A refund can solve a customer problem.
It can also become leverage.
Imagine receiving a defective product and complaining to the seller. The seller offers to make you whole—but only after you post a five-star review, change your existing review, or remove the negative one.
The money is no longer simply customer service.
It is attached to the public record.
The Federal Trade Commission’s current Consumer Reviews and Testimonials Rule prohibits businesses from offering compensation or other incentives conditioned on a particular review sentiment, whether the condition is explicit or implied. FTC guidance gives the simple example that a business cannot promise a coupon specifically for saying how much a customer loved the experience. See the FTC’s Consumer Reviews and Testimonials Rule Q&A.
Amazon’s current community rules are similarly direct: customers may not create, edit, or remove a review in exchange for compensation including refunds, discounts, gift cards, products, warranties, or services. See Amazon’s Community Guidelines.
The rating stops measuring the original experience
Suppose a customer honestly believes a product deserves two stars.
Then the seller says, “We’ll refund your $80 if you change it to five.”
The eventual five-star review tells future shoppers almost nothing about the original experience.
It may instead measure how badly the customer wanted the refund.
That is why conditional compensation is more serious than an ordinary attempt to resolve a complaint.
A seller is allowed to contact an unhappy customer, fix the problem, and ask whether the customer wants to update a review afterward. The FTC explicitly says ordinary complaint resolution is not prohibited.
The important distinction is whether the remedy depends on changing the public opinion.
Evidence is often unusually concrete
Unlike some reputation manipulation, conditional-refund schemes can leave excellent evidence.
A buyer may receive an email, marketplace message, package insert, WhatsApp message, or support ticket saying exactly what must happen before reimbursement is issued.
That message can establish the condition far more reliably than guessing from a suspicious ratings pattern.
Useful evidence includes the original review, the seller’s request, the promised benefit, timestamps, whether the refund was actually issued, and whether the public review changed afterward.
Without that documentation, a customer receiving a refund and later improving a review does not automatically prove manipulation. Maybe the seller genuinely fixed the problem and the customer voluntarily revised their opinion.
Reputation engineering can happen after the sale
Manufactured Consensus is not limited to fake accounts inventing happy customers.
A business can alter what the public sees by applying pressure to real customers who had real negative experiences.
The customer existed.
The purchase happened.
The disappointment was genuine.
What became manufactured was the version of the experience left behind for everybody else.