Why we don't publish fake ratings.

The shape of the problem

When you see a 4.9-star rating with "312 reviews" on a wedding portal, your brain treats it as social proof. Behind the scenes, that number can be any of several things:

• 312 real, verified couples who actually booked and reviewed (rare).
• 312 enquiries — never booked. Counts every form submission as a "rating".
• 312 social-media mentions scraped from Instagram, Google Maps, and other sites — none verified.
• A handful of real reviews padded with AI-generated ones.
• Family + friends of the venue owner asked to review.
• An entirely made-up number with no source.

Couples can't tell which. The portal doesn't disclose. The venue can't complain because their competitors are doing the same. The system stabilises on inflated, undifferentiated 4.7–4.9 ratings everywhere, and couples learn to mistrust ratings altogether.

The structural fix isn't "add more reviews"

The instinct is: collect more real reviews to dilute the fake ones. That doesn't work for a few reasons:

• Trust collapse is asymmetric. Once couples decide ratings are unreliable, no amount of real reviews restores trust. You have to make the verification mechanism visible.
• The incentive on collectors is bad. A directory that earns more when couples convert has incentive to inflate ratings to drive conversion. The collector has to be different from the financial beneficiary.
• Most couples don't review. Wedding is a one-time event for most people. They won't come back to leave a review unless prompted hard, and prompting hard correlates with positivity bias.

Our solution (and what you'll see now vs later)

For now, we publish nothing fake. Most venue pages show no star rating, no review count, no "trusted by" badge. We'd rather show nothing than show numbers we can't back.

When we eventually ship a verified-review system, it will:

• Be tied to a verifiable booking signal. Most likely: Stripe payment receipt or a signed booking record uploaded by the couple.
• Show the verified-booking count alongside the average. "4.7 stars · 23 verified bookings" is honest in a way that "4.9 stars · 312 reviews" isn't.
• Let venues respond publicly. A negative review needs a defence mechanism that doesn't depend on getting reviews deleted.
• Never let venues delete reviews. Only flag-and-investigate paths for fraud or harassment. Owner-deletes-bad-review is a major source of inflated ratings industry-wide.
• Decline rating data from social scraping. Instagram "hearts" aren't reviews. Google Maps reviews are Google Maps' problem.

What this costs us

Real money. Star ratings drive conversion. By not showing them, we lose conversion versus competitors who do. We're betting that, over time, couples who've been burned by fake ratings choose us, and venues that want to build long-term reputation choose us too.

The bet is also that AI search engines and aggregators reward honest signals more than inflated ones. When Perplexity or Claude is summarising "best wedding venues in the Hunter Valley," the source it trusts more is the one with verifiable claims — not the one with 5,000 fake ratings.

How to spot fake ratings on competing sites

Quick heuristics:

• If every venue on a portal has 4.7+ stars, the ratings are not differentiating real quality. They're padded.
• If a venue has 312 reviews but you can read fewer than 30 actual review texts, the count is inflated.
• If reviews all sound similar ("The venue was beautiful and the staff were so attentive" in slightly different words), AI-generated.
• If a portal makes it impossible to leave a negative review (form rejects 1-star, only verified-yes-prompted couples can review, no public dispute path), the ratings are managed.
• If you can't find a single 1-star review anywhere in the directory, that's a tell.

The wider point

Wedding is a high-stakes, low-frequency purchase. People remember being lied to. The directories that have spent the last decade inflating their ratings have built short-term conversion at the cost of long-term trust. We're betting on the long term.

— Mia