Designing Home Feeds
Let’s say you’re an app that wants a compelling, sticky homepage to bring your users back to everyday. Your content could be anything — photos from friends, a catalog of movies or an ecommerce site with millions of items.
You want to do your content justice, but there’s too much to show at once. So what do you do? For example, what if Amazon just put items as they came in on top of their home page?
Picking, reordering and arranging content can be really hard. As the example illustrates, one size does not fit all. Twitter orders its feed by time, with a small ranked section. Facebook preserves recency, but some life stories and photos are more important than links. Netflix doesn’t care about recency but buckets into genres.
So, which approach is the right one for you? Here are some guidelines to keep in mind when designing your own feed.
How important is recency / timeliness to your content?
The faster your content becomes irrelevant, the closer your feed has to be to strictly time-ordered.
Sometimes you can’t afford to reorder content in a non-chronological order. Think about how this applies to your pool of content. If you’re a Amazon or Netflix, time hardly matters, but if you’re Twitter, this is critical.
It’s important to think of this constraint first. The less you’re bound to time, the more flexibility you have in future design decisions.
How much better is your best content than your worst content?
If the pool of content is all high quality and curated, you probably don’t need Machine Learning to filter the signal from the noise.
Oftentimes, the best and worst content in the pool are not that different in how engaging they are. For example, Instagram is about visual stories. My photos are all about the same quality — somewhat decent photos of food for example. This is because I only follow a handful of people but as your follow graph grows, not all content is equally important and you need to algorithmically order your feed.
When the pool of candidate content is homogenous, you don’t need it reordered with a Machine Learning (ML) algorithm. But if the best (think wedding photos) is much better than the worst (reshared posts), then ML is a godsend for relevance.
How diverse are your audience’s interests?
The more similar your audience’s interests, the less you need to lean on personalization or interest graphs.
Global quality isn’t the only thing that matters. If your pool of content gets really large then you only want to show what’s personally relevant to the reader. Amazon can suggest I buy their top rated diapers all they want, but I’m simply not in the market for diapers right now.
When you take your objectively best content but not everyone wants it, you’re in the personalization business. Now, you can do this through a social filter like a Twitter-like follow graph, which is what Medium does for example or through Machine Learning on a topical interest graph, like what Quora does, or through inferred interests through implicit actions, like YouTube or Netflix.
Special-interest products like Hacker News or subreddits don’t need a personalization layer, so they don’t really need an interest graph. If anything you find new people to follow on those platforms that you wouldn’t have otherwise known of from your existing Twitter graph.
When you have the privilege of a diverse content pool as well as a diverse audience, then this becomes really important.
Does your audience have lean back or lean in intent?
The lazier and less “intentful” your audience is, the closer your feed has to be to a single sorted list. The more intent the explore, the more you can add dimensionality to the UI.
There rarely is a “reason” for you to go to your Facebook feed, at least not one with intent. Search, sure, but feed is designed to pass the time and given that you have no real intent, Facebook’s designers want to minimize the decisions you’ll have to make, and the simplest way to do that is to offer you a single infinite list of interesting stories.
Meanwhile, when you go to Netflix, you’re probably in the mood for a certain kind of movie. You might even have a genre in mind, like horror for example. Netflix could also have a few suggestions in mind for you, but you’re likely there with more intent, probably because you’ve decided to spend the next few hours of your life there. A multidimensional paradigm helps balance research with relevance and let’s viewers deep dive into interest spaces more easily.
Are you better off nudging decisions?
Sometimes your audience is so passive, you might be better off making decisions for them.
You don’t always go to Netflix with intent, sometimes you just open Amazon to see what’s cool or to get ideas. If your goal is to drive people to a single decision point, e.g. buy this product or watch this movie, then one of best ways to convert people is a single featured item.
The human mind loves defaults, especially when it lacks intent. This is why calls to action are such a powerful growth mechanism. This, sadly is why large banner ads work. This is also why cover stories and headline items work.
When does connecting to Facebook really matter?
Look at your content pool and ask yourself “Do I talk about this a lot with my friends?” Only if your answer is a very strong yes do you need a symmetric friend graph like Facebook.
I ask myself two questions when considering building a symmetric friend graph and using that to power a feed.
- Do you want to see what your friends are doing?
- Would they be comfortable sharing what doing on the network by default? I add the by default because that’s usually the only way to create enough content to create an engaging feed.
So let’s walk through some examples against these questions.
- Shopping: Rarely want to see what friends are buying, they almost never want to share by default.
- Music, Books: Do want to know what friends are listening to, they’re usually comfortable sharing.
- Movies, TV: As a lower artform than music or books, we’re not usually as proud of sharing all the movies and TV shows we’re binge watching, and especially how much time we might be spending on them.
So there it is, those are the most important factors I think of in when I design a feed. Of course, there are many others. Hopefully, what I’ve done here is give you one systematic way to start thinking about building feeds. Please feel free to fork this process into your own.
This is a longer, less edited version of a post I originally wrote for the Quora Design Blog. If you enjoyed reading it, please pass it along by recommending or sharing below.
