May we suggest...


The 4 Key Things That Went Wrong When We Launched A New Product and How We Learned From Them

As a social media startup, Buffer is constantly learning about what it takes to build great products.

Whether it’s giving a unique role to our Product Creators or evolving the structure of our teams, we try to treat everything as an experiment with opportunities to continually learn and improve.

One of our more recent sources of learning has been the launch of Pablo, our social media image creation app. Building Pablo has been a unique experiment that equipped us with some great product lessons.

In fact, some of the elements of launching Pablo that we initially viewed as mistakes or challenges have turned out to be great opportunities in disguise.

Here’s a look at 4 things that “went wrong” in launching Pablo that later turned out to have been exactly what we needed.

1. We launched way earlier than we planned

The Buffer team started working on Pablo in early 2015.

A task force was formed and quickly built a minimum viable product. The intention was to test it and steadily improve before officially launching Pablo—but that never happened.

Before we knew it (and definitely before we were prepared!) Pablo was added to Product Hunt and the news was out!


Brian Lovin, a Product Creator at Buffer at the time, describes what happened on the Daily Hunt podcast:

“The funny thing is, we actually didn’t really anticipate that we’d get on Product Hunt!…We wanted to learn a little bit more from a few people and interviewing actual customers, but of course someone submitted it to Product Hunt and it took off, so we’re kinda just rolling with it now.”

To some people, this might sound like a really big problem. The product was still in its infancy with a lot of bugs and we didn’t even know if we had product/market fit.

However, releasing it “too early” turned out to be a huge win, for a few reasons:

  • The Product Hunt community is full of incredible people who were willing to identify what was awesome about the product and cheer us on.
  • We were grateful for early-adopters who were willing to try Pablo, find bugs and let us know where we might need to improve.

Here are some examples of how this played out on the Pablo Product Hunt page (click to enlarge):

comment 1 comment 2

Launching on Product Hunt allowed us to accelerate our early product development.

By having Pablo out in the real world, we learned rapidly, shipped bug fixes at lightning speed and built a mini-community of Pablo enthusiasts.

And the benefits of this “mistake” weren’t only limited to Pablo. On the day Pablo launched on Product Hunt:

  • The Buffer website exceeded 100,000 visits in one day for the first time ever.
  • Buffer had its highest number of sign-ups and trials in a single day.

As Reid Hoffman famously said, “If you are not embarrassed by the first version of your product, you’ve launched too late.”

2. Our numbers flatlined after the buzz wore off

After launching on Product Hunt, a horde of people flooded into Pablo. Here’s what the user numbers looked like before and on launch day—a nice spike!

chart spike

But that graph only tells part of the story. Here’s what user numbers looked like just a few days later.

chart drop

Uh oh—Pablo saw a significant dip in usage.

But this big drop turned out to be another great learning opportunity.

At the time, Pablo was a minimum viable product with a limited use case in a field with lots of other options.

Attracting lots of users was positive because it signaled that people were interested in what the product might be able to accomplish.

But as the buzz tapered off, it served as a great reminder that there was still a lot of work to do.

Sean Ellis describes this predicament well:

“Even the greatest marketers can’t sustain growth on a weak foundation. Eventually, their growth curves crater.

“So what is required for a strong foundation? A must-have product.

Despite the drop-off in users, the initial spike in numbers gave us a large base of users to learn from. Their input played a key part in helping Pablo establish product/market fit and laid the groundwork for Pablo v2.

3. We didn’t know who owned the product

In the weeks after Pablo’s initial launch, there was an undeniable sense of excitement and energy for the project.

Here’s a note Brian wrote about his initial plan to move Pablo forward:

Brian message

But here’s where a third challenge came in: The Buffer team at this time was experimenting with self-management and working within units we called task forces.

As Courtney describes it:

“(Task forces) were fluid, dynamic teams formed for a specific purpose (for instance, developing Buffer for Video) that could be proposed or joined by anyone at Buffer, regardless of role.”

The task force approach had many advantages, such as allowing people to work on projects that they were most interested in. However, this setup also had significant challenges:

  • Task forces were project-based, so once Pablo had launched (i.e. the project was completed) it was a little difficult to determine who was responsible for its continued development.
  • Research would often be prevalent early on within task forces. Once product development was underway, research tended to slow down because it wasn’t an integral part of the task force structure. This setup slowed down validated learning within task forces.

This tension played a part in Pablo’s momentary stagnation, illustrated in another Discourse post from Brian:

task forces limitationsExamining this situation and others like it eventually led to us moving away from task forces (and a few other elements of self-management) at Buffer.

Our new team structure includes areas and circles, a setup that solves many of the challenges we faced within the task force structure.

areas and circles

So although we might have lost a bit of time developing Pablo further, we ended up unearthing and fixing a central tension that has since allowed us all to work faster, happier and more productively.

4. Our users told us they wouldn’t use it

Most recently, we spoke to dozens of users about their experience with Pablo and other image creation apps.

In those conversations, we started to notice a few themes emerging that gave us pause.

Here are a few snippets from those interviews:

“With our clients, sharing quotes is not the biggie — that’s why we hardly use Pablo.”

“Pablo doesn’t seem to be that beneficial because I’m not looking to add text to my photos.”

“Pablo doesn’t help the company much. It’s very personal tool, not a very great biz/enterprise solution.”

“I don’t like to see the quotes, so please don’t want to start off with quotes. More times than not they’re a nuisance.”

“I really only use Pablo to create quote images. I haven’t need to do that for a while.”

This felt like a big problem: Users were basically telling us that Pablo wasn’t all that useful for them.

But upon closer inspection, we noticed that what they were really telling us was about perceptions of the product.

They were telling us that they didn’t see Pablo as a ‘social media image’ creation app. They saw it as just a ‘quote image’ creation app.

That might not sound like a big difference, but we think it is! In the eyes of our users, Pablo was a great tool for adding text on top of an image. Users created many more types of social media images, so Pablo catered to one of their needs.

This limited use case helped our team realize that we needed to expand Pablo’s functionality if we really wanted to make social image creation easier and more intuitive.

As a result, we’ve been working on adding more images and creating new features like a library, image repositioning and more text options (all coming within the next month!).

We would have never discovered this invaluable information without talking to our users.

In this counterintuitive example, I take a lot of insight from Alex Turnbull, who talked with over 500 of his customers when he wanted to figure out how to improve Groove HQ.

In his own words:

“The ultimate ‘win’ from customer development is deep insights into how our customers think, feel and use our app. That insight is absolutely critical to the growth of any business, and it’s the biggest reason I took this project on. It had an immediate impact on how we approach our product roadmap and day-to-day decisions.”

What product lessons have you learned?

We still have much to learn, and you likely have some great lessons to add here.

What are some of your biggest takeaways from launching a new product? Have you ever had an element of a launch that you initially thought was a big mistake turn out to be a great opportunity?

I’d love to learn from you, so please share it all in the comments section!

  • Jevon Millan

    Would be interested to know what marketing surveys went before the launch of the product. Sounds simply like key buttons weren’t identified ahead of time.

    • Great one, Jevon! If my memory is correct, I believe we launched Pablo before we had really developed our customer research team at all, so we were definitely making lots of assumptions. Feels like we’ve evolved a lot since then!

    • Roy Olende

      Hey Jason, great question (and so thankful that Courtney jumped in for me while I was on paternity leave).

      We did some initial customer development (interviews) internally and with some Buffer users. As Courtney mentioned, this was in the early days of our research team. :)

  • salaamthebody

    This is an enlightening post, not just for behind the scenes on Pablo, but also because Pablo can essentially be substituted with any other product.

    • Totally agree; lots of universal lessons here! Thanks for checking this one out. :)

  • Susi Castle

    It was interesting reading about your internal transition away from task forces but what stands out the most from this post is how much self reflection you employ to learn lessons from every moment.

    “They were telling us that they didn’t see Pablo as a ‘social media
    image’ creation app. They saw it as just a ‘quote image’ creation app.”

    “Understanding the perceptions of the product” feels like a lesson that could be easily missed by someone only looking for a superficial lesson to be learned.

    I’ve been using Pablo a lot of the last couple of weeks so I’m looking forward to seeing those changes you mentioned being rolled out. Thanks Roy!

    I’m wondering how you go about interviewing users – I can imagine there are a lot of people who feel themselves too busy to reply (though I’m sure they want to be helpful). Do you survey people at random or do you find that you have a core group (like the Product Hunt community) who are always giving feedback?

    • Hey Susi, these are awesome questions! Roy went on family leave soon after he posted this so he’s not quite around to answer for a while. I wonder if this post might help out at all?

      • Susi Castle

        Hi Courtney, thanks for replying to me for Roy – what awesome news for him! Definitely one of the best excuses anyone can have for not replying :)

        That post gives a really interesting insight into development – thanks for sharing it. It’s also led onto some great other articles and now I feel like I’m in a TV Tropes loop, reading interesting article after interesting article. Help!

    • Roy Olende

      Hey Susi! Really grateful that the awesome Courtney jumped in here on my behalf. :)

      I hope that post was helpful. If you have any specific questions about how we reach out to users feel free to pop me an email: :D

      • Hi Roy, thanks for the reply and congratulations on your great news! Courtney’s reply was really helpful, so I’m good for now!

  • Michael Jenkins

    Nice to see that you took the time afterwards to reflect on what you did and didn’t do. Some worked out and others did not. I enjoyed reading about the process and feeling like I was there watching it in real time. I have not had much experience with Pablo but will look into it and look forward to seeing changes that will come.

    • Thanks, Michael! Hopping in for Roy as he went on family leave soon after he wrote this! Really appreciate the kind words and your willingness to check out Pablo; would love to hear how it goes for you!

  • What a great post! Thanks so much :)
    Any chance I’d be able to integrate Pablo in an app and search for created public social images by tag (to be reused in a different context)?
    There are plenty of stock photo websites and Pablo seems to aggregate a few, I’d be really interested in this aggregation functionality as an API :)

    • Roy Olende

      Apologies for my slow reply here — I just got back from a month-long paternity leave.

      Great idea! We’ve got no immediate plans to do that but as you can tell, we’re very open to these sorts of ideas! Cheers! :)

      • Congratulations on the new arrival! Please tell me somehow if you do start supporting it, thanks! :)

  • Kirrily Weatherstone

    Great reflections. I love that Buffer has decided to integrate an image product. I’ve used Canva quite a bit to create images to share on social media and in comparison to Canva I would have to agree with some of the customer sentiments shared above. I’ve only really found Pablo to be the better tools for picquotes. It has definite potential though as an image creation tool and I look forward to seeing how the Buffer team develops it further to meet customer needs ?

  • Shenny Solis

    For my line of work and how I use social media, pablo would be AHHHmazing, if it had the functionality of videos and text. I mean, almost like an editing tool for videos and photos. Nothing too crazy, it doesn’t have to turn into Final Cut Pro but say for instance I have a good shot of something and 2-3 pictures to go with it; I’d love to be able to quickly, simply and effectively make a 15 sec video with some text to go along with it. That would be outstanding! I saw the promo for Pablo, it does have a tremendous amount of potential. Continue to learn and reflect and I see Buffer being onto something great. For now, Pablo is just a tad bit under developed, however you’re lucky to have users that want to see you grow ;) Great job Buffer! Also when are you guys going to add Instagram?

    • Roy Olende

      Shenny, wow — awesome insights here! I can totally see how a simple video tool could make your life easier.

      Right now we’re working to make Pablo a really awesome tool, and perhaps sometime down the road we might be able to tackle your idea!

      And oh, one last one…we’re definitely hard at work trying to figure out how we could help users with Instagram. Hang tight!

  • Gareth Cotten

    Some insightful reflections here, Roy, nice one! I know you may well still away, but great piece to tee up your family leave with :-)

    Building on your lesson that Pablo was seen primarily as a ‘quote image’ creator, have you considered offering options for other common text-over-a-pic permutations? For example, having the option to use it to create memes – I currently used Memegenerator, which is functional, but uuuugly. Maybe another one where you’re able to layer over some SnapChat-like text over a pic (considering the rate that SnapChat uptake is growing)?

    Also, random observation, but now, when you search Google for ‘Pablo’, Pablo Escobar now pips Pablo Picasso as the first auto-suggestion. Just to show the power of current interest in the former compared to the eternal fame of the latter…

    • Roy Olende

      Hey Gareth, nice note on the Google search! :)

      Awesome insight. We’re definitely looking to expand the use cases and I foresee that you’ll soon be able to use it to create memes. :)

  • Would love Pablo to integrate with all the free stock photo sites.

    • Roy Olende

      Apologies for my slow reply here Joshua — I just got back from a month-long paternity leave.

      Yes, love this thought! We’re currently looking at broadening out library, so this falls in line with where we’re headed. Thanks!

  • Sylvia

    Roy, nice to meet you! Thanks for sharing about your process. I started using Pablo but ran into a few glitches, of course, a Buffer Happiness hero got in touch with me quickly but I think it couldn’t be resolved right away. I definitely like using it and had thought I would use it much more than I have, but that is probably just due to life and work getting so busy that I haven’t been personally posting as much :) Always great to see Buffer’s growing curve, and all the people and companies you’ve learned from – it gives me a lot of reflect on and integrate into my own thinking!

    • Roy Olende

      Hey Sylvia, thanks for these kind words! Glad to share the process. :)

80,000+ social media marketers trust Buffer

See all case studies