Reverse Engineering Addictive Consumer Products
Like most people who use the Internet (or phones), there are a few products that I use almost every day. It’s fair to say I’m addicted to some of them. For example, at least once a day (and almost always multiple times a day), I check the following:
- Hacker News
There are also a few products that I don’t check as often, but I know several people who visit regularly and get sucked in. For example:
- YouTube
- Imgur
- 4chan
As I tried to drill down into why so many people, including myself, love and use these sites/apps/services, I’ve found many common threads that tie all of the addictive Internet products together. I can only provide an anecdotal guess as to the importance of each of those factors, but I think they all contribute. More importantly, by reverse engineering what these factors are, we can begin to tie them into the products we build. Thinking about some of these aspects has caused me to rethink grasswire altogether, to the extent that I’m (even more) ashamed of the current iteration.
Of course, there are some products that are popular as a result of optimizing for the addictive products — for example, BuzzFeed drives most of its traffic from Facebook. I won’t be analyzing those products specifically, because optimizing to be shared on a product or network requires a different strategy, and a lot more reliance on factors you don’t control. It is possible for a product to shift from reliance on another product or network into becoming a destination in its own right (like Imgur riding the Reddit wave until it became relatively self-sustaining), but the former doesn’t necessitate the latter. BuzzFeed, for example, seems to be right on the cusp. Some people are starting to visit BuzzFeed directly, but anecdotally that site hasn’t tipped the scale yet.
Cadence
Your site design is a visual metronome from which visitors learn the proper cadence of their visits.
- Eugene Wei: Head of Product, Flipboard
Every site or app has a cadence — the rate at which the content (which really is the core of the product) flows or changes. In other words, how often you can check back and find something new determines, to a large extent, how often we check in on a product.
Among the most popular layouts on the Internet and mobile, as pointed outby FlipBoard is the “waterfall” design. Newest stuff goes to the top and slowly falls down over time.
We find the waterfall design in 8 of the 10 products listed above. And Pinterest only diverts from the “waterfall” in the sense that it has multiple columns. YouTube is more complex, but doesn’t stray too far.
Visit often for the latest and greatest, this waterfall won’t ever run dry.
For 8 of those 10 websites, the design is incredibly simple — one column of content, with all of the external stuff thrown around the outside.
From a design perspective, a lot of the Internet is actually incredibly simple. Most of the things that you use are basically big lists. The tricky part from a product perspective is how to get the list just right. Because, at the end of the day, that list is your product.
Reddit and HackerNews are particularly interesting examples of this. The algorithm that determines the ordering of the content on the front page is affected by time and simulated gravity. In other words, older content is pulled down as good content is pushed up. The delicate balance between these two things makes for a front page that is both interesting and timely. From a user perspective, that means the site is going to be different when I check it at 8 AM and 8 PM. Most of the top stuff will still be there, but I may miss out on a lot of what is cool, yet not cool enough to stick to the top of the page all day long.
I wonder how much of the use of these products is driven by FOMO. I may only have one shot to see that interesting article on HackerNews; I don’t want to miss it.
The easiest way to have a cadence is with a basic recency sort, in which the newest stuff goes to the top. Twitter and Instagram, follow this paradigm, but doing so requires a user to be able to prune what they see, and may only work to a certain level of scale.
Facebook, for example, is convinced that it is now too big for a recency sort. I want to be “friends” with my great aunt Gina, but I may not want to see everything that she posts. So Facebook has implemented an algorithm that shows me new content based on what I have interacted with in the past. I would argue that it needs some tweaking (my Facebook feed is 90% clickbait articles and random statuses about football games from two days ago), but Facebook could do some really egregious things before people would stop using it.
It’s interesting to note that Twitter is contemplating an algorithmic feed as well. I feel like that would “break” Twitter, but that’s mostly because I strongly dislike Facebook’s feed. I’ve talked to quite a few people who think the cadence of Twitter is too fast — one person I talked to even described it as “brutal.” Twitter can feel like a party you miss if you’re not using the productright now. Some less-hardcore users find it frustrating and ephemeral.
Speaking of ephemerality, there’s 4chan. 4chan content starts at the top and slowly falls into the abyss unless someone else comments on a post, pushing it back up to the top. This almost acts as an activity-based algorithm; by definition what you are seeing is what people are commenting on right now.
There are many ways of implementing a unique cadence or flow for a site, but the important aspect is that the product is always changing. I can feel justified logging onto Twitter ten times a day, because there’s always a new conversation taking place.
Signs of Activity
“Users like to feel a part of something. If they showed up to the website and the front page was blank, it just looks like a ghost town.”
- Steve Huffman: Co-founder, Reddit
Although it may not seem like it at first, in the beginning every consumer product is a two-sided marketplace. You need people to use the site, but you also need people to be using the site in order for it to appear interesting enough to be worth using.
Social proof is a very real thing. Whether you’re talking to VCs, who act like high school girls, or actual high school girls, people make decisions based on how other people think and behave. The only way to get people to use what you built, most of the time, is to have people using what you built. It’s an awful Catch-22.
Every successful consumer product that I can think of has user interactions front and center.
They’re not even a click away — they’re part of the piece of content itself. Every one of them has numbers of upvotes, points, comments, likes, retweets, favorites, repins, etc. on every piece of content, and on the home page. Every product I can think of consists of serving up a list of content that people can react to with little to no effort.
This seems obvious, but when designing a product we usually want to hide all of that because it’s not “pretty,” clutters up the page, or, worst of all, we don’t want to show the world that there’s nothing happening.
Common advice for businesses facing a chicken/egg problem is to go whole-heartedly after either the chicken or the egg. A consumer product is no different. Luckily we can borrow from what has worked for other companies to figure out how they were able to overcome the same challenge.
Faking the Chicken
Perhaps my favorite aspect of the Internet is that “users” and “people” are all just data. There is no fingerprint, no social security number, just an IP and what you choose to do with it.
That may seem philosophical, but at the end of the day a “user” is someone who is taking some sort of action in a product. If it’s your product, you control the circumstances under which actions can be taken. And you also control the definition of a “user.” So, you need 1,000 “users” on the site taking actions for the site to be welcoming enough to other users? Just be those users. Literally.
Steve Huffman of Reddit now openly admits to the fact that in the early days of Reddit he each of the co-founders had 1,000 accounts. They would submit, vote, comment, etc. If you were a new user, it appeared as if there were 3,000 active Reddit users. So you might sign up and join them.
Eventually you reach a point at which you don’t have to be 3,000 users. As the user base grows, you slowly wean the site off of the fake users, and all of the sudden you have a thriving, active community.
This takes months of hard work, but at the end of the day everything is pixels on a screen that you control. If you want it to look like there are users there, do what you have to do. The Internet will collectively forgive you later.
Exclusivity
Some of the most successful product launches of all-time happened as a result of exclusivity. The first one I ever participated in was begging for a gmail invite. Later versions include the million-person-long waiting list for Mailbox, Dribbble, ProductHunt, Facebook only allowing @harvard.edu email addresses when it first launched, etc.
This hacks user acquisition from a couple of angles: First, there is a (false) scarcity. Usually when you hear about a product you’re checking it out, deciding if it’s something you want to use. When you have to fight for an invite, however, as soon as you receive an invite your mind is in “use” mode not “explore.” This is a subtle but important difference.
Exclusivity also lets you generate a rush of use all on the same day. A good product launch from even the least-well-connected of entrepreneurs will probably result in 10,000 people checking out a site. If you can get all of them to do so on the same day, and with the intent to use instead of explore, you may have successfully avoided months of would-be fake users.
It’s also noteworthy that I would never “share” about how awesome Google Inbox is. But because I had a few invite codes, I think I’ve tweeted about it five times now to get them out.
Of course, the reverse is that you’re not letting everybody who wants to see your product see it; this could come back to bite you. I was annoyed that I couldn’t try Mailbox when it first came out. But then I got an invite, and I downloaded it and used it (until I realized it wouldn’t search email archives and deleted it, but that’s another story).
One-Click Interaction
In the early days of Facebook, we wouldn’t allow ourselves to eventalk about virality. We answered three questions:
How do you get people in the front doorHow do you get them to the “aha moment” as quickly as possible, andHow can we deliver core product value as quickly as possible.
- Chamath Palihapitiya, formerly growth at Facebook
We launched grasswire to a ton of traffic, and a lot of interest. Thousands of people signed up, 17,000 people visited the site in the first few days, and the average time on page was about four minutes. All it did was show a real-time feed of tweets, photos, and videos that were gaining traction for a particular keyword (in this case the Boston Marathon shortly after the bombing). I was ecstatic with the first day raw numbers (it was the first product I’d ever really brought to life).
But people weren’t using the site how we wanted them to. They would sit and watch like it was TV, but nobody was interacting with it like a website. They weren’t “favoriting” or fact-checking or sharing… they would just sit and watch.
I was talking to a friend who was ranting and raving about what we had built, when I finally asked him in frustration, “Why aren’t people using it though? Sure, they’ll sit on the site all day, but nobody is even favoriting stuff except for me and my co-founder.” His response? “Wait, you favorite stuff? What’s that?”
Unbeknownst to my friend, and apparently virtually everyone who ever visited the site, you could click on a piece of content to see a detail page. From the detail page you could take all sorts of cool actions, which would make up (we had hoped) the most valuable aspects of the product. There was only one problem — people never knew you could do that.
It’s difficult, as a product creator, to realize that people can’t read your mind. Most of them don’t really care about what you built. If I think something is cool, it warrants a minute or two of my time; certainly not long enough to train me how you want me to act, even if you have a great walk-through.
We had one metric we looked at — the number of repins. That meant someone understood the product and found something they liked. We optimized everything to increase that number. That’s the only thing we cared about.
- Ben Silberman, CEO of Pinterest (paraphrased from my notes)
Almost all consumer products are driving you to do one thing. They put that one thing front and center. That thing is usually typed out, also — you can either click a thumbs up on Facebook or you can click “like.”
When you look at your product with a repeating button it almost feels like overkill — “Are we really going to have that button on every single piece of content on our page?” Yes. You are.
Notice that every single product listed above has one main action you can take directly from the homepage. Even Pinterest, which is very visual, and certainly must have had concerns about displaying a big “Pin it” button across all of their images, makes the button virtually impossible to miss.
Facebook goes to the extent of leaving an open comment box at the bottom of every single piece of content, and leaving a status box at the top of the page. It practically begs you to interact with it.
In fact, every interaction that you can take on Facebook is displayed on every single piece of content. There aren’t really “detail” pages where the action happens (other than to see a photo in more detail). Unless you’re viewing someone’s specific profile, you’re always looking at the main feed. (And when you are looking at their profile, you’re still looking at a feed, it’s just a feed of the content they’ve shared).
Imagine what would happen to Facebook’s interactions if you had to click into every piece of content to take an action on it.
Notification of Interactions
I couldn’t decide if I should title this section “ego boosts,” “dopamine hits” or “notification of interactions.” I went with the safe one.
I’ll go back to a product every now and then if I think there’s going to be interesting stuff happening there. But if there’s going to be something that involves me? I’ll hit that site all the time.
The quickest way for any new user to reach the “aha” moment is to have a solid, positive interaction on your product. Sometimes that just means somebody paid attention to them enough to respond to a comment. Sometimes that means “karma.”
Karma, or imaginary internet points, are a fascinating phenomenon. Studies show that imaginary, worthless Internet points can keep us addicted to something in the same way flashing lights and chimes can keep us in a casino, losing our money. Points, badges, and other sorts of gamification all work much better than most of us would care to admit, but I think karma is by far the most interesting.
Karma is interesting because no matter what our number is, the base is arbitrary. We’re always looking for those next few points. When they come, it doesn’t matter if it’s 1 point or 100 points (it’s all arbitrary anyway), gaining those points releases dopamine and other endorphins that make us feel good. If you feel good, you’ll like whatever made you feel good, and keep coming back.
This is where the field of user acquisition gets a little bit scary — we haven’t evolved in such a way that our minds can fully process the worthlessness of worthless Internet points. Where exactly you fall on the ethical spectrum is purely up to you, but it’s definitely worth noting that seemingly silly things like that work.
Long-term View
Everyone who has built a product that becomes addictive will run into the “timeline question” at one point or another. It may happen unknowingly or subconsciously, but the timeline question usually sounds something like this, “We sent out a push notification and got 5,000 people to come to the site!” The natural reaction is to start sending out push notifications like crazy. The timeline question is, “At what point will people stop coming to the site?”
I noticed this happening with Quora. It’s almost impossible to avoid with LinkedIn. The incessant over-notification only serves to drive users away. Making it difficult to unsubscribe or turn off notifications is a sin of even greater degree.
Reddit and HackerNews have won a fanbase without even collecting emails. You’d be hardpressed to find anyone who adores a product or community more than those two users bases. But getting to that point is extremely difficult.
Twitter and Facebook survived in the beginning in large part thanks to emails that went out. If you interact with a site and never go back to see if there areany notifications, the site is dead to you.
I only include this to say, “Make sure you keep a delicate balance.” Initially we had no emails whatsoever for grasswire, but some of the older, less-technical audience started practically begging for them; we’ll have to build those into future iterations of the product. Other people email me saying, “I want push notifications every day.” Personally, I despise push notifications (in part because I feel like I’m falling into some marketer’s trap, in part because I put a high price on my focus). But I’m not our users. I’m not everyone. The rate of notifications is probably one of the things users should have complete and total control over. As one of the marketing buddies I used to work with puts it, “Don’t try to lose weight by chopping off a leg.”
