NoFakes · Verified review marketplace
Reviews you can only write after buying
NoFakes only accepted reviews from people who could prove a purchase. I designed the proof, the review, the trust signal and the rewards as one loop: worth the effort for the reviewer, and visible to the next customer.



Customers search local businesses and read opinions that are tied to a real purchase.
Results
The loop worked on both sides of the marketplace
Every number below depends on the whole loop working: a purchase, proof, a review and a reason to come back. Roughly one verified review was written for every two registered users, despite a verification step most review products avoid.
Figures at project close, from Amplitude analytics and internal commercial data.
The problem
Anyone could write a review, so nobody trusted them
Reviews drive buying decisions, but fake ratings are cheap to produce. Customers learn to discount every review. Honest local businesses lose too, because their real reviews look no different from bought ones.
Most platforms try to catch fakes after they are published. NoFakes made a different bet: no review without proof of a real purchase.
The design question
How much effort will people spend to be believed?
Proof makes a review credible, and every step of proof costs the reviewer effort. Rewards bring people back, and paid-for reviews are exactly what the product existed to fight. Every decision in this case is an answer to one of those two tensions.
The loop
Trust had to work as a loop, not as a badge
A verified badge means nothing on its own. It only carries weight if every step before it holds. I designed the product as five connected steps, so a change to one step could be judged by its effect on the others.
The first step happens offline, at a restaurant or a shop. Everything after it happens in the app, and the sections below follow one customer through it. Each reward brings them back for the next purchase.
Verify
Explain the proof before asking for it
Verification was the foundation of the product and its most demanding step. The risk was simple: someone photographs the wrong document, gets rejected and never comes back. So the flow shows what a valid receipt looks like before asking for one.


What counts as proof comes first. The upload comes second.
Both a till ticket and an invoice are accepted, on paper or digital. Showing both up front answers the most common question before anyone has to ask it.
Review
Keep the proof and the opinion apart
A receipt proves someone was a customer. It says nothing about what they thought. I kept the two in separate steps: the receipt establishes where the review came from, and the rating and comment carry the opinion.


The review form, empty and complete.
Separating them kept one rule simple to apply: proof decides whether a review can be published. What the review says belongs to the customer.
Signal trust
Compress all that work into one mark
The next customer never sees the receipt or the form. They see a business page. Everything upstream had to reach them as something they could read at a glance, without explanation.
NoFakes never accepted unverified reviews. Business pages still showed Google reviews for context, labelled with their source, but only reviews verified through NoFakes got the prominent place. The next customer could see the wider picture and still tell which opinions carried proof.

A business page groups verified opinions under one heading, with the mark beside each rating.
Reward
Give people a reason to come back
A trust product with no reviews is empty. Customers earned tokens for verified reviews and exchanged them for gift cards from brands they already used. The risk was obvious: pay for reviews and you get the reviews you pay for.
Tokens grew with the richness of the review: text, photos and video each earned more. The reward paid for effort and detail, the thing that makes a review useful to the next customer.

Gift cards show the balance, the tokens still missing and what has already been redeemed.
- Balance first.The token count opens the screen, so the value of past reviews is visible immediately.
- Progress, not just price.Each gift card shows how many tokens are missing, which turns the next review into a concrete step.
- History kept.Redeemed cards stay on the screen as proof that the rewards are real.
Businesses
Turn verified reviews into decisions
Businesses were the second half of the marketplace. A rating on its own told an owner very little. They needed to know how customers felt, which feedback was verified, and what to do next.
The business hub brought together an overall score, the rating distribution, verified and unverified reviews over time, replies and the business profile. From the same hub, owners could invite customers to review, which fed new purchases back into the loop.
Business dashboard, review management and profile.
What I learned
Trust is a system before it becomes a UI signal
The verified mark was the smallest thing we designed. It only meant something because of everything underneath it: a purchase, a receipt, a review and one rule applied to everyone.
The two tensions never went away. Every verification step protected the system and cost effort. Every reward brought people back and had to leave the reviews it produced just as credible. The job was not to remove those trade-offs but to make each one deliberately.
What I would do differently: test a lighter proof earlier. Reading the receipt automatically, before asking for a photo of it, would have shown how much effort people would really spend, before we layered more reward mechanics on top.




