PlatePost — Manoj Venkatesun

The solution

PlatePost adds a short video for each dish, so you can see it before you order.

Every restaurant got its own menu, but they all ran on the same template. Three things stayed the same: the item popup, the category bar, and the footer. The look changed for each brand, like fonts, colors and layout. One menu took me about two or three days.

MakiMaki Sushi menu with a three column grid of sushi rolls. Each roll has a photo on a white plate, with its name, ingredients and price beside it. — MakiMaki Sushi.

Cafe Yoto menu in rust red and monospace type. Round photo icons for each category sit in the top bar, above a four column grid of udon bowls, each with a small video icon. — Cafe Yoto.

Kei Coffee House menu with a black category bar of photo icons. Under an italic serif heading, a row of photo cards for drinks and food, each with a Best Seller tag, name, description and price. — Kei Coffee House.

Irv's Burgers menu with a blue top bar of category icons. Burgers and a hot dog sit on flat yellow, red and blue blocks, each with its name and ingredients below. — Irv's Burgers.

The problem

Most restaurant menus are just a name and a price, so you end up guessing what the food looks like.

I joined PlatePost in April 2024 as the only designer and built the VideoMenu template. I was the only designer for 21 months.

Brandon, the founder, handled marketing and brought in the restaurants. I was the technical side: I built the menus and worked with each restaurant's point of contact.

Key decisions

Designing for each restaurant

WYWH has three services in one shop: coffee, brunch and cocktails. So the menu starts by asking which one you want, instead of showing a wall of food.

WYWH Coffee menu home with three photo cards to pick from: Coffee, Brunch and Cocktail Lounge. Each card shows its days and hours, with an Order Online button below. — WYWH Coffee. Each service card shows its hours.

MakiMaki didn't have photos for the drinks. Instead of leaving empty gray boxes, I made those items a simple list with just the name and price.

MakiMaki drinks section and sake and beer section. Each item is a name and price in a three column list, with no photos and no empty image boxes. — MakiMaki's drinks, with just the name and price.

Cafe Yoto has Japanese dish names that are hard to say. I added a speaker button that says the name out loud, so you don't have to guess or point.

Making the card tappable

At Kei, people kept tapping the food cards and nothing happened. So I made the whole card tappable. It opens a popup with the video on top and the details below.

Each menu got its own tap hint: "Click to view video" at Kei, a small arrow at APE, and a play icon at Cafe Yoto.

APE Coffee Orange seasonal drinks in black and white. A large video still sits on the left. On the right, a list of drinks, each with a small photo that has an arrow in the corner. — APE Coffee, with the arrow on the corner of the photo.

Edge cases I designed for

What if a restaurant has no photos for some items?
Those items become a plain list with just the name and price. No empty gray boxes.
What if one shop runs more than one menu?
You pick the service first, like coffee, brunch or cocktails at WYWH.
What if a dish name is hard to say?
A speaker button next to the name says it out loud.
What if people tap a card that does nothing?
The whole card is tappable now, and each menu shows its own hint that there's a video.

Results

We ran a 4 week test at Kei. About 1 in 5 customers scanned the QR code (5,798 scans out of 26,836 orders). We were hoping for 1 in 4.

The average order went from $16.05 to $16.09. So people did use it, but the numbers can't tell us much more than that.

Table of the 4 week test at Kei, one row per week from Aug 25 to Sep 21. Columns show QR scans, total orders, scan rate, average order value and the difference from baseline. In total, 5,798 scans on 26,836 orders, a 21.6% scan rate and a $16.09 average order. A note says the baseline was $16.05, and week 4 fell below it during a back to school promotion. — The test at Kei, week by week.

How I got there

From my first client to my latest

WYWH was my first client. On that menu, the photos are hidden behind a small camera icon, so you have to tap to see the food. By the time I made APE, the photo sat right on the card and tapping it played the video. I got better at this with every client.

WYWH Coffee brunch menu. Each item is a blue name, a short description and a price, with a small camera icon on the right. One photo of a breakfast burrito is open on top of the list. — WYWH Coffee, my first client. You tap the camera icon to see the food.

What I wanted to build next

I also designed a second version of Kei's menu. It had photo categories, a Click to Order button, an AI pairing suggestion, and nutrition info. Brandon liked it, but it never shipped. I was also worried we couldn't keep the nutrition numbers correct.

Three phone screens from the second version of Kei's menu. The first shows photo category chips above a grid of drinks with Best Seller and Staff Picks tags. The second shows an open drink with a blue AI Suggest Pairing button and a black Click to Order button. The third shows a nutrition panel and ingredient icons. — The second version of Kei's menu. It never shipped.

What I'd do differently

I'd start building a proper app for the menus sooner, instead of making each one by hand. Adding a new restaurant would have been one row in a database, and that kind of app only gets harder to build later. I haven't built it yet.

I'd also work out how to measure the results before the test, not after.

PlatePost: Designing a video-first menu platform for restaurants

Problem

What I did

Result