Fix Tumblr

March 19th, 2011

For several months now, Tumblr has apparently been a victim of its own success. What used to be known as a super-popular, well-designed, streamlined blogging service with lots of internet-famous blogs hosted on it, is increasingly known as all that and also as a flakey, unreliable service. Folks who host their blogs on Tumblr are up-in-arms about the frustrating error-proneness and downtime of the service.

My primary product, MarsEdit, is a desktop blogging client that interfaces with Tumblr, among dozens of other types of blogs. The reduced reliability of Tumblr, and in particular of its API, has meant a deeply compromised experience for our customers in common. This means that I also suffer some pain in the midst of Tumblr’s flakiness, because I have to support an unreliable service and explain to customers that MarsEdit is affected by Tumblr downtimes as well.

Some of Tumblr’s greatest assets are the deeply respected bloggers who trust Tumblr to host their writing. They serve as implicit spokespeople for the service each time they publish an entry, and more explicitly so when participating in Tumblr’s own social network, raving about the site on Twitter, at conferences, etc.

As flakiness continues, the tone of endorsements is turning negative. I regularly see sarcastic snipes against the service in my Twitter feed, and even on blogs that are hosted by Tumblr itself. Garrett Murray’s frustration peaked yesterday when he posted a sarcastic revision to Tumblr’s increasingly famous “downtime” graphic. Steven Frank of Panic fame chimed in today on Twitter:

It occurs to me that Tumblr is also growing exponentially with no apparent income source. I should look for a new home, pre-dickbar.

Tumblr has a problem. Since late 2010 and for all of 2011 they have been suffering enough downtime and flakiness that a growing chorus of users is lambasting the service. Without judging whether that’s fair or justified, let’s accept that what used to be a widely lauded service is becoming a widely criticized one.

But how big of a problem is it if, as Steven Frank suggested in his tweet, the service continues to grow its membership by leaps and bounds? My theory is Tumblr’s continued success in signing up new customers is both thanks to and at the expense of their influential early adopters. Folks who helped to build Tumblr’s reputation over the past several years are now suffering, presumably because of Tumblr’s ravenous ingestion of new users. If this keeps up, the influential “power-bloggers” will quit Tumblr and move on to more reliable services. Tumblr will be left with millions of users, who I’m sure are perfectly nice people, but who don’t exert as great an influence in the web world.

What should Tumblr do? If the failure to rein in performance and uptime issues is connected to success in signing up new customers, then they should stop signing up new customers. Sound foolish? In Tumblr’s position I would do whatever it takes to bring back the level of service that users enjoyed before the “great downtime of 2010.” Happy, influential customers paved the way for Tumblr’s success, and bringing back that enthusiasm is the best way to perpetuate success far into the future.

If Tumblr turned off new user registrations today and added a “notify me when more new users are being accepted” sign-up form, it would provide breathing room to focus on fixing the experience for current customers. Framed correctly, it would also make those customers feel cared for and important, something they probably aren’t feeling so much at the moment. Yes, for prospective customers it would be a slap in the face. Nobody wants to feel shut out. But if given a choice, protect your existing, not future customers. Web services build buzz all the time with limited, invitation-only beta testing intros. It would be a step backwards for Tumblr, but it would also re-establish a sense of exclusivity that would pay dividends after issues are resolved and open enrollment returns.

It’s easy to armchair-quarterback another business when you don’t know any of the details. I’m sure the challenges at Tumblr are diverse and hard to pin down to my convenient diagnosis of “too many users.” But if you’re bailing out a sinking boat, the first thing to do is stop admitting new passengers.

I hope Tumblr figures out a way to solve this for the long run. My customers depend on it. Their customers depend on it. And the longevity of the company itself depends on it.

8 Responses to “Fix Tumblr”

  1. Jane Quigley Says:

    I know I sound naive – but I’d like to know what the VC’s accountability is. They hand over millions to a very young entrepreneur and then seemingly walk away.

    A very sharp contrast to Tumblr/David Karp is Squarespace – Anthony Casalena had great mentors around him that seem to have taught him a lot. Watching him speak in public now is much different than a few years ago, when he was very shy (just like Zuckenberg). Anthony has a strong vision and business model and I feel that all the money he now takes goes to the right places. Squarespace’s customers come first.

    This is a HUGE difference between a free service and a paid service – the more you rely upon $$$ from your customers, the more important it is to keep them happy.

  2. Brandon Leedy Says:

    I couldn’t agree with you more. I actually became a user of MarsEdit specifically _because_ Tumblr would go down as I was writing a post. I would click “post” only to find that Tumblr was down and then lose the post I had been writing, with no way to even save as a draft! MarsEdit has solved that problem and much more.

    To be honest, if Tumblr doesn’t gain some more stability soon, I’m with Steven in looking for a new CMS. As a design student about to graduate, I’m building my site and portfolio that I’ll be using to get a job. I need a stable CMS to host my blog and work. It just looks bad to send your potential employer to a dead site. Also, I’m not invested monetarily in the service, even though, like Garrett, I’d like to be. In fact, the only investment in blogging that I have made is in MarsEdit, and that can follow me to any CMS. I normally hate being apart of the somewhat fickle flock of web-service users, but the downtimes are really starting to become ridiculous. For me, it’s not a matter of putting up with “scaling downtime” (like in the early days of twitter) this is repeated downtime without any real sign of improvement. At the end of the day, I’m not a blogger who drives even a moderate amount of traffic, so my thoughts might seem like complaining… but I am a user nonetheless and I am finding it hard to stick around.

  3. Scott Says:

    Just a thought on how to work around an unreliable service such as Tublr: Clicking “post” will initially attempt to connect via the appropriate API. If a connection failure is detected, then the post is added to a background queue which will attempt to post at some reasonable pre-defined interval. A dialogue should appear indicating the failed connection, allowing the user to “Keep trying to post in the background” or “Manually try again later.” A non-modal notification of success when it finally goes through should occur.

    Using the correct language in the notifications to the user might serve to deflect some of the support away from you and to whichever service is having issues.

  4. Sam Says:

    An alternative to closing registration is scaling back on features and turning off parts of the site that are the biggest resource issues.

    BTW, your comment form layout is broken on Google Chrome for Mac :)

  5. charles Says:

    I don’t follow the Tumblr ups and downs very much, but it seems they are in trouble if they don’t solve the downtime issues soon enough. It’s probably telling though that you call the users “customers”, since they don’t pay… Compare that to the Pinboard recent story ;-)

  6. Justin Williams Says:

    Scaling back registrations is exactly what they won’t do sadly because growing their subscriber base numbers is what is important for acquisition. Having 10 million Tumblr accounts is more important than having 6 million happy ones.

    Tumblr is Geocities with Ajax and hipsters.

  7. Matthew Brown Says:

    Also; scaling back registrations is what LiveJournal did, and was probably key in starting that service’s decline.

    i’m increasingly of the belief that free services are worth less than I pay for them. Paid services get more money if they get more popular, and economies of scale kick in. Free services, not so much; it may not bring much of an ad money boost, and the ad sales companies don’t give you your money for months.

  8. km0i Says:

    i used MarsEdit to post blog in but the image wont show in facebook when share it from bloggers weblog…

    the image will only show in facebook when the post is from the blogger editor…

Comments are Closed.

Follow the Conversation

Stay up-to-date by subscribing to the Comments RSS Feed for this entry.