YouTube - Seth Godin: Sliced bread and other marketing delights

Seth Godin: Sliced bread and other marketing delights

Comments (1)

Denis De Mesmaeker

Denis De Mesmaeker Oct 14, 2010

"be remarkable"! Find the people who want to listen to you.

10 Ways Newspapers Can Improve Comments


10 Ways Newspapers Can Improve Comments

The other day Bob Garfield had a good kvetch about dumb comments on newspaper websites on his show, On The Media, and I posted my two cents, but I still don’t feel better. I think that’s because Bob’s partly right: comments do suck sometimes.

So, instead of just poking him for sounding like Grandpa Simpson, I’d like to help fix the problem. Here are ten things newspapers could do, right now, to improve the quality of the comments on their sites. (There are lots more, but you know how newspaper editors can’t resist a top ten list.)

  1. Require Accounts

    Anonymity is important in journalism, but not for comments.

    There are a lot of good reasons to allow anonymity, especially in the news. Sometimes a source needs to speak out against an employer or the government without being named. Fine. But there is no reason, really no reason at all, to allow people to post comments without having to first sign up for an account.

    Simply requiring an account will remove 80% of your comment problems. If allowing anonymity is important, you can allow the user to remove their name on a specific comment, while still requiring them to be logged in. (In other words, the user must log in so the system knows who they are, but they can opt to leave a comment as “Anonymous” if they choose. Anonymous comments could then be held in a special moderation queue for approval to guard against any bad uses.)

  2. Set and Enforce Rules

    Nobody likes finding out about a rule after they’ve broken it. Write a human-readable set of community guidelines (Flickr’s are excellent). Make all new members agree to it when they sign up, and link to it prominently from every comment form. This way, if you have to take action later, you can say “We warned you.”

    Then enforce the rules. Delete bad comments and publicly promote the ones that are great. There’s a common misconception that moderating comments makes you more liable. This is not true. Managing your community does not have any baring on your DMCA compliance, safe harbor standing, or any other legal issue.

  3. Employ a Community Manager

    If you can’t name your community manager, it’s probably you.

    You wouldn’t let a writer put their work in the paper without having someone check it, so why let commenters do so? If you’re going to have people posting comments to your site, it should be someone’s job to moderate them. Think of them as the editor of the Comment Desk.

    You don’t have to read every comment before it goes online, but it should be somebody’s responsibility to remove any comment that runs afoul of the posted community guidelines. Like graffiti in an urban space, bad comments lead to more bad comments. But the Community Manager should be more than a cop – they should be a vital connection between the staff and the community. They should lead the community by example, participating in the discussion and being helpful, and also do a daily “community weather report” for the staff, feeding the community’s input back into the newsroom.

  4. Sculpt the Input

    Just because your users can post comments doesn’t mean you can’t help them shape them.

    Back in the day, when we had people posting comments to Fray, we were constantly tweaking the form’s automated responses. If you tried to post something too short, it asked you to expand on it a bit. If you posted something too long, it asked you to edit yourself down. If you posted in ALLCAPS, we de-capitalized it (Flickr does this now). These are easy things for computers to do, and they make a huge difference.

  5. Empower the Community to Help

    If you think bad comments bug you, they bug the good commenters twice as much.

    Yes, you should be paying someone on staff to be the Community Manager. In addition, you can also enable the community to help. Give every post a “This is Bad” button. Then give the community manager a private page where they can see the comments with the most bad votes and take appropriate action.

    For bonus points, give each post a “This is Good” button, too, so they can also tell you about the good ones. Remember that your members are not the enemy: they want to help you keep the place clean, too.

  6. Link Stories to Comments

    The worst thing you can do is separate the “community section” away from your content. That creates a backchannel, where people feel safe being inappropriate because, why not? They’re at the kids table, anyway.

    So link stories to community conversations as closely as possible. This will give the conversation a central topic.

  7. Enable Private Communication

    The internet didn’t create the angry letter to the editor, but it definitely put it into overdrive. And that’s okay – sometimes people need to vent. Your job is to direct the venting.

    Some papers’ comments are so crazy because there’s no other way for the reader to respond. People will gladly communicate with you privately if you gave them a way to do so.

    So create a form people can use to email the editors, and link to it from the comment form. Say: “If you’d like to say this privately, go over here.” (Props to Vox, where there’s a “Send private message instead” link on every comment form.)

    You may get some angry email this way, but it’s better in your inbox than on the website where it will just start, or add to, a fight.

  8. Participate …

    Get your writers involved in the conversation. People chill out a lot when they know they’re being listened to by the writer (and they act out a lot more when they think no one’s listening). I know, writers can find this an onerous addition to their workload, and have probably already decided that they hate their comments. Too bad. This is part of journalism’s evolution, and you’re either on the boat or you’re not.

    One great way to get writers on board is to give them the ability to moderate comments on their own stories. They can do this on their blogs, they should be able to do it on their stories, too. (With supervision by the Community Manager, naturally.)

  9. … But Don’t Feed the Trolls

    Members participating with good intentions are generally pleased when the authority figures are participating. Unfortunately, that can also bring out the trolls – bad users who are playing a game called “suck up as much of your time as possible.”

    School your writers in the ways of online community. If someone is trying to get a rise out of you, don’t fight back, no matter how tempting. A good Community Manager can help train writers on how, and when, to join the fray.

  10. Give Up Control

    Newsrooms are top-down places, but the internet is not. Get used to the fact that people online won’t do things just because you told them to. In fact, the only thing you can absolutely count on is that something will happen that you didn’t expect. When it does, you’ll be defined by what you do next. Be ready to be surprised.

As you can see, embracing community tools on your site takes work. If you just turn on comments with open-ended tools and no oversight, of course the result won’t be pretty. That’s because you haven’t done the job of an editor – to lead by example, direct the conversation, and sculpt the results.

The real reason comments on newspaper sites suck isn’t that internet commenters suck, it’s that the editors aren’t doing their jobs. If more newspapers implemented these 10 things, I guarantee the quality of their comments would go up. And this is just the basic stuff, mostly unchanged since I wrote Design for Community seven years ago.

Imagine what we could do if we could get past the easy stuff.


PREVIOUSLY: This is Not a Comment

UPDATE: Just One Question for Jason Schultz: Does moderating comments on a website make the website owner more liable?

What is the Ideal Checkout Login?


What is the Ideal Checkout Login?

When planning a conversion rate improvement project, its a great idea to start with tweaking your checkout process.

But where in the checkout process do you start? Many are tempted to begin by shortening the number of steps in the checkout process. While this has improved conversion for many sites and is a good idea to test, it shouldnt be the first thing you optimize.

More important is getting the customer into the checkout process to begin with. Typically, there is a gate between Proceed to Checkout and Shipping/Billing information in the form of a login/guest checkout screen. This is most likely presented in 1 of 5 formats:

1. Login screen (no guest checkout option)

2. Login screen with guest checkout option

3. Login screen with guest checkout plus account creation option

4. Default to guest checkout flow, with option to sign into account

5. The Amazon (single email field with radio buttons for password)

The problem with all of the above except for the Amazon approach is customers must decide between input sections. This requires the customer to read and make a choice. Sadly, you can never count on everyone to read instructions, meaning you will always have a segment of customers who will trip over login screens like this. Heres an example from Jakob Nielsens usability blog:

In our study, a person who was already a registered user on Kayak had great trouble logging into the site on this screen:

How can that be? The distinction between sign up and sign in is clearly marked! Sure, if you read the blah-blah. But users dont. Their eyes and mouse go straight to the field where they can type.

Thats what people want to do: they want to get things done. They dont want to read.

The golden rule of usability is dont make me think, and the more options you give the more you make the user think. Using Amazons approach, Sears streamlines its login screen, with even less instructions text to read than Amazon:

Only when the user selects No does the text You will have a chance to register after checking out appear, saving existing account holders from extra reading and thinking.

A second runner-up in the simplicity category is the default guest checkout (skips the login in screen) with optional account login box. We tested this approach against our original 3 option process with the Vancouver 2010 Olympic Store and achieved a 257% lift in checkout completion.

I challenge you to test your existing login screen against the Amazon and the default guest checkout first, then look at other points of friction (like presence of a progress indicator) to further improve results after youve determined the best login approach for your site.

Accordion Forms

Testing Accordion Forms
Testing Accordian Forms

Three point five pages. Its my usual answer when someone asks me how long should my web form be? And believe it or not, many people ask. It may be the most common web form design question I get asked. So Im not exactly ecstatic that my tongueincheek answer mostly draws blank stares and very few laughs. You see, it turns out this is a topic that many people take quite seriously.

And take it seriously they should. Web forms are the linchpins of most online businesses and applications. Whether they are responsible for checkout on e-commerce sites, communication on social applications, or any kind of data entry on the web, forms allow people to complete important tasks. And web form design details can have a big influence on how successful, efficient, and happy people feel about the process. Especially details like form length.

I often get asked how long web forms should be because web designers and developers want to know:

  • Do onepage forms convert better than multiple page forms?
  • Will breaking up a long form across several pages simplify the process of completing a form?
  • What about rich interactions that manage multiple sections of a form on a single page?

Let me level with you up front. There is no magic number like three point five that will work in all cases. Your audience, their intentions, your systems, your policies, and more, will ultimately define the solution thats right for your product. But you can arm yourself with informed design solutions so that when the time comes, you can pick the right answer for your specific situation.

One such solution is accordion forms.

Accordion Forms

Accordion forms use dynamic interactions on a single web page to hide and reveal sections of related questions as people go through the process of completing a form. This allows people to focus their attention on one set of questions at a time without requiring them to navigate between different web pages.

Accordion forms have the potential to be really effective because, in general, hiding irrelevant form controls from people until they need them results in forms that are easy on the eyes (i.e., less distracting) and they can be completed quickly. Couple that with the time saved on page reloads and getting people orientated on new pages, and accordion forms begin to look like an interesting solution for helping people complete lengthy forms on a single page.

To illustrate the difference between a typical multipage form and an accordion form, we need look no further than the Apple stores recent checkout redesign. Apples previous checkout process used several different web pages to capture customer, shipping, payment, and account information.

Screenshots of the old Apple Checkout Process

Fig 1: The Apple stores previous checkout process spanned several web pages.

Apples redesigned checkout process dropped multiple pages in favor of a single accordion form that reveals one set of questions to customers at a time. As you can see in the video below, people are first asked about their shipping information. Once complete, this section slides up revealing questions about payment options. Once payment details are entered, the account information questions are revealed, and so on, until the form is complete.

Video 1: This sped up video of the Apple store checkout process illustrates an accordion form in action. So, Apples using accordion forms; should you?

Putting it to the test

To better understand how accordion forms influence conversion and usability, I worked with Etre, a Londonbased usability firm, to test 24 average users (ranging in age from 19 to 48) on four variations of a typical e-commerce checkout form. We asked people to complete a purchase by entering information about themselves, their shipping address, and their payment information. This is essentially the same information that Apples checkout process requires, without the additional step of creating an account.

Screenshot of Four Typical Ecommerce Checkout Forms

Fig 2: The four variations of a typical e-commerce checkout form we tested: a) multiple pages b) single page c) accordion with headers d) accordion with buttons.

The control version of our four forms separated the questions people had to answer across multiple pages. We also tested a version with all the questions on a single web page, and two versions of an accordion form that revealed additional sets of questions as people completed the prior set. One of our accordion forms required people to click on section headers to expand each set of questions on the form. The other made use of a continue button within each section that, when clicked, automatically expanded the next section and collapsed the previous one.

For each variation, we measured success rates, error rates, completion times, satisfaction ratings, and standard eyetracking metrics. We presented each form randomly to minimize familiarity bias.

But enough setupwhat did we learn?

Constant conversion

Let me give you the good news up front. Each of the forms we tested had a 100% pass rate. Thats not all that surprising when you consider the task we gave our participants was familiar to them since each participant we tested had been using the internet regularly for more than three years.

This tells us that simply porting the same questions you have on one long web page or across several web pages into an accordion form isnt likely to increase conversionat least, not for common tasks like e-commerce checkout.

The flip side of this, of course, is that accordion forms arent likely to negatively influence conversion either. When you consider that most innovative user interface solutions typically have an adverse influence on conversion rates until people get used to them, this is an important finding. Accordion forms dont seem to have this effect. They might not move the needle dramatically upward but they arent likely to tank your conversion rates either.

At this point, I should pause and remind you that we tested 24 people on our generic e-commerce forms. To really test conversion, nothing beats testing your actual audience on your actual site.

Need for speed

Where we did see a noticeable difference between accordion forms and the single page and multiple page forms we tested was in completion speed. People were fastest with the buttonbased accordion form we tested (option D in Figure 2 above). Surprisingly, about ten seconds faster than even the single page form (option B).

While completion speed may not be paramount for e-commerce, there are plenty of other situations online where quickness matters. Online auctions or exchanges come to mind, where the failure to act quickly could result in a lost opportunity. Accordion forms just might fit the bill in these situations.

Accordion around

The primary difference between the two accordion forms we tested was in how people made their way between the different sections. Version C (Figure 3) required people to click on the header of each new section to expand the content within it. In this layout, a final action to complete the entire form was always visible.

Screenshot of an Accordian Form

Figure 3: In this accordion form (version C), people needed to select each header to reveal the next set of questions.

Version D (Figure 4) instead required people to click on a primary action button within each section of the form. Selecting this action would collapse the currently open set of questions and expand the next set (which also contained a primary action within it that moved people forward).

Screenshot of an Accordian Form

Figure 4: In this accordion form (version D), people needed to select a primary action button to reveal the next set of questions.

In our testing, very few people thought to interact with the headers as they made their way through our forms. The vast majority of our participants quickly gravitated to the primary action buttons and clicked them to move on. Even our explicit labeling of the forms final action (in Figure 3) button as Make payment and complete order didnt help. People selected this action to move forward even though they had not yet entered any payment information. This, of course, resulted in some unintended errors.

The lesson here is that even when innovating, you should try to embrace conventions. Though accordion forms may be new to many people, clicking on a visually prominent button to complete a web form is not. The form design in Figure 4 builds on this existing behavior by expanding the next section when someone clicks on the primary action on the form.

The form design in Figure 3 breaks with this behavior by forcing people to interact with nonstandard form controls like section headers. This requirement made accordion forms feel foreign and complex to people instead of familiar and easy.

Being accessible

While our user testing didnt specifically target people with disabilities, access for everyone is an important design consideration. Single and multiple page forms, when properly implemented, are familiar models that keyboardonly and assistive technology users can usually handle. Accordion forms, however, may be different story. Luckily support for WAI-ARIA, which allows complex controls such as accordions to be properly expressed, is growing quickly (it is being baked into HTML5) and may ultimately make accordion forms more accessible.

More to learn

Our quick test seems to indicate that accordion forms may have an interesting future online. They can bring multiple page forms together into a single set of dynamic interactions thatwhen designed welldont negatively influence conversion rates but do reduce time to completion. Of course, theres a lot more to explore about where, when, and how we should use accordion forms. But consider this design solution a quiver in your bowready to use when the right situation arises.


Id like to thank Etre for all their work scripting, running, and reporting on this study.

10 Worst Things That Can Happen In Your Checkout


10 Worst Things That Can Happen In Your Checkout

1. No persistent cookie

As I wrote previously, persistent cookies that remember whats in a customers cart are important for customers that take more than one visit to complete a purchase. Without persisted carts, those who abandon the checkout process with the intention of returning later must start all over again.

2. Required registration / account sign in

According to Forrester Research, 23% of shoppers will jump ship when asked to register with your site. If thats not bad enough, even your registered users lose heart when they cant remember login information. Forgot password? links are not the answer. Analyzing the database of a very large online retailer, Jared Spool found only 25% ever tried to finish the checkout after requesting password information! Safest bet is to offer guest checkout with the option to save customer information at the end.

3. Trapping customers in your checkout

Many conversion gurus have preached the virtues of removing site navigation during checkout to minimize distractions and keep customers focused on the big shiny calls to action in the checkout process. But this makes it difficult for customers who want to go back and add items to their order (which increases average order value). For example, I recently made a cross-border purchase. The shipping charges to Canada were so high, I wanted to add to my order to justify the shipping cost, but was unable to do so without opening a new tab and visiting the sites home page again.

While Im all for removing clutter, and game for every site testing with and without navigation, I dont think its necessary to remove all means for the customer to back out of checkout. At the very least, hyperlink the site logo to the home page.

4. Surprise shipping costs

Speaking of high shipping charges, this is the number one reason for cart abandonment according to Forrester Research. While how much you must charge for shipping is a business issue, not a usability issue, showing shipping charges too late in the process was also a customer peeve. 59% of consumers expect you to offer a pre-checkout tax and shipping calculator on the shopping cart page.

5. Asking for too much information

12% of consumers abandon checkouts that ask for too much information (again, thank you Forrester). A classic example of asking too much is Foot Locker, which makes favorite sports and shoe size required fields. Huh?

Take a look at your checkout forms and ask yourself which fields are untouchable and which fields you could sacrifice for a split test.

PS stay away from captchas too. Only 1% of checkouts use them, theyre difficult for customers to decipher, and dont put it past spammers to find workarounds for them.

6. Poor error handling

Forms that contain input errors are culprits to cart abandonment because customers need to understand what to correct. You must make the error message very easy to spot, very easy to understand and very clear why the error happened. Many checkouts fall flat in this area. Here are some examples from top e-tailers:

American Apparel packs so much above the fold that you cant see where the error occurred when the page refreshes. If youre lucky, you might notice the yellow strip containing the error message.

Chapters Indigo highlights all the errors at the top, which forces the customer to figure out which fields are associated with each error.

Worse, Skull Skates uses a dialog box that must be closed before you can continue. You cant bank on customers to recall each error if there are more than one.

Williams Sonoma and American Eagle Outfitters do a better job, respectively.

Beware of browsers that can mess up the display of errors, making them near impossible to read.

Industry jargon like AVS mismatch should also be explained. Most consumers dont know that the address verification service requires an exact billing address match to their credit card statement, including middle initial if applicable. This is especially problematic when the Shipping Address can be copied to the Billing Address section with one click and the online retailer uses AVS.

And dont forget login screens, a common place for errors.

7. Not explaining the CVV

Like AVS, CVV (or CVC) is an ambiguous acronym. Make sure you explain it.

8. Not optimizing for page load speed in checkout

Do you focus your performance optimization efforts only on your home page or pages? Dont forget to optimize your checkout, both for page load speed and your payment gateway performance.

9. Pre-checking fields

While it wont impact conversion, it can erode trust post-checkout when customers unwittingly opt in to your marketing program and receive what is in their minds unsolicited email (aka spam).

10. No contact information

If a customer is experiencing a problem in your checkout, a prominent customer service contact number (or live chat link) may save the sale. So why do just over half of retailers (according to Jupiter Research) display one in checkout?

There are many other checkout nightmares that could happen. These are my top 10. I welcome your ideas in the comments.