The $300 million button: “Continue” instead of “Register”

• The primary source of the story is Jared M. Spool’s article of January 14, 2009. The company’s name isn’t in it.
• The problem was the required registration behind the button. New customers didn’t want to register; returning ones couldn’t remember their password.
• The fix was small: “Register” went, and “Continue” came with a short explanation.
• The figures are as Spool reports them. The article doesn’t explain how they were measured.

The story: two fields, two buttons

The story is told by Jared M. Spool. Luke Wroblewski asked him for a form example while writing his book Web Form Design: Filling in the Blanks. Spool wrote this text, and Wroblewski published it in his book.

The form was perfectly ordinary. When customers pressed the button to pay in the cart, a sign-in screen appeared before the card details. It had two fields: email address and password.

Under the fields were two buttons, “Login” and “Register”, plus a “Forgot password” link. The team had built the form so that returning customers could buy faster.

What the test showed

Spool’s team ran usability tests with people who actually had to buy something on the site. Two groups got stuck in two different places.

First-time shoppers resented being asked to register. As one shopper put it in Spool’s article: “I’m not here to enter into a relationship. I just want to buy something.”

Some new customers didn’t even know whether they had created an account before. When their guess was wrong, they got frustrated. It is a good example of a difference in mental model: the team was thinking of an “account”, the customers of a “purchase”.

Returning customers fared no better. Most no longer knew which email address or password they had used. Many had also changed their email address over the years.

What the database showed

The team then looked at the customer database. 45% of customers had more than one entry, some as many as 10. Anyone who couldn’t sign in simply registered again.

The site received about 160,000 password reset requests a day. 75% of those who requested a password never tried the purchase again. The password recovery screen didn’t bring customers back; it lost them.

The change and the result

The designers removed the “Register” button and put “Continue” in its place. Next to it came a short message saying that you don’t need an account to make purchases and can create one during checkout if you like.

According to Spool, the number of customers purchasing went up by 45%. The extra purchases brought in about $15 million in the first month. By the end of the first year, the extra revenue was $300 million. That is where the story gets its name.

Spool doesn’t name the company. He describes it as a “major e-commerce site” and a “$25 billion retailer”.

Source check: what we know and what we don’t

On this page I only report what is in Spool’s article. The table shows which claims have a source and which don’t.

Topic	Status	Note
Author and date	Known	Jared M. Spool. The article was published on January 14, 2009.
Who ran the tests	Known	Spool says “we” ran the tests and names no one else.
Structure of the form	Known	Two fields, two buttons, one “Forgot password” link.
45% duplicate entries, 160,000 password requests a day, 75% lost	As Spool reports	Said to come from an analysis of the database. The raw data isn’t published.
45% more purchases, $15 million in the first month, $300 million in the first year	As Spool reports	The measurement method isn’t in the article. We don’t know whether there was a controlled test.
Name of the company	Unknown	Spool doesn’t give it. He describes it as a “$25 billion retailer”.
Difference between new and returning customers after the change	Unknown	The article gives only the overall result.

I haven’t found a source that confirms the company’s name. That is why I name none on this page.

Because the measurement method is missing, I don’t read the $300 million as a firm result of an experiment. A promotion or a seasonal effect may have played a part in the same period. The article says nothing about that.

For me, the most solid part of the story is the data on duplicate accounts and password resets. You can measure both problems in your own database today.

Reconstruction: try both flows

Spool’s article has no screenshot of the form, only a description. From that description I drew two flows side by side: the old one with required registration and the new one where you can continue as a guest.

Use the buttons to go through both flows to the end. It also works with the keyboard: Tab to the button, then Enter. The counter above each flow shows screens, form fields and clicks.

Try both flows

Before: “Register”

6 screens, 12 form fields.

1. 1. Cart

   You have 1 item in your cart.

   Checkout
2. 2. Sign inEmailPassword

   Forgot password

   LoginRegister
3. 3. RegisterEmailPasswordPassword (again)
   Create account
4. 4. Shipping addressFull namePhoneStreet addressCity, state and ZIP
   Continue to payment
5. 5. PaymentCard numberExpiration dateSecurity code
   Place order
6. 6. Order received

   Done. But first you had to create an account.

After: “Continue”

5 screens, 10 form fields, account optional.

1. 1. Cart

   You have 1 item in your cart.

   Checkout
2. 2. Sign in or continueEmailPassword

   You don’t need an account to buy. Just continue; if you like, you can create an account during checkout.

   LoginContinue
3. 3. Shipping addressEmailFull namePhoneStreet addressCity, state and ZIP
   Continue to payment
4. 4. PaymentCard numberExpiration dateSecurity code

   Optional: create an account with these details.

   Place order
5. 5. Order received

   Done. Creating an account was your choice.

A reconstruction I drew from Spool’s description of the form, not a screen from the real site. Go through both flows to the end and watch the counters. The “Login” button stands for the shopper who can’t remember her password.

At first glance the difference looks small: one screen and two fields. But the counter doesn’t show the real cost.

In the old flow, instead of payment the shopper meets a memory question: did I have an account? Anyone who can’t remember the password goes through “Forgot password” into their inbox. In Spool’s case, 75% of those who took that path didn’t come back to buy.

In the new flow, the same shopper doesn’t have to remember anything. She types her email address once along with the address and goes to payment. If she wants an account, she ticks a single box on the last screen.

Open this demo in a team meeting and ask everyone to try both flows. The discussion moves from personal opinions to numbers.

The lesson for today

Don’t require an account before payment. Keep the path to continue as a guest open and visible. Make the account optional and move it to the end of the flow.

That is exactly what the fix Spool describes did. The account stayed; it just got out of the customers’ way.

The problem didn’t stay in 2009. On the Baymard Institute’s cart abandonment list, among the reasons other than “I was just browsing”, “The site wanted me to create an account” stands at 18%. On the same page, the average abandonment rate across 50 studies is 70.22%.

A button label makes a promise

To customers, “Register” reads like a commitment. “Continue” only names the next step. Same button, different promise.

Compare a label with the goal the customer has at that moment. Someone at the payment step wants to buy, not to create an account.

No memory test at checkout

A password requires recalling information from memory. The sixth of Nielsen’s 10 heuristics, “recognition rather than recall”, says to reduce users’ memory load. Asking for a password at checkout means asking for the hardest task at the most sensitive moment.

Look at your own data

Instead of putting this story on a slide, measure it on your own site. Start with these questions:

• How many password reset requests come in a day, and how many of them end in an order?
• Are there several accounts for the same person (same phone number or address, different email)?
• Where do sessions go that leave the sign-in screen at the payment step?

Don’t do guest checkout halfway

Opening the guest path may not be enough on its own. If the guest option stays a small text link, it is easy to miss. If the main button still says “Register”, the shopper heads for that door again.

Move the account offer to the order confirmation. At that point you already have the email address and shipping address. Asking for a password is enough.

Think about tracking guest orders too. A page where the order can be tracked with the order number and email saves customers without an account a call to support.

Once you have these numbers, you see the size of the problem in a few hours. How to review a checkout step by step is in the UX audit guide and the checkout pattern.

A five-minute test for your site

Open a private window in your browser and add a product to the cart. Go to checkout and count how many screens, fields and decisions you meet before you reach the card details.

Can you continue without creating an account? Do you have to look for the way to do it? If you answer “no” to the first question or “yes” to the second, you have a button story too.

I can review this flow for you: UX audit for your checkout.