Amazon iPhone app and feature-incomplete products

Amazon recently released their native iPhone application. As far as iPhone apps go, it's definitely good - at least on par with the eBay app (which is fantastic). But this is not a review of the app itself, I wanted to focus specifically on an interesting feature included in the app called Amazon Remembers. It allows you to take pictures of items you're interested in, post it to Amazon, and it then gives you the ability to buy the item from Amazon right from your phone.

The objective is great, but I have some thoughts on the implementation, which I think severely decreases the usefulness of the feature. First, a brief walkthrough of how it works:

	Clicking on the "Remembers" tab in the bottom brings up this description - succinct, well-written and with a clear call to action. Note how expectations are downplayed - "we'll even try to find you a link...". They obviously don't want to set expectations too high - understandably, because Amazon might not carry the product, or they may not be able to identify the product from your photo. I think this kind of transparency is really important - setting expectations appropriately is crucial to establish trust with customers.
	I tested the feature by taking a picture of a book I'm currently reading. The process is simple - you take the picture with the phone's camera, and upload it to Amazon. The "what happens to my photos" link is interesting, and addresses not only privacy, but also provides a little more info on how they go about finding the product on Amazon. They seem to rely heavily on a community of users who look at the photos and try to find matching products on Amazon. In my opinion this is both the app's greatest strength and its central drawback - it improves accuracy dramatically over automatic image recognition, but the asynchronous nature takes away a big chunk of the usefulness of a feature like this. More on that later...
	About an hour after uploading a photo of the book, I receive an email to say that my book was found on Amazon. The email includes a link to view the item in Amazon iPhone app (or Web browser if you prefer).
	Clicking on the iPhone app link takes me to the book, and I can buy it easily from there.

This is, without a doubt, great technology - it works great and it's easy to use. But as we know by now, cool technology isn't enough if there isn't a real user need being met. As I mention above, the central problem with the implementation of this feature is its asynchronous nature. Relying on a user community to review/provide input on the image definitely improves accuracy - but it creates the time lag that takes away much of the utility.

Let's step back and think of a few use cases for product image recognition on mobile devices:

• Price comparison to see if you can get it cheaper online than in the store you're in.
• Reviews to help you decide which product to buy.
• Longer manufacturer descriptions or details about a product you are looking at in a store.

The problem is that fulfilling these needs is useful only while you are in the store. If you have to wait (even an hour) to get the information, you're going to move on and either buy the product in the store anyway, or decide you'll check online later without also having the store experience of being able to view the product in person.

This brings me to my main point. This is a classic example of a product solution where developers failed to stay focused on the real user need, and therefore went out the door with a feature-incomplete product that will now probably not get the usage it deserves. The tragedy is that it's so close - once automatic image recognition gets more accurate, this app will be perfect (SnapTel, for example, is already doing a great job at this). My fear is that, if this current implementation doesn't get a lot of usage, the whole idea will be dropped, instead of what should happen - developers going back to users and understanding what their needs are. Someone else will come along and figure it out, yes, but Amazon has a chance here to be industry leaders if they stick with it.

Larger implication

So if you're looking for a larger user experience design implication in this example, it would be that launching products that are not feature-complete is dangerous for three main reasons:

• Fixing issues after a product launches is expensive
• It's painful to make changes later on - users get used to the product and are extremely resistant to change (Exhibit A: the "New Facebook")
• If it doesn't work at first, the entire product may be eliminated because the business benefit isn't immediately apparent

This is why a product design approach that allows developers to fail early, cheaply, and often, is so important. We can't test-and-learn our way into great products.