Improving Twitter
I have been writing this post slowly over the past couple of weeks. Yesterday, Twitter announced “Project Lightning”, which is very much aligned with what I discuss in this post. I am excited for Twitter and think it’s a huge step in the right direction. When Lightning is released I will follow up with thoughts on what they got right and what could be improved.
Twitter has focused its product design around the idea that real-time information consumption and real-time conversation are what people want. But what people really want is relevant real-time information and conversations. Twitter’s incomplete value proposition manifests as a bad user experience, leading to poor retention and ultimately a lack of active user growth. In this post, I will explore ideas like rooms/channels, personas, curation, and unbundling as a means of improving Twitter’s value delivery.
Real-time
Twitter chose to focus its product design and consumer-facing marketing around the idea that real-time information consumption and real-time conversation are what people want. Therefore they built a product with recency at front-and-center; users could see a chronologically sorted feed of every tweet from every person they follow. Said another way, Twitter created a “personalized” firehose. It’s personalized in the limited sense that one can choose who they are following (but that’s it).
Relevancy
Twitter successfully delivered on its promise of delivering real-time information and conversations, yet people aren’t happy and the service isn’t growing fast enough. Why? The main gripe is that a user’s stream is too noisy and it’s hard/unclear how to overcome that. Consumers expect Twitter to deliver relevant real-time information and conversations. It’s not that Twitter needs to automagically deliver individually-tailored data immediately, but it must at least be easily configurable to deliver what one wants to see when they want to see it.
The true power of Twitter is well articulated by Austen Allred: it allows you to be in a room with whoever you want, whenever you want. You can listen to them think out loud, eavesdrop on a conversation they are having, or even converse with them. This is incredibly powerful.
Relevancy in this view of Twitter is the idea of being able to create rooms. You don’t want to be in Times Square trying to listen to everything that everyone around you is saying. You want to be in a room confined to a particular conversation topic or a particular set of people. Equally as important, you’d like to be able to effortlessly switch between rooms whenever you want.
Unfortunately, relevancy isn’t included in Twitter’s value proposition and therefore has not manifested as a first-class construct in the product. Twitter doesn’t natively support the concept of a room. They focus on people. You have to stand in Times Square and listen to everything that everyone around you is saying, and all you can do is allow more people to stand there by following them or ask them to leave by unfollowing them. Therefore, Twitter’s new user onboarding is essentially as follows: pick a broadly defined topic that you’re interested in and from that we’ll suggest 40 people to stand in your Times Square. It’s no wonder that active user growth is lousy.
Digging deeper
We can further see that relevancy is not central to Twitter’s product design by thinking about the following question. What is the best way to skim work-related information during the day and read hobby related tweets in the evening on my iPhone? Not via the Twitter app. The best way on my iPhone that I can switch between my work and hobby contexts is by switching between a list of users who Tweet about work-related things and a list of users who tweet about my hobby. To navigate to a list of users requires tapping the “Me” icon at the bottom of the app, then tapping a small unlabeled gear near the top third of the app, then tapping the word “Lists” in the middle of the app, and then scrolling through an unsortable list of lists and choosing the one I want. Ouch. To make it worse, if you’re a new user and somehow manage to find your way to the list of lists page, it’s empty! No curated lists await you, no suggested lists, no onboarding video, no link to this Twitter help page about using lists on the web app (even though the majority of users and ad revenue comes from mobile app users). Nothing. Since I have no prepopulated lists, I’m resigned to just scrolling through my feed. The only way to see relevant data is to hope the people I’m following are writing what I care about and have done so recently. Otherwise, I’m going to be scrolling for a long time.
It’s painfully obvious that relevancy is an afterthought in Twitter’s app design.
TweetDeck
The closest thing to letting me create rooms and switch between them involves getting off of my phone and leaving Twitter entirely to go to TweetDeck (or another third-party app like Tweetbot). TweetDeck allows me to create lists of users, put them side by side, and easily scroll among them.
Note that TweetDeck is more powerful than Twitter when it comes to letting you control the list of attendees in a room. Because Twitter is user-focused, you can only create a static list of users. With TweetDeck you can create a column based on a search term (hashtag or phrase) which enables you to create a dynamic room of people based on whoever is tweeting and including that term. A search just returns all results which is still too noisy, so there is definitely room for improvement on this front. For example, curate tweets around a topic or term (more on this later).
Twitter bought TweetDeck back in 2011. Amazingly, even though TweetDeck is better at delivering value and has been under Twitter’s wing for four years, new Twitter users will not discover TweetDeck unless someone shows it to them. There is an astonishingly tiny amount of content on Twitter’s website about TweetDeck, and it’s incredibly hard to find unless you search for it (which requires that you know about TweetDeck in the first place). Twitter doesn’t want to tell you about TweetDeck because they haven’t introduced ads into TweetDeck. Twitter is missing a big opportunity here, as users can be even better targeted in TweetDeck. Inside TweetDeck, I could be shown enterprise software ads in my SaaS thought leader column and shown diaper ads in my #DadLife column.
Rooms
The Twitter mobile app must enable users to create and effortlessly shift between “rooms,” where a room is defined as a feed of relevant tweets from a configurable set of people. The configuration of a room’s membership should be more powerful than just a list of users. A list of users is a fine start, as it allows one to restrict membership to a handful of people who fit a particular profile such as famous NBA stars. But the membership of a room should also be dynamic, such as a room full of any Capitol Hill staffers who happen to be discussing foreign policy.
From an implementation perspective, enabling dynamic room membership sounds difficult; hashtags or search terms can be used to filter members based on a topic, but they don’t filter members based on their profile. Here’s where the idea of allowing users to have one or more definable “personas” would be useful. The membership of a room could be defined as a particular type of highly-ranked personas who are talking about a particular range of topics. Here a “persona” is the description of the profile a user is posting as. A concrete example of a user assuming different personas would be Jessica Alba switching between her actress persona and her entrepreneur persona depending on the content of the tweet she is posting. A persona enables users to tweet from their different vantage points in life and allows followers to more effectively filter the tweets of a particular user. In a future post, I’ll dive into why Twitter should enable personas, how they could manifest in the product, and how to simplify the implementation.
Twitter must add room controls to their app home screen so users can switch between and create/delete rooms. While more sophisticated room creation is the end goal, to get the ball moving today Twitter could start with defining room membership using their existing user lists.
Curation
Of course, an empty drop-down of user lists is not helpful for a new user. Along with a drop-down of rooms, a button to add or create a room is imperative. The problem with creating a Twitter list today is that I don’t know who to add to a list. For example, I know that I’m interested in learning about product management and want a room with the best product management thought leaders out there. But how do I know who those people are? Twitter needs to take a page from the Spotify playbook and introduce curated rooms in-app. Yes, Twitter recently launched curated lists for logged out users on desktop which is a great start and especially useful if you happen to not be on mobile and not be logged in. But access to curated lists must be available and easily accessible in-app.
Side note: Spotify has absolutely nailed it with their curated lists, which are both far-reaching in terms of genres but also deep and nuanced within each category. This is not only a brilliant solution to onboarding but is great for discovery (and in my case, retention). Curated lists deserve their own blog post because they are simple but so powerful.
Channels
A curated room could be marketed as a “channel”. (By the way, Twitter does very little customer-focused marketing but really ought to). These channels could even manifest as their own one-page microsites with heavy branding and incredibly relevant ads. During the NHL Finals, one should be able to go to twitter.com/nhl-finals for a full-blown interactive NHL experience powered by Twitter, featuring commentary tweets from analysts, NHL stars, and President Obama. Imagine immediate post-game ads for t-shirts that say “Chicago Blackhawks: 2015 Stanley Cup Champions;” I’m much more likely to click on that right after the game than I am an advertisement for a “Sales Hacker Conference in Atlanta” (I’m neither in Georgia nor sales).
Stronger Interest Graph
Allowing a user to switch their firehose feed out for a more targeted, tailored feed on demand yields very specific insight about what I’m looking at and when. This information is great for strengthening Twitter’s interest graph; Twitter could know what room I’m in, how long I’m there, how often I visit a room, and how much I engage per room. In the short term, this data would help Twitter sell even better-targeted ads. Long-term, this behavioral data could be used to better deliver even more relevant information to the end-user.
Unbundling
Relevancy should be a key ingredient of Twitter’s main value proposition and must manifest in the app going forward. Yet there are still other useful characteristics of Twitter, such as being the best place to discover real-time news. One option Twitter must explore to take advantage of these additional value propositions is to “unbundle” Twitter. For example, is a tiny little search bar the best possible interface for a real-time search engine? Consumers and advertisers would benefit from a completely separate website that delivers the best possible real-time search experience and results. New users would benefit from the clearer positioning and separation of value propositions (i.e. for the best real-time search I know to go here and not the Twitter app). Advertisers would again have the opportunity and real-estate to provide even more relevant ads. Similarly, is a reverse chronological firehose the best way to read conversations (even with their improvements)? Or is there a better way to unbundle conversation view into a separate website and/or app? A future blog post will explore this idea in more depth.
Good news
I have full confidence that Twitter’s engineering team is capable of enabling relevancy in-app. The technology that powers Twitter is absolutely amazing. While product innovation has been slow, we must acknowledge the challenge of improving Twitter while keeping the fail-whale at bay. That would be difficult, although money and a bunch of experienced and smart people seem to help with difficult things.
Summary
Twitter needs to enable users to easily discover relevant, real-time information by adding the ability to discover, create, and switch between rooms/channels within the app. To further cut down the noise they could implement user personas. Finally, Twitter should even explore unbundling their various value propositions into separate websites and applications.
Twitter Lightning is promising to bring “event-based curated content” to Twitter, which sounds very much like curated channels I suggested above. This is great news and the timing is impeccable. I look forward to Lightning’s release.
