Best possible post for each network; smart and effortless.
Crowdfire is continuously and rapidly evolving into a smart, personalised assistant poised to give creators the growth their creation deserves. A few months back we announced the launch of our latest feature — Tailored Posts. It is the most polished product of our do-it-for-me strategy.
Tailored Post solves the problem of sharing content on each network, smartly and effortlessly. It auto-tailors perfect social posts for each network based on what you are sharing.
You can see how it works below:
Three years ago we launched Publish, a scheduling feature, quite like Buffer or HootSuite with a USP, “ auto-post at best time for maximum engagement ”. This took away a load of deciding when to post away from users and kept things quite simple to one button. Everybody loved it. Publish took care of the ‘When’ to post for our users. Digging deeper into Content Marketing.
Content marketing is also What, Where, When and How to post.
This past one year, we focused our energy on bringing post ideas that are — more relevant to the user, drive more traffic to his creations and keep his timeline engaged. We built daily Smart Posts — sharing content that you create (like videos, blog posts, products on your online store), recycled smartly and evenly distributed to drive more traffic. We also built article recommendations which are based on your twitter trends, topics you want to be famous for. Our users loved it, soon they drove more traffic to their creations, promoting their content multiple times, their timelines were not empty anymore, We saved them a lot of time. This contributed to a major shift in metrics for publishes from self compose to recommended articles. We are now in business of post ideas.
While working on how to help creators share their content everywhere smartly, I felt the experience of how to post different content for each network was also broken in this world.
A perfect for each social network — was now our responsibility since we were also recommending content to our users.
Any passably good social media manager would tell you that what works for marketing on one social network does not work for another. For example, you can get away with posting over a dozen times a day on Twitter, but the same’s not true of Facebook. Or if you create a video on Youtube, you could share it right away on Facebook but would have to create a shorter teaser version for Instagram. Also, some posts work well with multiple photos.
So, a great social media manager would actually spend time tailoring a post for each of the different social media networks its audience is on. They will add in the correct hashtags, write up the perfect copy, create accompanying creatives that would garner the most engagement, schedule them for the right times and so on. But, if a business owner were to do it by themselves, they can’t-do the tailoring — which is, clearly, a full-time job.
You see, when you are building your business from the ground up online, whether you are a YouTuber, an Etsy seller, a Blogger, a Freelancer contractor or any other form of one-man or few-men army, you have one goal and that is to build your business. You are focused on creating your next big video or crafting a trendy product for your store. This takes up almost all of your time— as it should.
Marketing your content should not be your full-time job.
An elegant solution to our problems.
We imagined it be like a pro-active assistant which detects your latest YouTube upload and recommends you a smartly tailored social post for each network. It could have a teaser video for Instagram, a 140 (or 280 soon!) character tweet for Twitter and relevant hashtags to go with each post — all you have to do is tap on a button to send the post across all your networks.
The vision of creating the perfect post for each network sounded great, but it was very challenging to keep things simple for our users.
Our goal was to get this complex problem solved in least number of steps. The ideal Minimum Viable Flow for Tailored Posts would go something like this:
Crowdfire: “Siddharth, should I promote your latest video?”
Siddharth: “Yes. Thanks!”
End of story. Beyond this one response, the user shouldn’t have to do anything and Crowdfire would promote that video by crafting the perfect post for each network and scheduling it for the right time.
However, AI is not intelligent enough yet to understand every person and we needed an interim step where the user can preview how Crowdfire will do the promotion with an option to edit the posts Crowdfire created.
Hence we adopted our UX philosophy as preview and share, this is different from review and share … which feels like more work each time. The difference is huge, one is more like a preview of what is going to happen while other is review of each action. This is unnecessary in most cases. Other challenge was also to consider giving user good control over his actions.
So, here’s how it works right now:
Seems simple enough right? However, step 2 & 3 is where the magic and challenge lie.
Every social network has it’s own posting and marketing strategy. So, how do you create perfectly tailored posts from one piece of content?
And when you do, how do we present all the different crafted posts to the users, such that they can easily preview and approve?
Once we craft the posts for each network, previews for all the audience networks are easily accessible by a simple swipe gesture. This is a big step as we really wanted to avoid buttons / tab navigation.
In The Preview Card:
Each card is designed to give the user an idea of how the post will look on different platform. This is important since we are composing the post for the user. Each platform has its own way of interpreting a post and we simulate all such cases. Typically each preview card has over 16 cases to handle like image post, text only, text with a link, video post, error states etc.
Our tailoring logic happens on the server side and front end clients merely simulate post previews for each network based on the content params; views also adapt when the user edits a post. Much of the tailoring of each post depends on the source of the content. It matters if the content is a YouTube video, WordPress blog post, Tumblr post, SoundCloud track or an Etsy/Shopify product.
So, how did we define the perfect post for each source?
Let’s take YouTube. We started by documenting social media post styles of new and famous YouTubers, borrowing from their experience on how social posts should be.
Turns out there are many ways of sharing a YouTube video. You could do an image post, where you share the thumbnail of the video and post it on Facebook and Twitter with a link, but in this case, the YouTube player is not embedded so it does not play within Facebook or Twitter. But, if you share it as a link with no accompanying image then it plays within the network.
The posting strategy is different for Instagram as well, where posting a teaser video might get you more viewership to your YouTube video. The least you could do is create a thumbnail for Instagram and post.
We studied the click-throughs from different post format through experiments until we arrived at ideal posts for each of the social networks.
No. of experiments = Multiple content sources x Multiple social networks x Multiple types of posts
This has been a tediously manual task to begin with. We plan to automate this with machine learning for other networks & sources we integrate into Crowdfire, but for now we needed a logical start for our training models.
We classified the tailoring of the posts: by type, by content, for growth, and by frequency.
Tailoring by Type:
This is the first step in the tailoring process. This step defines how the post should look on the network it is being posted to. Essentially, we create a model of cases mapping to the source of the content.
Major factors that decide tailoring here are the source of this content, the social networks it has to be posted on and the types of posts possible on that social network.
Tailoring by Content:
Next, we focus on the content itself. The objective here is to make the post interesting enough to attract clicks. This is an intensive process that took weeks of experimentation, iterations, and engineering to achieve.
This is where we:
Tailoring for growth: We go on to tailor the post such that it would garner maximum engagement. This includes analysing it and including network-specific hashtags that would work best and increase discoverability. You will see a lot of improvement here in coming months.
Tailoring by frequency: Here we decide how often this content should be shared across the different networks. For instance, if it is your new blog post, you would want to share it more often this month (once a week) while all your old evergreen content can be shared 2 times a month. This is a powerful feature which deserves a post by itself, I will share more about this next time. For now, you can read more about it here.
And, there you have it! The perfectly tailored post is ready for the user’s purview.
This five-step tailoring process and the segregation of the sources, networks, types of content, frequency and the final post makes this feature extremely agile and scaleable. I feel good that the product, design, and engineering came together to solve a problem elegantly.
Tailored post has simplified the mechanics of publishing content everywhere.
That was easy? Do share your thoughts and suggestions below and take a go at Tailored Posts by downloading our iOS or Android app. Don’t forget to connect your Medium blog on Crowdfire.
To shoot the breeze with me about products or product designs, hit me up @BuddhaSource — DMs are open.