The sad demise of Twitter counts means tough decisions for us
- Summary:
- Twitter counts have gone and that's left us with a tough decision. We've turned off the public display of social shares because the lack of Twitter counts means our display is misleading. We can't allow that.
Today is a sad day. We've lost the means to count Twitter shares. On 22nd September, Twitter announced, among other things, that:
The Tweet button has displayed share count over the last five years by querying a JSON endpoint hosted on various domains. These private JSON endpoints have been used by third-party developers over the years to retrieve a simple share count of any URL. These endpoints will be shut down next month when the Tweet button removes its share count feature. The Twitter REST API's search endpoints1.6k are the best way to gather ad-hoc information about a URL shared on Twitter.
That announcement got plenty of play with developers almost universally expressing varying degrees of being pissed off. The follow up post on October 6th explained the technical reasons for the change, arguing that:
Additionally, the “count API” has never existed as part of our public, supported and documented API endpoints; it was only intended for use by our own web widgets. We’ve often cautioned in our developer forums that use of such undocumented endpoints shouldn’t be relied upon, as we cannot commit to supporting them.
Regardless, many services were taking advantage of those end points in order to serve up Twitter share numbers. Those just got switched off as advertised. Or rather they were switched off on 20th November although it seems that some services saw their access to the endpoint fade away over a matter of days. There are, as always, two ways to look at this:
- Twitter certainly gave developers plenty of notice this was going to happen and offered an alternative approach. Developers however were far from certain the proposed access method will work as intended.
- The technical reasons seem reasonable but the fact Twitter has made no commitment to supporting such a feature in the future flies in the face of de facto counts that can be obtained for other popular services.
Now - in fairness to Twitter, views are divided about whether these counts are useful or otherwise. I tend to the view that they are and here's why.
Social sharing count numbers across different network platforms is a useful metric in our case because we see our content getting amplified in other places. Having that number gives us a reasonable indication of where content is being amplified which in turn means we can focus attention on those networks when considering engagement.
We have multiple authors and each author makes a commitment to amplify selected stories to selected networks. We leave the final decision to them but there is no doubting that a carefully worded and timed Tweet has a distinct impact on readership. That's important because we want to be sure that content reaches people at the moment when they are most likely to be interested. In short, we are applying some science to what we do and the priorities we set. Having the numbers both before and after we make those decisions is useful because we can then determine what additional action is needed.
We also know that different content appeals to different networks but the extent to which that matters varies enormously. Right now, it is very hard to predict what will fly and what won't on a piece by piece basis although we have good insights into topic areas that resonate at given points in time.
When you take all that together, having those basic count numbers matters and the fact that services we use have lost access means that any display of an aggregate social sharing count is now misleading.
We are looking for a solution to this and are working with service providers to see what the official REST API can do to overcome the problem.
In the meantime, I made the decision to keep the AddThis sharing buttons (which are a primary metric source but not the only one) on the site, including the Twitter share button. But in the short term at least, we will not be displaying a social share count. Unlike some other services which seem to be blithely ignoring the problem, I'd rather not show something that is obviously incomplete than show a misleading figure.
More generally, I am disappointed that the services we use have not attempted to find a workaround. They have failed to understand that the metrics which they were supplying as part of the service for which we use them matter. Equally, I am disappointed that Twitter seems to have ignored the many calls for them to continue this element of their service or, at least, chosen to ignore for the time being. Thanking developers for feedback is hollow when an alternative isn't in sight or a better explanation isn't on offer.
Here's the rub. There are many people out there thinking Twitter is on the wane and see this kind of thing as another example of how Twitter is out of touch with those who use its service. To me, Twitter has always been something of an anomaly. It started off as a badly engineered skunkworks that suddenly found itself in the limelight and which accidentally became popular but which has never really grown up. Don't get me wrong. Twitter performs a very useful service and has a lot of super smart people. But its inability to imagine a monetization strategy outside of a limp advertising model has always struck me as the inevitable result of a management that is one short of a six pack.
I said back in 2008 that Twitter's chances of proving successful as a business were limited and that an early pivot to a paid subscription model was both logical and sensible. Of course that was at a time when ad supported models were all the rage and where VC funding was pouring into "grow and then monetize" strategies, almost all of which assumed that software would be free sand that advertising would be the supporting pillar. Twitter plowed on with what seemed like a remarkably generous but ill conceived developer approach that in essence allowed a thousand flowers to bloom. The adjacent services made money off Twitter's free back while Twitter languished with its own self inflicted problems. In the meantime, the superior Facebook experience trumped everyone and services like Twitter have been left scratching their heads ever since.
Technical reasons side, this latest move seems to be as much a reaction to what happened in later years as Twitter became less and less developer friendly as it sought ways to pull revenue towards itself.
In short, Twitter has created a situation that makes it extraordinarily difficult to imagine how it can adequately turn a profit without doing what it's doing now - hacking away at the cost base. In that sense, Twitter counts look like a casualty of that rationale. I don't expect Twitter to make any sort of admission in that regard and Andy Piper, a technical evangelist I have known for years, and who has to try communicate the issue, seems hamstrung into simply repeating what has been said in the most recent past. It is a very tough spot top occupy at the moment.
As always in these circumstances, we and others will find ways to work around the problem but it won't happen today, or tomorrow. But at least you know our thinking and rationale behind our decision.