Twitter’s lack of growth over the past few months has quickly become its defining narrative – witness Inside Twitter’s plan to fix itself from Quartz, which despite the headline, fails to actually explain anything about said plan.
As with most things I write about Twitter, I have no particular inside knowledge of the company’s plans, but I’ve written over and over about its core failing, and promise. In 2008 (!) I suggested “TweetSense“, and in 2011, I wrote Twitter and the Ultimate Algorithm: Signal Over Noise (With Major Business Model Implications). It opens with this:
My goal in this post is to outline what I see as the biggest challenge/opportunity in the company’s path. And to my mind, it comes down to this: Can Twitter solve its signal to noise problem?
I go on to say that it most certainly has to, because solving the problem allows it to attach sponsored advertisements (promoted tweets in particular) to just the right timelines in just the right context. I called the solution “TweetWords” – because AdWords came before AdSense. Twitter’s promoted tweets product did in fact evolve toward interest-based targeting – alas, in one way only, as far as I can tell. Advertisers can target Twitter users based on their interests (as expressed by what they tweet, retweet, follow, etc.), but they can’t place their promoted tweets contextually into timelines (IE, in a manner that “fits” with the content around them). **Update. Twitter has had keyword targeting – a key step in contextual ad targeting – for a year now. I missed this. My apologies.
So far, there’s no such thing as TweetSense or TweetWords – where ads are contextual to the stream in which they appear. It seems Twitter has not focused on this particular problem – and it may not have to. Revenues are doing extremely well, and Twitter is clearly opening up new forms of advertising based on larger formats, video (Vine), and cards.
But if the core problem of understanding individual timelines as context is not going to be solved, it’d be a shame – because solving that problem will address Twitter’s core signal to noise issue as well. Here’s more from that 2011 post:
If Twitter can assign a rank, a bit of context, a “place in the world” for every Tweet as it relates to every other Tweet and to every account on Twitter, well, it can do the same job for every possible advertiser on the planet, as they relate to those Tweets, those accounts, and whatever messaging the advertiser might have to offer. In short, if Twitter can solve its signal to noise problem, it will also solve its revenue scale problem. It will have built an auction driven marketplace where advertisers can bid across those hundreds of millions of tweets for the the right to position relevant messaging in real time.
I still think this is a huge opportunity for Twitter, and not for revenue reasons. I get a ton of value out of the Twitter platform, but I don’t turn to it for news and happenings anymore. I follow too many people, and managing multiple screens on Tweetdeck is just too much work. Instead, I depend on great curators like Jason Hirschorn and his team at MediaReDEF – essentially the morning newspaper for folks like me – and a number of machine-driven services that consume my feed and spit back the most popular shared stories (News.me, Percolate, etc).
I find the machine services are predictable, but Jason’s service is top notch – he’s an Editor’s Editor. His stuff, along with folks like Dave Pell, have become my go to these days. But Twitter can’t get the mass market users on its system via human curation – or can it?
Back when Twitter was small and the signal was high, I found a lot of value in my Twitter feed. Individuals who were great curators were my favorite follow. Over time my feed clogged with too many other types of folks – and I’ve never found a tool that can help me get back to those halcyon days where the best stuff rose to the top. Twitter’s Discover tab is interesting, but lacks instrumentation. Wouldn’t it be cool if Twitter somehow elevated the best curators on its platform in some way – promoting their work and helping them gain audience? Sure, it’d feel a lot like the old “who to follow” of the old days (and there was much to criticize with that system), but given how much Twitter now knows about its own platform, it might be a pretty powerful half-step toward giving people a better handle on the richness the platform has to offer. It’d be a great, lightweight way to start using the service, and for power users who have bankrupted their feeds (IE, me), it could really change the game.
I’d love a service on Twitter that pointed out the best curators for any given topic where I’ve indicated a strong interest (and my interests have already been mapped by Twitter, for purposes of promoted tweets). Further – and this is important – I’d love for Twitter to break out those feeds for me as part of its core service – a sort of Headline News to its constant 24-Hour barrage. It’d mean a break with the one-size-fits-all mentality of the main Twitter stream, but I think such a break is overdue.
Chances are, Twitter’s already explored and dismissed these ideas, but…are they crazy?
I remember the days when people were complaining that all they could see was Robert Scoble’s great curation on Twitter.
Twitter aimed for the stars & brands. The majority of curators were never considered for verified accounts or for Twitter’s onboarding process
Whilst I have returned from a 2 year hiatus to Twitter, in all honestly that hiatus has been 4 years.
Some of the advertising aspects with retargeting intrigue me, especially things like permission (taken from an email context) and how we will manage “do not mail” across multiple campaigns.
I think there’s about to be a lot of innovation on the product/content side at Twitter. It’s certainly the focus.
Further – and this is important – I’d love for Twitter to break out those feeds for me as part of its core service – a sort of Headline News to its constant 24-Hour barrage. It’d mean a break with the one-size-fits-all mentality of the main Twitter stream, but I think such a break is overdue. http://qr.net/rtcX
Managing twitter noise is spot on…scaling is always an issue for any successful business and managing (organizing) information traffic is a growing concern. Does Twitter really have a clear vision for its platform? I’m really not clear on where they are going, seems experimental at the moment…
Nonetheless, I still believe that Twitter is arguably the most valuable user platform in social; staying focused and relevant on product engagement is the most valuable lane for them, they should avoid trying to stitch-in unnatural revenue formats.
Twitter has been supporting contextual ads for almost a year now:
https://blog.twitter.com/2013/introducing-keyword-targeting-in-timelines
I utterly missed that. Keyword targeting is a good first step – ned tomes more semantic work. I’ll update the post.
They also support some simple semantic analysis features (e.g., sentiment targeting)
https://blog.twitter.com/2013/introducing-broad-match-for-keyword-targeting
That’ll teach me to write late on a Friday> Thanks.
Great post John. I’m of the mind that Twitter should stick to doing what it has done best though – and keep the whole experience simple. I stopped using FB when it became a living hell of rubbish in my timeline and I don’t get that in Twitter – I just get too much. I just signed up to Cronycle – I think they’ve nailed this problem. It’s a paid service but they seem to have got the balance right between me filtering my own streams into useful sets and then being able to curate and share those sets as you mention here. I’d rather use a service like this for the heavy lifting and keep Twitter as simple as it can be. Just my two cents!
Great observations, as usual.
I want the old relevance back in my Twitter stream, too. Honestly, I don’t know how anyone uses Twitter today without good search tools and well-groomed lists.
Twitter can do so much more for their users, for society and for their own business if they did a better job sifting relevance from noise.
For example, they could incorporate signals of intent into their platform – and not just for advertising, a la AdWords.
I know it’s dangerous to debate the meaning of ‘relevance’ with John Battelle, but I define a tweet’s relevance as:
the right content
+ the right timing
+ matches my context
+ meets an immediate need (intent) of mine (if it exists)
That last part – intent – includes stuff like me asking a question about a book, complaining about AT&T or saying I want to buy a BMW. Intent signals also include discussions of personal life events like a job change, a new baby, a pending divorce or moving to a new city.
As you well know, intent is a different kind of signal than the topics and people I’m interested in (my interest graph).
So far, social networks have focused mainly on interest-based relevance (keyword-based targeting), leaving intent signals to the cookie lords (web tracking companies). But there’s a lot of intent expressed in social media, too.
I know this because my company mines Twitter for signals of intent, today.
Signals of intent aren’t in every tweet, of course. But when they do exist, they are often the most important contextual signal a person can give.
It’s still early days, but I think the kind of intent-mining we do at NeedTagger will eventually make it into Twitter – especially into their advertising.
FWIW, I know the team at Twitter understands your points and mine. They just haven’t incorporated them into the product we see today. Yet.
Let’s hope they’re on it right now 😉
I am sure they have thought of everything we have and more! The hard question is understanding which signal(s) are worth investment now, because signals tend to change over time, and until they tip, it’s hard to say there’ll be enough of them – and a good approach internally – to validate the opportunity cost of missing other signals.