Twitter recently announced its Q1 2016 results. While its revenue grew by 36% to ~$600m, and losses halved to ~$80m, the numbers didn’t impress Wall Street analysts or tech pundits. After all, revenue growth is slowing, audience numbers are barely moving and there are newer more exciting kids (snapchat, medium etc) on the block. The stock dropped as much as 13% on the news, and is now trading at its all time low, below $10b market cap.

The growth challenges that Twitter is facing isn’t news of course. There has been a slew of stories on this, peaking in the last year as these led to the then-CEO Dick Costolo being replaced by co-founder Jack Dorsey. For some representative samples, look here, here and here. Fast-forward 10m ahead, and it doesn’t look like Dorsey has been able to move the needle much. The fundamental challenges (anaemic growth in audiences led primarily by complexity of the product) still remain, and they aren’t going away too.

What really can Jack Dorsey, and the A-team at Twitter do?

For one, the team needs to change the framework through which they view Twitter. Secondly, they need to understand that Twitter’s challenges aren’t specific to twitter alone, but are the challenges of online media, and cannot be seen in isolation. Solving Twitter’s challenges is really about solving online media’s challenges.

Let us look at each of the above two steps.

(1) A framework for understanding Twitter
Of late, Dorsey has been pushing the framework of ‘live’ to understand Twitter. In February announcing Q4 2015 results, he said “Twitter is live: live commentary, live connections, live conversations.” As a strategic approach this meets at least one of the key tenets of strategy: clarity. It says what you will focus on, and what you will not. Yet is this framework correct? How does one validate it?

Dorsey’s definition can be evaluated only if we agree that Twitter is a media product, to the exclusion of anything else it can be. But should Twitter be a media product to the exclusion of all else?

Around the time of its emergence, Paul Graham, of ycombinator fame, wrote a wonderful blogpost about understanding Twitter as a protocol. This framework, viewing Twitter as a one-to-many communication protocol where the recipient is unknown, is a good lens to understand it. It is not just Paul Graham – see links , 2 and 3 which talk about viewing Twitter as a protocol.

Let us take email – a one-to-many protocol with known recipients. Imagine if there was just one email service or product where the protocol and application was integrated. How much innovation would then have resulted? Would Gmail, Blackberry or even Mailchimp have happened? I doubt it.

Email’s growth and the numerous innovations that have emerged around it have arisen due to the separation of the protocol and the product(s). For a protocol that is over 40 years old, not a year goes without some new product launching around the protocol. Product creators are free to build on the protocol, add bells and whistles etc. If email had evolved akin to Twitter then it would have become perhaps an instant messaging platform (live conversations, live connections)!

A bold step thus for Twitter would be to separate itself into two
1) a protocol co; we could also evaluate whether it should also be a personal information manager or PIM as advised by Dries Buytaert, where all the personal information of the user is held. This enables the user to update profiles, privacy settings etc
2) a product co that creates several Twitter apps / products, and competes with other creators of apps and products, which are built on the Twitter core api / social graph. Arguably, you would create several Twitters – a live Twitter, perhaps a delayed feed Twitter with the highest RT’d and liked tweets, corporate Twitters etc.

What is also interesting is that products built atop Twitter such as Nuzzel, one of the really interesting leveraging of the Twitter api, are able to convert noise to signal through effective filtering. As an example take the concept of following in Twitter. To really go through every tweet, one will need to put a realistic limit on the number of people you follow – 200 or perhaps 300. However the more people you follow on Twitter, say if you follow 500-1000, the better the data generated by Nuzzel (i.e., more interesting the recommendations), even though it becomes harder and harder to consume all those tweets per se.

Let a thousand Twitters bloom!

(2) Separating curation and content doesnt work in the long run
This year the Pulitzer prize for feature writing went to Kathryn Schulz of New Yorker for her piece The Really Big One, on an impending earthquake in the northwest U.S. I thought it was an outstanding piece. Did it deserve the Pulitzer? Perhaps. It helped that it was in the New Yorker and the recommendation letter was written by David Remnick, the legendary editor. Would it have won the Pulitzer if it had been in Aeon or Nautilus, two outstanding online science publications out of the many that exist online? Well, I am really not sure.

The New Yorker or its peers such as The Atlantic or Wired offer a curatorial service as much as a content package. In fact, the two, the curation and the content bundle are so tightly tied together as to be indistinguishable. In their case the curation is tied to the creation of content itself, given that these pieces are commissioned.

In the case of Aeon, the curation is post facto creation and even publishing. Once the story appears on the site, recommendations on 3rd party platforms such as Twitter, Facebook, Reddit and the links in newsletters etc all drive readership. The fact that it is on Aeon itself carries no signal. There is no curation power inherent to Aeon, at least for the average joe or jane reader.

In fact, many readers who come to popular stories via a recommendation do not even know which site they are reading the story on. They read it, and then close the link. The wrapper, such as Aeon or Prospect magazine has no inherent meaning. The website itself has got atomized into individual articles. This is true of most websites barring perhaps a dozen destination news and feature brands.

So curation / recommendation has got disconnected from the content in two ways. It increasingly happens after content has been published, and not before or during the publishing. Secondly, it is moving out to 3rd party platforms such as Facebook, Twitter, Reddit etc.

As a result, given the immense power of the curators, the playing field is shifting away from content providers. We can see the impact of this in Facebook’s Instant Articles, and its competitors such as Apple News, which want to host content on their platforms. While the reason suggested is the slow loading time of the page, the real reason is that Facebook understands that it is the reason these articles command pageviews, and it thus wants to capture a share of the revenue that these sites get via the eyeballs that it is sending their way. Google has an interesting variant of this whereby it allows publishers to publish content directly to Google and have it appear instantly in search results.

Not having a platform to host content and thereby monetize is like New Yorker providing a contents page, and saying now please write to the writer or his agent to source the article.

What does this mean for Twitter?

Clearly after Facebook and Pinterest, Twitter is the biggest source of referral traffic for websites. I wonder if Twitter could create a platform for hosting content, especially content that is getting traction via likes and RTs on its site. If Twitter doesn’t want to compete with Facebook and other platforms on hosting mainstream publishers’ content, it can identify niches such as one-off blogposts that get popular (such as Chris Goodfellow’s theory of MH370 crash – initially published on his Google+ page, and then on Wired)

If it can do this, it can compete with Medium effectively. Medium has succeeded in providing infrastructure that makes content creation easy. It hasn’t however been able to make content discovery as easy. It remains largely a Silicon Valley echo chamber. Given how broadbased twitter is, I can see it becoming the pre-eminent content discovery and consumption platform rapidly.

To summarize
1) Twitter needs to encourage diverse experiments around its api and protocol, harking back to its earlier days, deconstructing itself into a protocol co, and a separate products co.
2) In addition to protocol and product, Twitter should explore becoming a platform for blogposts that are going viral, thus monetizing the interest graph its users create daily.