May we suggest...

Inside Buffer

Helping a Feature Find Its Fit: The Story of Buffer’s Content Suggestions

Important update: We’re retiring the Suggestions feature – beginning August 1, we’ll start turning off Suggestions in phases. Read more…

One of the coolest things about working at a true startup like Buffer is how many jobs each of us gets to do in the course of a day or a project.

I joined Buffer thinking that most of my time each day was going to be spent writing, and then I discovered that I could also be working in search optimization, paid ads, conversion optimization and lots more on any given day. Pretty awesome!

One of my favorite projects I’ve had the privilege of taking on so far has been Buffer’s content suggestions. I’d love to walk you through the evolution of this experiment and share with you where this feature could be going next.

Starting out: A small experiment with suggestions

suggestions notification

The idea of curating great articles, photos and quotes for Buffer users to add to their queues originally arose to help new Buffer users understand the app more quickly and easily.

We started out in early 2014 with 5 suggestions on weekdays—generally a mix of marketing, inspirational and lifehacking content with a great quote thrown in. Curating content was one of the first things I took on when I joined Buffer in February.

The process was beyond lean—just a spreadsheet we used to store our daily links, which I input into a CMS.

suggestions spreadsheet

As we curated more content and expanded into weekends as well, we quickly discovered that the feature was helping all sorts of customers, not just those new to Buffer. That felt great!

content suggestions love

Growing from experiment to feature

If five suggestions a day was good, could more suggestions be even better?

In May 2014, we expanded the suggestions feature to 25 new posts per day across five different content categories (decided from a quick qualitative survey): marketing, lifehacking, inspiration, business and startups, and design.

suggestions categories survey

As shares of these suggested posts grew to up to 20,000 shares a day, the privilege of hand-curating this daily bounty was mostly mine (with lots of backup from my amazing teammates).

content suggestions marketing

I had a blast reading widely from a huge variety of online sources and informative newsletters, all of which we have shared with you in various posts on the Buffer Social blog.

As our small experiment continued its journey into a feature, we began to hear from some of the content creators we featured in our suggestions and see blog posts pop up about their experiences.

Their enthusiasm was so great that a number of content producers began to ask how their pieces could be considered for inclusion, so we created a quick form (now disabled) to allow anyone to alert us to awesome content. Over the time that the form was active, more than 1,000 pieces were submitted. (And we looked at all of them!)

Content suggestions go mobile with Daily

By the time June rolled around, we focused on making content suggestions a mobile experience as well. We added content suggestions into the latest version of our mobile app, and then introduced a fun, Tinder-for-content concept with our app, Daily.

Daily

 

Daily ramped up even further the number of shares that content suggestions received, as it made sharing content as easy as a simple swipe or click.

Listening to customers’ feedback

Our content suggestions experiment was working well for some people, but the feature certainly was not perfect. We heard from many of our awesome customers and community members that suggestions would be more useful if they were more tailored to the user’s profiles and interests.

A Qualaroo survey we followed up with revealed the same ideas:

suggestions improvement ideas

Another challenge was beginning to present itself, too: With Buffer’s large customer base receiving the same 25 suggestions each day, the number of shares each post was getting sometimes became overwhelming to the content creators we intended to honor. Here’s a look at the Twitter stream of the author of one content suggestion on the day it was promoted:

sharing content suggestions

We were hearing a diverse number of experiences with content suggestions, and it was becoming quite interesting to think on where to take them next.

Finding product fit

Luckily, around the same time Buffer adopted a new advice-based process for making decisions, meaning that I (as the primary decision-maker) got to ask all my teammates (and even some folks outside the company) for their great advice on how to move forward.

I acted first on advice from Leo to check out Sean Ellis’ simple survey method for determining product fit. Here’s how Sean describes it on his blog:

“I ask existing users of a product how they would feel if they could no longer use the product. In my experience, achieving product/market fit requires at least 40% of users saying they would be “very disappointed” without your product. Admittedly this threshold is a bit arbitrary, but I defined it after comparing results across nearly 100 startups. Those that struggle for traction are always under 40%, while most that gain strong traction exceed 40%. Of course progressing beyond “early traction” requires that these users represent a large enough target market to build an interesting business.”

When we created a survey like Ellis suggests using Qualaroo and ran it on the site for a few hours, it was almost immediately apparent that we had not achieved product-market fit with content suggestions.

product market fit survey

Based on this clear message, and knowing the amount of time that was being devoted to content suggestions every day, I made the decision to close down content suggestions totally.

I created a phaseout plan and we even went so far as to turn suggestions off for many users before I changed my mind and gave them a reprieve. (Turns out I could use a bit of improvement on my decision making!)

Preparing for the next evolution

So. What to do with a time-consuming, lovingly hand-curated feature that most users either don’t know about or don’t feel very strongly about?

It seems as if the ideal solution would be to reimagine content suggestions altogether. And to do that, we needed a bit of extra time.

So the latest evolution of content suggestions is designed to set the stage for a larger scale revamp in the future while countering some of the existing challenges we’ve encountered.

Fantastic Buffer developer Niel worked with me to randomize our existing pool of suggestions so every user sees a different selection each day.

Here’s a look at some of his technical specs:

technical specs for random suggestions

Today we have a pool of 5,000-ish content suggestions, and each Buffer users sees a slightly different set of 25 each day. This has improved our feature in a few big ways:

  • It evenly distributes social media posts for each suggestion (no more Twitter floods!)
  • It allows posts that users might have missed to be seen again (previously, if you missed a day’s suggestions they were gone forever)
  • It drastically cuts down on time spent curating content suggestions

I still love curating great, exemplary content, and doing it without an arbitrary quota each day makes sure the quality of content suggestions stays as high as possible.

Seeking out customer conversations

While Niel was working on randomizing suggestions, Happiness Hero Patrik has been talking with a wide variety of customers—some of whom love suggestions, some of whom see lots of room for improvement.

We appreciate all their viewpoints so much, and they’ve led us to a much deeper understanding of how we can make this feature soar.

Here are the validated ideas we’ve uncovered so far, from our working Hackpad:

customer interview validations

To the next phase of content suggestions

With great insights from our users, I’m really excited to see where content suggestions will go in the future. Some ideas we have so far include:

  • Personalized suggestions based on who you follow on social media and/or the feeds you’ve chosen in your Buffer account
  • Personalized suggestions based on content categories you choose
  • Curated suggestions by experts in tons of different fields
  • Working with an existing content curation tool to add similar functionality into Buffer
  • Creating our own content curation algorithm

It’s been a very exciting journey so far with this Buffer feature, and I know the best is yet to come.

If you’ve ever worked on a challenge like this one or a similar feature, how did you approach it and iterate on it? I’d love to hear your insights. And if you’re a Buffer user, it would be great to hear any thoughts you have on content suggestions!

  • Morning Courtney! What an undertaking! Good luck with the project. From a user perspective, the curated suggestion list is a lifesaver when I’m strapped for time and scheduling posts for the week. I’d fall into the category of ‘disappointed’ if the feature was discontinued. Love the idea of being able to enhance/add to the curated suggestion list by pulling in top content from selected followers/Twitter lists/feeds in our buffer account.

    Would also love to learn how top content from the Bench blog can be considered for inclusion in Buffer’s suggested content. Starting Jan, we have some content in the pipeline that’ll be hugely helpful to readers in the lead up to tax time – good for the ‘Busines & Startup’ category. Thanks v. much for letting me know!

    http://blog.bench.co/

    • Hi Cameron, thanks so much for your insights here! It’s super helpful to know how and when you use content suggestions and how they could be made even more useful for you. I’ll definitely keep an eye on the Bench blog for some amazing content to share!

      • Courtney, thanks for replying so quickly. You’re too kind, and can’t wait to see the next iteration of the content suggestions feature.

        Also: thank you for considering the Bench blog! Amazed/delighted/excitedly nervous that we’re on your radar :D

        Have a good rest-of-the-week!

  • I am a big fan of Content Suggestions Courtney..Something i look forward to read and share everyday.. This helps me see & discover new blogs which would have been taxing otherwise..

    I would also love a feature of suggestions based on upvotes..Something which shows up like (25 friends in your network have shared this) based on the user followers/following

    Something which would help content discoverability and shareability right within Buffer

    • Hi Omkar! Wow, what an interesting idea to unearth cool content through your network of friends. I’d love to mull that one over. Thanks so much for the suggestion, and for checking out the post!

      • Wow, I would love to see the trending posts rated by number of shares, re-shares etc. that would be awesome as well. Maybe the first 5 will be the trending ones followed by the others. Seems like you are working on a killing feature where lot of people will be spending their time scheduling and reading the posts :)

  • Content suggestions are immensely helpful to me and I will be very disappointed when/if they disappear. Do you see the feature development overlapping with the current implementation or will I have to go without suggestions for a time?

    • Hi Emily! Good news! Content suggestions should not disappear on you. Even though they are not quite a product-market fit, I heard from enough customers who love them that it didn’t feel right to remove them. They’ll stay in the current format until we are able to do a larger-scale revamp with more personalization. :) Hope that helps!

  • I never used content suggestions for the Twitter account I have connected to Buffer, because the suggestions were never relevant to the audience of a chick lit book blog. But suggestions based on who the account follows would be a HUGE game changer for me! One of my 2015 resolutions is to share more from that account, and not just my own content, so the content suggestions would make that so easy!

    • That sounds like a great resolution, Brittany! Here’s hoping we will be able to help you out with your mission in 2015!

  • Love the thought process here. Can’t wait to see what’s next. Thanks for sharing!

    • Hi Carol, thanks so much for your kind comment. Looking forward to the next phase!

    • Me too, I am so excited about the new features of this tool.

  • Wow, thank you for sharing the process and evolution of suggestions! I found Daily to be a huge boon when selecting great content to share. I’ve even received some very pleasant compliments on it, which I pass on to you :)

    I think it’s great to use data as a flashlight to illuminate challenges or obstacles, but never as a hammer. From what I’ve gleaned from reading this and through my experience, having suggestions coming in based on whom you’re following would be well worth considering. Integrating that with lists might be even more powerful for focused and relevant suggestions.

    • Hi Christopher! Very glad to hear that Daily has been working well for you, and I love your analogies about data. We’re trying to be both intuition- and data-driven here at Buffer, which is an interesting experiment! Love your idea for lists, too. Definitely one for us to think over!

  • Oh this is so exciting! I have always loved the suggested content feature but when ever user was seeing the same thing it became clear it wasn’t a good option to rely on (because then your stream ended up looking like many other Buffer users). It’s great to know that this has now been switched so we all see different posts. That really makes a difference and I’m excited to add back in the suggested content feature to my everyday social media work! :)

    Side note: I love the idea for Daily, I hope to see it as an Android feature soon.

    • Hi Kate! This is such awesome feedback, thank you! I’m glad to hear we’re moving in the right direction to make suggestions more useful to you. And yes, I agree that Daily on Android would be awesome!

      • My pleasure. I’ve been a huge fan of Buffer ever since I first heard of the company. The whole team works hard to bring great solutions to Buffer customers and it is an amazing thing to watch. I’m looking forward to seeing what is coming next!

  • Conrado@MailTrack.io

    Courtney, great article as usual. I asked Adam the other day what happened with with your content collection form. Is there still some way to suggest you articles to be included in the suggestion tab? Thanks!

    • Hey there! Thanks so much for checking out the piece! Right now there is not a formal process for suggesting articles because we’re not doing too much curation as we head into this next phase. So sorry about that! I’ll definitely keep you posted on the blog and social media if there might be opportunities as we move forward. :)

  • Wow! I always wanted to know how this works! Thanks so much Courtney, you’ve made my day. I absolutely love it. And of course you’ve included all the details as always. You keep such high standard of your blogs at Buffer that I start thinking you’re perfect and the best. Which of course you are, but you know :)
    I use the tool all the time and love the suggestions and have also noticed you offer 25 posts now instead of only few of them. And it is better and better. If you’ll craft it the way you say you will – like to offer content my audience would love to see – that will be almost epic feature I must say! Including developers and some AI will be a killing combination, I’m so much looking forward to that.
    What I would love to see too is to include some trending hashtags for each content offer because what I do is to add it to the Buffer schedule and then I must go to the scheduler and add hashtags manually for each post. Maybe something like “suggested hashtags” to add to each post on which I can only click and they’ll be added to the end?
    Thanks again for the article, I love this feature and it is so great you also support other content creators by this.

    • Hi Petr! Cool, I’m so glad this post was able to answer some of your questions about the content suggestions process! It’s awesome to get your insights as a user of this feature, and I’m so glad you feel good about where we might be headed! The idea of suggested hashtags is really neat; I’d love to see how we could incorporate that idea!

      • Awesome Courtney, I’m so much looking forward to the improvements and changes in the Content suggestions. Going to schedule some content right now. Can’t wait to see what Buffer prepared to me for today :)

  • Great post Courtney!

    We’ve gone through a similar exercise and continue to do so at Blitzen. We started with a simple Google form on our landing page to gather information around use cases and the apps they use. We used this data to help decide whether we’d take a more vertical or horizontal approach for our product, which online form features are necessary, and which apps we need to integrate with (Buffer is one!).

    We’re now iterating on this process with our product development coming along. We’ll be removing the Google form and inserting a form build with our own integrated form builder. We’ll gather new information, this form will be integrated with our favourite web apps such as MailChimp and Pipedrive, and it will essentially be the first publicly available Blitzen form (testing on ourselves). We’ll see how the experiment goes.

    As for the content suggestions feature – I think your ideas for the future are bang on. In fact, as I was reading the post, that’s what I found myself thinking. I thought I was going to have an opportunity to deliver some incredible insights for the Buffer team, but you completely nailed it! :)

    Looking forward to it!

    Cheers,
    Josh.G @ Blitzen

    • Oh wow, Josh! Thanks so much for opening up so thoughtfully about your process at Blitzen! It’s super helpful to hear from other folks working on similar issues, and I love the lean way you’re approaching things. Glad you feel like we’re on the right track! I’m really looking forward to pushing ahead!

  • Anastasia Astridge

    so is there an opportunity to get content suggestions for kids/family/education or are you still focused on the 5 initial content areas?

    • Hey Anastasia! Right now we have suggestions for the 5 initial content areas but are hoping to expand that really soon and get some categories that get much more specific to what each user likes to share!

  • Mehdi Yousfi-Monod

    Hello Courtney,

    Thank you for your very interesting content suggestion story at Buffer. In my previous position I worked on social user interests analysis and I’m convinced your feature could benefit from interest dedicated algorithms.

    The technology I worked on used users’ textual publication content to create thematic signatures representing users’ interests. It was based on the Tweet content and, if there was a link, the Web page content. Such a Web page provides more content than short and possibly ambiguous Tweets.

    Thematic signatures can be generated from any textual content: A user’s wall, a web page, a full web site, a user’s biography… So for curation I can instantly compare the signature of a user with any document coming from any source flow and tell which of them match enough the user interests.

    Such a technology can answer the question of personalized content. It should be used in conjunction with a “popularity” metric to filter out ordinary contents that match your interests but are not exceptional. The above mentioned suggestions about the retweet count among your community would be great to help calculating that “popularity” metric.

    I would love to discuss that approach with you if you’re interested.
    Best regards,
    Mehdi

  • Field of Success

    Kevan, thank you for getting back to me so quickly and redirecting me here… I would have been on here daily adding my articles in hopes of getting added to the suggested content. I read this entire article by Courtney you sent me, great stuff Courtney! This took time to write, and I guess it was a pin drop to the task of updating the way suggested posts work. You guys are setting the bar for businesses and entrepreneurs as far as customer interaction and feedback goes. I will be signing up for the Awesome plan hopefully sooner than later… I am waiting till I have enough of my own content written and till my Twitter following grows a little more… My goal is by Summer to have enough of an audience to upgrade…I am brand new to both twitter and your app which I found shortly after I launched my website in December. So far I love the app! You guys built a great tool! Thanks!

    • Hey there! So happy to hear that Kevan got you directed to the right spot and answered your questions! I’m wishing you tons of success with building up your content and audience. We’d be happy to have you with us at Buffer anytime!

      • Field of Success

        Thank you Courtney! One question, maybe its part of the Awesome plan. But is there a way that you can add your posts from your page to some kind of log in your buffer so you can just click to add them into your queue? In other words say in the future when I have 100+ articles can I see these on a separate page set up with the photos and everything so I can just click to add them to my buffer queue?

  • I generally find the suggestions from Buffer interesting to me specifically. I’ll save them to Instapaper, but almost never want to share it via social.
    If Buffer had a combo of suggestions based on following/followers, some interest topics to personally chose ourselves, overall Buffer user shares, and some Buffer curator picks, it would definitely be hit closer to the center of the target for what I’d share.

  • Jay Doubleu

    So…can we, or can’t we change the suggestion types? I’m super-confused about what value “Suggestions” provides…as it’s not useful if it’s only “Buffer’s Ideas”.

    Can someone definitively clarify — what practical functionality does Suggestions provide today for someone?

    Can we get suggestions without having to use a cellphone application?

    • Hey there Jay! Right now, one cannot change the suggestion types. I agree that would be the most useful way to use this feature, and we are working on creating a way to personalize suggestions right now. Right now, they’re only useful for the segment of people who want to post about the five areas we cover (lifehacking, marketing, business/startups, inspiration and design) and we’d love to make them a resource for everyone! Excited to keep going on this and let you know when we feel like we’re there! To answer your other questions, you can see suggestions in the web app as well as on mobile.

  • Dwayne Matthews

    Great article! Courtney we have created a curator algorithm that learns from user feedback. If you or someone from your team would like to chat about it let us know…..

  • Jennifer Karin

    Looking forward to seeing this new functionality. Suggested Content separates Buffer from other dashboards, in the best possible way.

80,000+ social media marketers trust Buffer

See all case studies