Hacker News new | past | comments | ask | show | jobs | submit login
The $300 Million Button (uie.com)
113 points by astrec on Jan 14, 2009 | hide | past | favorite | 43 comments

It must be Best Buy. Their checkout copy is quoted in this thread from 2005, and matches the article exactly.


"you do not need to create an account to make purchases on our site. Simply click Continue to proceed to checkout." "to make your future BestBuy.com purchases even faster, you can create an account during checkout."

Best Buy had close to $25 bil in revenue in 2004 and 2005.

I think an implicit lesson in this story is that you should have a system to track the way visitors travel through your site (access patterns). With this in place they could have seen at a glance from the very beginning that a lot of customers balked at the register form immediately before checkout.

I agree, but knowing the bounce rate of that page doesn't necessarily indicate that the login form was the issue.

You may be surprised at how high the the "cart abandonment" rate is on ALL e-commerce sites. With nothing to compare to, perhaps they just assumed those people were a lost cause.

Are there any good standard tools already out there that handle this type of tracking?

This vividly shows how important usability is.

It also shows how much users resent registration - which comes up here at least once a day.

It's not quite as obvious for a shopping site. Users are going to all the necessary information anyway, but if you pose the question as "Would you like to save this information" at the end instead of "Create an account to get started", it makes a world of difference.

Indeed. Though there are some fairly easy steps to reduce this as much as possible.

1. Give guests full use of all the features of the site. Posting, editing, befriending, etc. Store this information by identifying them by the combination of IP + username (which they have to enter for each post) or something similar. This will encourage them to participate (no barrier to entry) and let them decide whether or not they want to become a full member.

2. If they register, link everything they posted while they were guests to the new account, i.e. change the username on all the old posts (if necessary), etc. That way they don't lose all their stuff.

3. The only required information should be a username and password. An e-mail address should be optional, with a warning that if you don't provide one you won't be able to recover your account if you lose your login details. Naturally you should be able to provide one later if you decide you really don't want to lose the account. This keeps the "You just want to send me spam" people happy.

4. No mandatory confirmation email. When you click submit on the registration form you should be logged in and ready to go. Opening your mail client/website is an unnecessary and annoying step.

I might have missed one or two points, but adhering to these rules should take the majority of the pain out of registering.

Yep. Why does it seem like most new sites still roll their own reg? OpenID/Facebook Connect/etc all seem like better choices.

It's a usability nightmare. No one knows what OpenID is, I only heard of Facebook Connect from you mentioning it.

I once tried to establish an OpenID account for stackoverflow, but it was too much of a hassle so I didn't bother registering. Users can just reuse names and passwords. It sucks from a security perspective, but unless you can make it transparent to the user, you're pretty much boned because no one will want to use it.

Because facebook may not be around next year. In fact, given the way things are going for them, I would not be surprised if your site outlived theirs.

When it's something as crucial as keeping track of your customers and users, don't "outsource" it unless there is a very, very compelling reason.

Facebook covers only a small percentage of people online.

Another of my favourite examples from Luke Wroblewski's book is "A Case Study of Explosive Ordnance Disposal" by Jack Moffett:

"When “Number of Deaths” is one of the field labels, you can be certain that every design decision is an important one."

Fairly vivid too!

This must be Amazon right? I mean, you can count on one hand the number of retailers that do $300m total online, let alone some multiple.

Not Amazon. The article said that customers purchasing went up immediately by 45%, $15M per month. Since the company was at $25Bn it was obviously a brick-and-mortar company.

Probably. Amazon's market cap is ~$21b now, which makes sense with "the message I received from the CEO of the $25 billion retailer".

Amazon wouldn't make such an obvious mistake. besides they would have found it in a/b split tests which they use extensively.

Nearly every big company uses a/b splits extensively.

Also, by your logic, every time a big revenue increasing optimization is found, one could say "if this were true, (that company) would have already found it had it existed."

Reminds me of the story of the economist, who finds a hundred dollar note on the ground: "Must be fake --- otherwise somebody would have already picked it up."

You're right...

I went out and searched for screenshots myself, and found this one: http://www.webreference.com/programming/xul/amazon.png which shows it more clearly.

Damn, I hate being wrong :-(

And sorry Matt for being so persistent in being wrong. :-)

Ha, no problem.

Not that I know anything about Amazon specifically, but I wouldn't be so sure. Sometimes it's hard to step back and change something as fundamental as requiring users to register before they purchase. Even if people inside the company had said it was a good idea it might have taken and outside "expert" to get it done.

in usability this is a pretty common mistake that most people in the field would know about. besides, as I said, even if they are dilletantes they would have caught it in split testing.

They may have in the past. As it is now it'll ask you for an email address, make the sale, and remember any relevant information if you want to sign up after, including your CC if you want to make it your account's default payment option.

They definitely would have years ago.

No they wouldn't. Here are a few reasons:

- The checkout path is the most important part of a site that sells consumer products, the people you lose here are ready to buy: If you lose them during the process it's money right out the window. For this reason it's the most optimised part of any serious retail site.

- This is a pretty standard trap, and any usability professional will know this. I would be very surprised if Amazon, even years ago, was designed by Jeff Bezos 14 year old cousin.

- Amazon is known for running split tests and numbers on their site to see how something works. They have done this since they launched. The thought that they somehow didn't do so on the checkout path, the most important part of the site, is pretty far fetched.

- Jeff Bezos is a numbers guy - he used to be a stock broker, and is crazy about optimising and testing.

It's like saying Steve Jobs approved the new line of Ipods being Scottish tartan with the user interface outsourced to Microsift.

he used to be a stock broker,

He was a quant. Huge difference. Any idiot could be a broker.


Well, they almost certainly did if this story is true. Please point me out one other company that's worth around $25b and has the online sales to do improve by $300m per year. Just one. Regardless of what you think about Amazon specifically, who else fits this profile? It's not as if there are $25b companies doing hundreds of millions in sales online that you don't know about.

This is a standard trap now, but he was a numbers guy back when it wasn't, and he hired someone to optimize and test. They found that error.

> It's like saying Steve Jobs approved the new line of Ipods being Scottish tartan with the user interface outsourced to Microsift.

I know you're trying to be funny, but soon after Jobs' return to Apple after the "NeXT exile," things really did seem that surreal for a bit...



It's nice of Matt to find counterexamples to your arguments. But you should know that pointing out that you use circular logic is enough to show that your argument is not valid (in other words, no counterexample is really needed). Now, this doesn't necessarily say anything about whether this company is Amazon or not, but it does show that your proof, that there is no way this is Amazon, does not hold water.

(The first two quoted lines are my own interpretation, others are quotes.)

"Amazon is a big company. Big companies don't make amateur mistakes. That's why they're big."

"Therefore, Amazon doesn't make amateur mistakes."


(Repeating this type of logic over and over is...)


"The checkout path is the most important part of a site that sells consumer products, the people you lose here are ready to buy: If you lose them during the process it's money right out the window. For this reason it's the most optimised part of any serious retail site."


"This is a pretty standard trap, and any usability professional will know this."


(Continuing to post more arguments feeling you must be right since no evidence has been provided to the contrary, is...)


"The thought that they somehow didn't do so on the checkout path, the most important part of the site, is pretty far fetched."

http://en.wikipedia.org/wiki/Straw_man (nobody said they thought Amazon absolutely decided to skip usability testing of checkout functionality. Also, for example, they could have tried but did not think of changing it, or ran out of good subjects who hadn't already used Amazon's checkout page, bad selection of candidates, poor interpretation of results from the split tests, or had lots of bureaucracy or competing work or designs of checkout incompatible with such changes, human error, that this looks easy in retrospect but not something they thought of at the time, person in charge of that being complacent, etc.)

"new line of Ipods being Scottish tartan"

That's just the customized Carnegie Mellon version.

I appreciate the core point about the form. But something in me hesitates to believe that this all happened quite the way he says.

I did usability for a retail site two years ago that had pretty much the same problem, and solved it the same way.

I would say that the numbers are probably accurate.

Ok, you I trust. That other guy... :)

The site this article is referring to could be "Coolhorse" http://www.coolhorse.com/mm5/merchant.mvc?Screen=OINF&St...

EDIT:Although I do not see how this one could make 25B a year.

I think there's some shopping cart software that has similar language in it. That's probably what Coolhorse is using.

looks like some good ol' oscommerce

It's interesting that users assume you are jerks when you ask them to register no matter what your brand, or the information you ask for.

Registration has become synonymous with ways to sneak in email newsletters.

Jared Spool is also a fantastic speaker. If you ever get a chance to see him give a talk (e.g., at a SIGCHI meeting), by all means do.

This is more about obstacles in their checkout process rather than any button, but still important lessons to keep in mind.

Wow this is very interesting but makes perfect sense. People just don't like registering for websites.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact