A reviews module used by millions of diners, every month, to decide where to eat.
TheFork connects diners with restaurants across the world. Reviews are one of its most-used features — the single signal diners lean on most when comparing options.
60,000
Restaurant partners, in 12 countries
30M
App downloads
20M
Reviews posted by the community
+20M
Visits generated per month
The problem
Same restaurant, three different review experiences · iOS · Android · Desktop
Reviews existed — but they weren't doing any convincing.
Diners had reviews available, but the experience was fragmented across platforms and buried the one line that actually sways a decision — a standout quote, a simple "would you come back?" signal — under generic star breakdowns and long text blocks.
Cross-platform inconsistencies
iOS, Android and Desktop showed different visuals, navigation and even different rating scales for the same restaurant — the community score wasn't reading as one trustworthy number.
Not enough useful information
The restaurant profile surfaced a single review. With very little context about the rating itself, diners had no fast way to judge if a place was actually right for them.
Accessibility issues
Low contrast, small tap targets and inconsistent hierarchy across the reviews section made it harder to use for a meaningful share of diners.
Inconsistent UX writing
The product spoke differently to its cross-platform users — labels, tone and structure shifted from screen to screen, weakening the sense of one coherent product.
Research & learnings
Reviews are read before the decision — and today, over half of diners don't trust what they find.
User journey · When diners read reviews
30%Read reviews when comparing options
20%Read reviews right before the meal
5%Read reviews after the meal
Survey
+20.8% CVR
for TheFork users who read reviews inside the app.
Survey
52% say it's not enough
of users feel the reviews section today isn't useful enough.
Baseline
4/5
is the score diners use as the baseline for identifying a good restaurant.
Behavior
Both sides, always
most users read negative and positive reviews before trusting a score.
Research takeaway
Diners don't want more reviews. They want the one that tells them what they need to know — fast.
Beyond the score, many users specifically look for photos shared by other customers before they trust a restaurant's reputation.
Reviews mission · What I do
Reviews are crucial in making a decision.
Food, value for money, service and noise all determine whether a place is perfect and meets the user's expectations.
Regarding the restaurants, the reviews improve their rating following the highest standards of service, food and atmosphere — so the module has to work for diners deciding where to eat, and for restaurants building their reputation.
01
Impact
Reviews should visibly move the decision to book — not just sit as background noise on a profile.
02
Engage
A reviews experience diners actually want to read, and restaurants want to be rated in.
03
Design standards
One consistent, accessible system across iOS, Android and Web — same score, same voice.
Ask for recommendations, check guides, get inspired by photos, filter/sort
Might search reviews across different platforms — TripAdvisor, Google, etc.
30%
Read a review
Compare & choose
Compare ratings, read reviews, check photos and menu items, verify availability
Secure a table
Choose the best price/deals, ask for special requests
20%
Read a review
Prepare
Share booking info with other diners, check address, verify how to get there
Diners invited to a dinner often skip the comparison step entirely
Enjoy the meal
Seat, wait, eat, drink, discuss, take photos
Pay
Pay by card, cash or phone, tip, split the bill
5%
Read a review
Share the experience
Leaves a review and photos, shares the experience on social media
Often checked against other reviews, to see if opinions align
Based on user survey data · when diners read reviews across the decision cycle
Design exploration · Two directions
Vertical scroll or carousel — I prototyped both before betting on one.
Two proposals came out of the wireframing stage: a vertical-scroll summary that reads top to bottom like a feed, and a horizontal carousel that lets diners swipe through the highlighted quote, the would-return stat and individual reviews as separate cards. Both kept the same information architecture — the difference was how much diners had to scroll versus swipe to get through it.
Highlighted quote
The single most compelling line from real reviews, surfaced above the fold.
"Would return" stat
A simple, structured trust signal instead of raw star math.
Structured review cards
Reviewer, score, date and photos in one scannable, consistent format.
One system, three platforms
Same rating scale, same components and same UX writing on iOS, Android and Web.
01 · Vertical version — full summary, rating breakdown, sorted reviews
Both directions shipped as an A/B test against the old reviews module. The carousel version — leading with the highlighted quote and the would-return stat, then letting diners swipe through the rest — outperformed on both the engagement and the business metric, confirming the original research: diners don't need more reviews, they need the right one, fast.
+25%
In read reviews
+6%
In CVR
52% →
"Not useful enough" — the baseline this was built to fix
The winning carousel, in detail · summary, frequency-word filters, sortable reviews
For diners
Decide faster
The most persuasive detail is visible before scrolling — no more digging through star breakdowns.
For restaurants
A fairer reputation
A consistent scale and standardized card mean a restaurant's rating reads the same everywhere.
For the product
One design system
A single, accessible reviews component shared by iOS, Android and Web — validated with data, not just opinion.
