Here's the minute that decided how Bounty works.
I'm at a Save-On Foods. There's a yellow tag on the shelf with a price that looks like it might be a deal. I want to know if it actually is. I pull out my phone.
I open a browser. I type the product name. The first three results are the store's own website, an aggregator, and an ad. I tap the store's site. It loads slowly. The price on the page is the same as the price on the tag, which tells me nothing — of course it's the same, they're both coming from the store. I back out. I tap the aggregator. It shows me a flyer. The flyer has the same price. I open another tab to check a different store. Same flyer aggregator. Same kind of layout. The competing store has the product on sale this week too, at a price that's suspiciously close to the first one.
I stand there for a second and realize something I hadn't put into words before. The online flyer and the paper flyer have pretty-much the same prices. The competing store's flyer has nearly the same prices. Everywhere I look at this product, every source is showing me a number that came from somewhere very close to the store I'm standing in. It's not exactly collusion, but it might be... besides if they are publishing the same flyers out there, it's easy enough for them to sync up. There is no independent number anywhere.
I look at my cart. There are maybe 8 things in it. I have just spent two and a half minutes checking one of them, and I don't actually know any more than I did when I started. Shouldn't be this much of hassle to know if something is "a good price".
I put my phone away. I get out of there.
What that minute taught me
That minute is the entire reason Bounty is built the way it is.
If a tool requires me to type, search, sort, scroll, tap, and switch tabs to answer one question about one item, I will not use it. Neither will anyone else. The tool has to live inside the gesture I would already make if I had any way to make it useful — point my phone at the tag and have the answer be there. Beep - just like the checkout.
So that's what Bounty does. You point the camera at the tag. The app reads it. You get an answer. There is no search box. There is no list of results to sort through. There is ideally one tap (or as few as possible), and there is one app. Ooh, that's a rhyme... One Tap, one app... (ick)
What it costs
The honest cost of that decision is that reading shelf tags with a camera is much harder than looking up a barcode in a database. The OCR has to work on tags that are torn, faded, lit from the wrong angle, half-covered by another sticker, printed by a label maker that ran out of toner three years ago. Most of the engineering work in Bounty so far has gone into making that one gesture reliable enough to trust.
I made the trade on purpose. The hard engineering is mine to deal with. The friction is the user's, and the user will not deal with it. If Bounty asks anyone to do what I did in that aisle that day, it has already failed.
— Elmer