Adding a Review Filter to Bokksu's Product Page — Turning a Buried 4.8 Average Into an Answer at the Point of Doubt

Hypothesis: Meeting a specific doubt with the specific reviews that answer it — instead of asking the shopper to find and filter 12,036 reviews themselves — shortens the path between doubt and evidence to one tap, without touching the reviews themselves.

Brand of the Week

Bokksu built its name shipping curated boxes of Japanese snacks, candies, and teas sourced directly from small, often multi-generation family makers — an authenticity story in a category that's since filled up with lookalike snack boxes. Plans run $32.99–$39.99/month depending on term length, so signing up is a recurring commitment, not an impulse buy, which raises the bar for the proof a first-time visitor needs before checkout.

The challenge

We took a peek at Bokksu's product page, and three things caught our attention.

12,036 reviews at a 4.8 average sit 3,564 pixels down the page — with no star rating or review count anywhere above them. Verified: zero proof elements before the widget.

A star average answers "is this good in general." It doesn't answer "is this worth it for me" or "does shipping actually work" — the specific doubts VOC shows shoppers are actually asking.

The VOC also shows active distrust of on-site reviews generally, so a generic average doesn't just fail to reassure — it reads as one more thing to be skeptical of. Anyone who scrolls down looking for real answers ends up a long way from the only buy button on the page.

So what would we A/B test?

Control: Bokksu's product page as it stands today — picker, detail copy, and a Yotpo review widget sitting 3,564px down, with no proof of any kind above it.

Variant (built in MidaGX): Five jump links inserted at the seam between the picker and the detail copy — "Worth the price?", "Arrives okay?", "Snack quality?", "The flavors?", "Any problems?" — each one clearing any active filter, applying the matching native Yotpo topic filter, and scrolling straight to the result. On mobile, a sticky bar carries the selected topic and live count as the shopper scrolls, with a CTA that fires the real add-to-cart button.

Our hypothesis

Meeting a specific doubt with the specific reviews that answer it — instead of asking the shopper to find and filter 12,036 reviews themselves — shortens the path between doubt and evidence to one tap, without touching the reviews themselves. Nothing is copied or rewritten; the native widget stays the system of record, which matters given how much of the doubt here is about authenticity in the first place.

The CRO principle underneath

This is the proximity principle at work: a trust signal only converts if it appears at the moment the doubt is live, not as a blanket average sitting somewhere else on the page. The average was never the problem — its distance from the decision was.

Want to run a test like this on your own site? Get started with Mida for free — now available even if you don't have an account with Mida.

Related patterns

This is the second brand-of-the-week teardown in the series. The first looked at a different kind of on-page distance problem: Adding a Comparison Table to SuperSaaS's Pricing Page.