL)), and date subtraction with DATE_DIFF. Column names follow a generic Shopify-style export; rename to match yours.

The proof file row

CLAIM (EXACT WORDS)   | WHERE IT RUNS (URLs, templates, feeds, packaging)
KIND (map row)        | EVIDENCE (link to the document itself)
SOURCE AND DATE       | INDEPENDENT OF US? (yes / no)
WHAT IT SUPPORTS      | the widest wording the evidence allows
OWNER                 | NEXT REVIEW DATE | RISK SCORE (chapter 3)
STATUS                | live / qualified / pulled, with date

The review request email

SUBJECT   How Is Your [Product] Working So Far?
PREVIEW   two minutes, and every review gets published, good or bad

Hi [First name],

You've had your [product] for about [two weeks] now. Other customers
decide whether to buy it by reading what people like you say, so we'd
be grateful for a few honest lines. What worked, what didn't, and who
you'd recommend it to.

[ Write a review ]

We publish every review that follows our review policy [link], whatever
the rating. If something isn't right, reply to this email and a person
will help, and you're still welcome to review it.

[Brand] · [Address]
You're receiving this because you ordered from us. [Unsubscribe]

No reward, no suggested rating, same email to everyone. For the reminder a week later, add the true review count as a social-norm line.

The review request SMS

[Brand]: Hi [First name], it's been two weeks with your [product].
How's it going? Tell other customers here: [short link]
Reply STOP to opt out

Send only to customers who have agreed to marketing texts, within your sending hours.

The incentive disclosure

ON THE REVIEW   "This reviewer received [a free product / a 10% code]
                 for writing a review. The reward did not depend on
                 the rating or what they wrote."

The reply to a negative review

Thank you, [name], for telling us and other customers. [Fact, without
arguing: e.g. "The medium does run small in this style."] We've [what
you did or changed: e.g. "added a sizing note to the page"]. If you'd
like a replacement or a refund, write to [email] and [first name] on
our team will sort it out.
[First name], [Brand]

Never offer anything in exchange for changing or removing the review.

The review policy

HOW WE COLLECT REVIEWS
We email every customer [N] days after delivery to ask for a review.
We don't choose who is asked based on how we expect them to rate us.
[If applicable: Reviewers who received a free product or reward are
labeled, and the reward never depended on their rating.]

HOW WE CHECK THEM
Reviews marked "Verified buyer" match an order. [Say what you do with
reviews that don't match an order.]

WHAT WE PUBLISH
Every review, positive or negative, except those that contain: personal
information, abuse or harassment, content unrelated to the product,
or content we can show is false. The same rules apply at every rating.
We log every review we don't publish and why.

HOW WE CALCULATE THE RATING
The average of all published reviews for this product[, including
those imported from (source)].

The supplier evidence request

For [product / component], please send by [date]:
1. Where it is made, and where each significant step happens.
2. Country of origin of each input over [5]% of cost.
3. Certificates and test reports for [claim], with dates.
4. Notice of any change to the above, before it ships.

Claims and skepticism (front, chapters 2 and 3)

• Phillip Nelson. “Information and Consumer Behavior.” Journal of Political Economy 78(2), 1970.
• Michael R. Darby and Edi Karni. “Free Competition and the Optimal Amount of Fraud.” Journal of Law and Economics 16(1), 1973.
• Gary T. Ford, Darlene B. Smith and John L. Swasy. “Consumer Skepticism of Advertising Claims: Testing Hypotheses from Economics of Information.” Journal of Consumer Research 16(4), 1990.
• Federal Trade Commission. Policy Statement Regarding Advertising Substantiation, 1984. Health Products Compliance Guidance, December 2022.
• PowerReviews data as reported by the Spiegel Research Center, “From Reviews to Revenue,” 2021.

Reviews and sales (chapter 4)

• Judith A. Chevalier and Dina Mayzlin. “The Effect of Word of Mouth on Sales: Online Book Reviews.” Journal of Marketing Research 43(3), 2006. NBER working paper.
• Michael Luca. “Reviews, Reputation, and Revenue: The Case of Yelp.com.” Harvard Business School Working Paper 12-016, 2011, revised 2016.
• Sherry He, Brett Hollenbeck and Davide Proserpio. “The Market for Fake Reviews.” Marketing Science 41(5), 2022.
• Tom Collinger, Edward Malthouse, Ewa Maslowska and Yorgos Askalidis. “How Online Reviews Influence Sales.” Spiegel Research Center, Northwestern University, 2017.
• Ewa Maslowska, Edward C. Malthouse and Vijay Viswanathan. “Do customer reviews drive purchase decisions? The moderating roles of review exposure and price.” Decision Support Systems 98, 2017.

Who writes, herding and drift (chapters 5 to 7)

• Jakob Nielsen. “Participation Inequality.” Nielsen Norman Group, October 2006. Clay Shirky, Here Comes Everybody, Penguin Press, 2008.
• Nan Hu, Paul A. Pavlou and Jennifer Zhang. “Overcoming the J-Shaped Distribution of Product Reviews.” Communications of the ACM 52(10), 2009.
• Leif Brandes, David Godes and Dina Mayzlin. “Extremity Bias in Online Reviews: The Role of Attrition.” Journal of Marketing Research 59(4), 2022.
• Eric T. Anderson and Duncan I. Simester. “Reviews Without a Purchase: Low Ratings, Loyal Customers, and Deception.” Journal of Marketing Research 51(3), 2014.
• Spiegel Research Center. “Self Motivated vs. Prompted Reviews,” 2021.
• Lev Muchnik, Sinan Aral and Sean J. Taylor. “Social Influence Bias: A Randomized Experiment.” Science 341(6146), 2013. Counts from Smithsonian, 2013; Sean Taylor’s summary.
• Matthew J. Salganik, Peter Sheridan Dodds and Duncan J. Watts. “Experimental Study of Inequality and Unpredictability in an Artificial Cultural Market.” Science 311, 2006.
• Xinxin Li and Lorin M. Hitt. “Self-Selection and Information Role of Online Product Reviews.” Information Systems Research 19(4), 2008.
• David Godes and José C. Silva. “Sequential and Temporal Dynamics of Online Opinion.” Marketing Science 31(3), 2012.
• Wendy W. Moe and David A. Schweidel. “Online Product Opinions: Incidence, Evaluation, and Evolution.” Marketing Science 31(3), 2012.
• Evan Miller. “How Not To Sort By Average Rating,” 2009, and its companion on ranking items with star ratings.
• Federal Trade Commission. FTC Brings First Case Challenging Fake Paid Reviews on an Independent Retail Website (Cure Encapsulations), February 26, 2019.

Asking and showing (chapters 8 and 9)

• Gordon Burtch, Yili Hong, Ravi Bapna and Vladas Griskevicius. “Stimulating Online Reviews by Combining Financial Incentives and Social Norms.” Management Science 64(5), 2018.
• Baojun Gao, Jing Wang, Xiaojie Ding and Yue Guo. “The Pitfalls of Review Solicitation: Evidence from a Natural Experiment on TripAdvisor.” Management Science 71(2), 2025.
• Federal Trade Commission. Featuring Online Customer Reviews: A Guide for Platforms, January 2022.
• Google. Maps user-generated content policy: prohibited and restricted content, read September 2026.
• Danit Ein-Gar, Baba Shiv and Zakary L. Tormala. “When Blemishing Leads to Blossoming: The Positive Effect of Negative Information.” Journal of Consumer Research 38(5), 2012.
• Federal Trade Commission. Fashion Nova will pay $4.2 million, January 25, 2022; order finalized, March 2022.

Origin, handmade and effort (chapter 10)

• George E. Newman and Ravi Dhar. “Authenticity Is Contagious: Brand Essence and the Original Source of Production.” Journal of Marketing Research 51(3), 2014. Yale Insights summary.
• Christoph Fuchs, Martin Schreier and Stijn M. J. van Osselaer. “The Handmade Effect: What’s Love Got to Do with It?” Journal of Marketing 79(2), 2015.
• Ryan W. Buell and Michael I. Norton. “The Labor Illusion: How Operational Transparency Increases Perceived Value.” Management Science 57(9), 2011.
• Ryan W. Buell, Tami Kim and Chia-Jung Tsay. “Creating Reciprocal Value Through Operational Transparency.” Management Science 63(6), 2017.
• Federal Trade Commission. Made in USA Labeling Rule, Federal Register, July 14, 2021; 16 CFR Part 323; Made in USA enforcement, 2021 to 2024; warning letters, July 8, 2025.
• Federal Trade Commission. Williams-Sonoma will pay record $3.175 million civil penalty, April 26, 2024.

The rules (chapter 11)

• Federal Trade Commission. Final rule banning fake reviews and testimonials, August 14, 2024; 16 CFR Part 465; warning letters to 10 companies, December 22, 2025.
• Federal Trade Commission. Adjustments to Civil Penalty Amounts, Federal Register, January 17, 2025 ($53,088 per violation from that date).
• Davis Wright Tremaine. On the FTC’s revised Endorsement Guides, July 2023. FTC, Notice of Penalty Offenses on endorsements, October 13, 2021.
• Federal Trade Commission. Consumer Review Fairness Act: What Businesses Need to Know, 2017.
• Federal Trade Commission. Green Guides; request for comment, December 20, 2022. BWD Strategic, Green Guides status, checked September 23, 2026. Packaging Dive, “Green Guides updates still pending,” February 2025.
• European Commission. Questions and answers on Directive (EU) 2024/825. Irish Department of Enterprise, Empowering Consumers for the Green Transition.
• European Commission. “Protecting consumers from misleading reviews: 55% of screened websites violate EU law,” January 20, 2022.
• Digital Markets, Competition and Consumers Act 2024, Schedule 20; commencement on April 6, 2025 by S.I. 2025/272. Competition and Markets Authority, Fake reviews guidance (CMA208) and Consumer protection enforcement guidance (CMA58), April 4, 2025.
• Competition Bureau Canada. Keurig Canada to pay $3 million penalty, January 6, 2022.

Grading the canon (chapter 12)

• Noah J. Goldstein, Robert B. Cialdini and Vladas Griskevicius. “A Room with a Viewpoint.” Journal of Consumer Research 35(3), 2008.
• Gerd Bohner and Lena E. Schlüter. “A Room with a Viewpoint Revisited: Descriptive Norms and Hotel Guests’ Towel Reuse Behavior.” PLoS ONE, 2014.
• Sheena Iyengar and Mark Lepper. “When Choice Is Demotivating.” Journal of Personality and Social Psychology 79(6), 2000. Benjamin Scheibehenne, Rainer Greifeneder and Peter M. Todd, “Can There Ever Be Too Many Options? A Meta-Analytic Review of Choice Overload.” Journal of Consumer Research 37(3), 2010.