Lessons We Learned from Our Biggest UX and Design Mistakes

Lessons We Learned from Our Biggest UX and Design Mistakes

We’ve finally hit the 500,000-user mark at Buffer, a product that helps you share on your social media networks more efficiently. About two years ago when we started on our path to building Buffer, we knew we’d be meeting obstacles and making mistakes along the way.

One of the main things we’ve kept in mind is that making mistakes is unavoidable and that if we choose to learn from them, they’ll be helpful in giving us good guidance on how to move forward more effectively.

And I believe that it’s partly because of these mistakes that we were able to get to where we are today.

The Experience That Shaped How We Build Our Product

Before I discuss the biggest lessons we learned from some of our UX and design mistakes, I want to talk about one of our primary product development principles:

"Validate first, code later"

Let me tell you how this came about.

When the idea of Buffer for building a smarter way to post to Twitter and other social networks came about, our founder Joel Gascoigne immediately started coding instead of first validating his idea (which is how he used to typically approach things).

A few minutes into his coding session, he realized that that wasn’t the way to go.

So he tried something else.

What he did was to put up a landing page as if his product already existed, even though he hadn’t finished everything yet.

He then went ahead and spread the link of his site across Twitter.

People whose interest got piqued would click on a button on the landing page to sign up for an account. They were then presented with another web page that said Buffer wasn’t finished yet and asked the interested user to leave his/her email address so that he/she can be get an email as soon as Buffer launched.

This strategy worked out quite well and we got our first paying customer within 7 weeks of coming up with the idea.

Three key product development principles were created from this experience:

  1. Keep the first version of a feature or product as minimal as possible.
  2. Be prepared for a long journey with lots of course-redirection.
  3. Turn any idea for a feature into a hypothesis that first needs validation from the user.

Now that you know a little bit about our product development principles, the following lessons that we learned will make a bit more sense.

Lesson 1: User Flows Should Focus on Retention, Not Revenue

A key user experience (UX) lesson that we discovered early on was that we needed to concentrate on keeping our customers rather than generating revenue.

This is how we learned this lesson. Our initial landing page’s user flow was as follows:

  1. When you first see our landing page and click on the sign-up button, we would show you a "subscription plans and pricing" page.
  2. From there, you would be required to choose either the "Free" plan or one of our paid plans.
  3. After you’ve picked the right plan for you, you would be able to sign up by providing your details (e.g., your name, email address, and so forth).

With this user flow, the rate of people choosing one of the paid plans upon signing up is high, thus allowing us to generate revenue early on.

However, we quickly learned that users who picked to pay for the product before getting the chance to use it sent our churn rate through the roof.

Why did this happen?

We found out that people would pay, but then they wouldn’t even start to use Buffer, and eventually they’d cancel their subscription.

So we changed our sign-up user flow and philosophy of acquiring users. We decided to say to ourselves, "Let’s get the person to start using the product first so that they can experience first-hand the value of our product, which will hopefully encourage them to upgrade to one of the paid subscription plans."

That worked out a lot better for us.

As a result, the user flow through our website is incredibly more seamless now:

As a result of this user flow redesign, two important things happened:

  1. More people signed up because we didn’t dilute the funnel with a pricing page as an interstitial.
  2. More people upgraded over time because they started to actually use the product, find value in it, and stick with us.

Since then, we’ve made many more product changes that give out more free features, and our core product features are still free. Only after we’re sure you’ve really seen lots of value from using Buffer that we would encourage you to upgrade to a paid plan.

Making user flows that focus on user-retention was an important discovery for us.

Lesson 2: Social Sign-in is Better Than Email Sign-in

After dozens of failed A/B tests to improve the conversion rates of our landing page, we came up with the idea to enable social sign-in (i.e., signing up for a Buffer account and logging in using your Twitter, Facebook or LinkedIn account).

After all, Buffer is for posting on Twitter, Facebook and LinkedIn so it only seemed logical that our potential users would be able to quickly and conveniently sign up using one of their existing accounts.

For comparative purposes, below is our initial landing page that would let you sign up with your email and a password:

Our change from the email sign-in to social sign-in is one of our biggest growth hacks, allowing us to eventually increase sign-ups by 50%. And around the time of this sign-in switch, we went from 500 daily sign-ups to 800 daily sign-ups practically overnight.

Lesson 3: Test All of Your Assumptions Early

Let me tell you a story about how we failed miserably with a new browser extension redesign idea.

One of the essential parts to Buffer is our browser extension for Chrome, Firefox and all other major browsers. We know that if you’re a user and use the extension, you’ll have the best experience with the product. Sharing from any web page and easily adding everything to your "buffer" to share on Twitter, Facebook and LinkedIn emerged as the most powerful use-case.

So it was only natural for us to focus on this element of the product and try our very best to improve it, as we clearly had some ideas on how we could make it better.

But when we got going with the production of a better browser extension, it seemed like we fell back into our old habits and made the same mistakes we knew we should be avoiding.

Here’s the sequence of how we (incorrectly) approached building our new extension:

  1. We identified some issues with our existing browser extension and then we brainstormed on what we wanted to change.
  2. We spent a lot of time and resources to design and develop a fully-functional first version.
  3. We tested the first version late in the process and realized people were getting extremely confused when using it.
  4. We canned the new idea and never launched it.

This was the new layout of the browser extension that we never pushed to production:

Getting into the habit of testing every single assumption you have as early as possible is something we now have seared deeply into our brains since that failure.

To make things right, this is now how we approach building new products and features:

  1. Identify issues with the existing product and brainstorm what you might want to change.
  2. Talk to users about whether they actually have the same issues.
  3. After validating it in a discussion with our users, quickly build wireframes or a simple prototype that doesn’t take longer than 1-2 days to produce.
  4. Talk to users again and see how they interact with your prototype.
  5. Iterate further to a somewhat usable feature/product that you can show to more users.
  6. Track engagement/growth/revenue metrics and talk to users again about their experience.

Lesson 4: Be Clear with User Interface Labels

The last lesson that I want to share with you is something that I’ve been thinking about for quite a while: It’s the issue of being clear with your labels, buttons, help text, etc. versus being clever with them.

Des Traynor from Intercom defined this problem incredibly well in an article:

Situation: You’ve built a great feature that solves a real problem that you know your users have. They’re not using it though. Usually, it’s because they haven’t seen it, or they saw it and didn’t know what it did.

That situation is something we continue to face.

The key example I wanted to describe here is the following: Inside Buffer, you can connect multiple social accounts together so you can post to them all from one location, e.g., you sign up or login with your Twitter account but you also want to connect your Facebook and LinkedIn account to Buffer to make posting from one place easier.

What it used to look like was this:

We thought to ourselves that having a clever "plus" (+) sign icon will indicate to people that that’s the way to add your other social network accounts.

And yet, we repeatedly received emails from our users asking us if there was a way to connect their Facebook or LinkedIn account in Buffer.

So what we did was to change various aspects of the "connect" button by making it a placeholder icon, using a bigger plus sign, and many other design tweaks.

What was the eventual design solution?

The solution was simple: Using the text "connect more accounts" in plain, written words, which is a lot clearer and more effective than having (what we thought was) a cleverly devised icon to represent this UI task.

We learned that picking the clear solution over the clever solution — even though the former might not be as pretty or as unique or as cool — is always what’s better for the user.

Conclusion: Everything Is a Hypothesis That Needs to be Validated

We believe that our company and our app is still in its early stages, so a lot of the design processes and methods we have are still very experimental.

The concept of not being attached to a single idea and treating every design or feature iteration as a hypothesis that needs validation is the overall biggest thing we’ve learned.

We plan to build many more features that people will hopefully love. And we also expect to build many more that we’ll have to throw away.

We think this way of thinking will give us the biggest potential for building something our users truly want.

Related Content

About the Authors

Leo Widrich is co-founder of Buffer, a better way to post to Twitter, Facebook and LinkedIn. He also blogs about insights on lifehacks, business and productivity on the Buffer blog. You can say "hello" to him on Twitter @leowid (he is a super nice guy).

Tom Moor is co-founder at Buffer, a smarter way to share on Social Media. At Buffer Tom leads design and UX, constantly trying to make the interface easier to use, whilst focusing on the little details. Follow him @tommoor on Twitter for more great chats on startups and design.

This was published on Feb 4, 2013


David Barker Feb 04 2013

Interesting article. Can’t emphasise enough this idea of: validating first before doing the work. Saves so much wasted effort. Do as little as possible to get the required information. Once you have analysed the information, then go ahead and build.

Sumit Feb 04 2013

Thank you for sharing.

Zsolt Pentz Feb 04 2013

Great article! I love how the experience pushed you to a new way of thinking and how you could realize and then also solve the design problems. Nice to share these thoughts!

Robert Reiz Feb 04 2013

Many Thanks for writing this. We did many of the same mistakes, too. I wished I have read this 1 year ago :-)

Amazing article! Thanks for giving a glimpse into the struggles and principles that are usually learned the hard way. Keep up the good work.

Ryan Bollenbach Feb 04 2013

Really useful article here guys.

The 2 golden nuggets I found this article shined on:

1. So we changed our sign-up user flow and philosophy of acquiring users. We decided to say to ourselves, “Let’s get the person to start using the product first so that they can experience first-hand the value of our product, which will hopefully encourage them to upgrade to one of the paid subscription plans.”

2. Our change from the email sign-in to social sign-in is one of our biggest growth hacks, allowing us to eventually increase sign-ups by 50%. And around the time of this sign-in switch, we went from 500 daily sign-ups to 800 daily sign-ups practically overnight.

It would have been awesome to read about more statistics about things that were the bomb vs things that bombed. If you guys any other insight on this, please let me know! – Instead of focusing on conversion pages, it’d be neat to hear about application works changes and other contextual insights.

Thanks for sharing!

@Iconic88 Feb 04 2013

Awesome share Leo and Tom! I know many people if they utilise your learnings, they’d go along way to improving their own online services.

Keep up the buffering!

Anthony Burokas Feb 04 2013

Great post.
Validate first is a very good lesson to share. Very often we think we know the right way, but when it comes to a product that others use, the question is instead, does it work for them. Not us.

There’s a lot of #4 in #3- and a good bit of it in the new interface. If I could tweak one thing, I’d take my picture off the icons and just have the social media logos. And then instead of grey or not grey (again, clever, versus clear) I’d circle them in green or have a red X or \ through them. Because with the sun glare on some screens, it’s hard to tell whether something is on or off.

And I appreciate that you don’t post a lot of business stuff, but the stuff you do post, or link to, is quite good. A very good signal to noise ratio.

And never, ever tetweet someone tweeting you. God, I hate when companies do that. Some break their arm patting themselves on the back.

Russell Rothstein Feb 04 2013

Hi Leo. Great post as usual. I love your product (and happily pay for it). Question – what tool/site/method do you use for usability testing?

haider Feb 05 2013

Very interesting article.

alyssawilliams494216 Feb 05 2013

This is really a very nice article to describe how to learn from your past mistakes in the designing. Thanks for sharing this.

shofik Feb 05 2013

Our change from the email sign-in to social sign-in is one of our biggest growth hacks and many type ……

Cubicle Ninjas Feb 06 2013

Excellent article! I will be implementing some of these ideas into my own project.

Ryan Bradley Feb 06 2013

I definitely like the new layout of Buffer. When people try it first, they will convert more often. That’s why supermarkets put out samples, if you eat, you buy.

Amir Talai Feb 06 2013

great piece, thanks so much.

I think you are using wrong terminology with social-signin.
It’s not about what TYPE of signing you offer, it’s about the SPEED and TRUST of the signin. IMHO you could offer Google login as well – is that email or social (G+)? Doesn’t matter. Just make sure it’s FAAAST and easy and that security/authentication is handled at the ‘trusted’ mega-corporation of user’s choice. Email signin is still very valuable for your paranoid customers who do not want to share their every movement with said mega-corps.

Coalesceideas Feb 09 2013

social login is the best now a days, but sometimes it will break your website, like what happened when Facebook encounter issue

I found this article very interesting, thanks for sharing the lessons!

Mike Storzieri Feb 10 2013

Loved the story. We test everything we can, although our clients push for quicker results at times, you cannot rush into programming like you pointed out in the beginning of this story.

Tamara Bredemus Feb 11 2013

Thanks for taking the time to share what you learned. Clearly laid out and great tips.

Hrishi Mittal Feb 11 2013

Your first lesson is quite counter-intuitive especially given how Joel originally validated the idea. Thanks a lot for sharing your lessons – very useful stuff!

ranjit Feb 13 2013

Nice article… Thanks

Mitch Feb 13 2013

I have been exploring for a little bit for any high quality articles or weblog posts in this kind of house . Exploring in Yahoo I finally stumbled upon this website. Reading this information So i am satisfied to convey that I’ve an incredibly good uncanny feeling I found out just what I needed. I most indisputably will make certain to do not overlook this web site and give it a glance on a constant basis.

Why isn’t six revisions a responsive website?

Houdas Feb 14 2013

Great article, great story. A lot of inside-tips, which is always valuable.

Frans Ekman Feb 14 2013

I find it interesting that the churn rate was high for those that payed. I would have guessed that those who paid would in a way have been “pot committed” and would have tried to get whatever they can out of the service. But apparently you can never know before you test.

Jared Feb 16 2013

I’ve always said that UX is a lot of trial and error to find the right match.

Christopher Wakare Feb 17 2013

Thanks. Test ALL assumptions, The Pricing Page views were quite insightful.

Ganar Dinero Feb 23 2013

Amazing article! thanks so much.

Debbie Feb 24 2013

“Everything Is a Hypothesis That Needs to be Validated”. – Most businesses don’t have the time and resources to sit on this stage and yet everything else is rather useless if one doesn’t do the ground work.

Thank you for sharing both sides of the process.

I love the share of lessons learned. It is so obvious that a website that converts people to pay without really wanting the product is going to have a high churn – obvious with hind sight. This is about ‘qualifying’ the users first, something that people know they have to do with PPC but don’t think so clearly about with organic visitors.

Obat Maag Feb 28 2013

Thanks your article.

Roye Okupe Feb 28 2013

Hmm, I have been looking for a service to handle my tweets, looked at HootSuite for a bit and was not really comfortable with it, after reading this I might try Buffer.

Great post!

Devon Designs Mar 01 2013

My take-away from your post is this: don’t focus on short-run gains, focus on client’s satisfaction.
It may sound mundane but it’s actually very hard to act upon.

haider Mar 01 2013

Very informative and useful article.

Dmitri Jun 16 2013

Great insight!

Few comments:

1. Validation is important but confronting user with (tedious) sign-up to see empty content can also repel the user. It is frustrating how many sites demand up-front sing-up without showing anything but promises. Why not delay the sign-up until the user actually see the value?

2. Yes, facebook sign-up is friendlier than Email/Password.
The former needs a single button click, whereas the latter forces you to type in your complete email address (without misprints), then to come up with a clever password that you shouldn’t write anywhere, then to validate your email and then possibly to sign in again. So how surprising is this difference?
Apparently it still is, otherwise how can one explain that so many sites still go the old way?

Dmitri Jun 16 2013

Some more feedback:

Maybe the request to post on my friends facebook wall on my behalf should be delayed? So many apps are now asking for it, without telling what they actually post there, which I find intrusive.

This comment section is closed. Please contact us if you have important new information about this post.