February 27, 2011

Tumblr’s Future

Some Suggestions on How to Handle the Transition from Free to “Freemium”

Since my earlier posts on niggles I had with Tumblr, I’ve thought about the topic a lot — specifically, how Tumblr can handle the inevitable transition from a free to a paid/freemium model. (This entire post is based upon the assumption that such a transition is inevitable, which I think is the only logical conclusion to reach. Tumblris, after all, a business.)

Tumblr actually has a bit of a problem here: a lot if what people would have paid for — custom domains and video hosting, for example — are already for free. As such, I’m not sure they can monetize these things; it would take a very brave company to start charging for them once users have become used to getting these features for free, and there would probably be a lot of defections to Posterous if they did.

The way to avoid this is to introduce new features, targeted squarely at the geeks who’ll actually pay for a blog, and charge for them. Below, I’ve compiled a list of ideas that could become pay-for features, the reasoning behind why people would want them, and why they’re worth implementing. It should be noted that these are not in any particular order, so when you read this, don’t assume that I consider photo slideshows more important than guaranteed uptime or better structure.

Guaranteed Uptime

This one is important. Tumblr’s history with regard to uptime is middling at best. Even ignoring last December’s massive server meltdown — which, to be fair, was a pretty significant amount of egg-on-face for the platform — Tumblr is slightly wobbly at best. It goes down, very briefly, at random intervals, and gives no indication of when it will be back. (“We’ll Be Back Shortly” is friendly, but unhelpfully non-specific.)

A platform of Tumblr’s size, complexity, and current rate of growth is almost inevitably going to have these problems, and it’s something that needs to be accepted as a fact when using the service. As it’s free, it’s not as if you’re not receiving a service you paid for; you’re not receiving a service you haven’t paid for, which is a small but important distinction.

When you do pay for it, though, the situation changes: it becomes much more important that your blog is always available, and that, in the event of unavoidable failure, visitors are given some kind of idea of when to expect it back. Moving Tumblr’s paying users onto stronger servers, or ones with more redundancy, and giving them an accompanying guarantee of uptime, reassures them that their money isn’t being wasted and that their site is safe in your hands.

High-Resolution Photo Uploads

This one is a no-brainer, really. As it stands, Tumblr’s “High Res” photos actually top out at 1280 pixels wide; upload a 5 MP photo with a standard 3:2 ratio, and it gets resized to 1280×851 image. That’s a higher resolution than the average webcam delivers (usually 640×480 or thereabouts), but its disingenuous to call that high resolution — it’s a medium resolution at best. Fortunately, it’s easily fixed: just allow images of a higher resolution. The cost-per-image of serving them is higher, but the feature is only for paying users, so they’re covering the increased cost incurred. Serving up a 1280×851 image for a free user makes you a loss ever time; serving up a 3008×2000 image for a paying user doesn’t.

Better (Read: Non-Flash) Slideshows

As it stands, Posterous is vastly better than Tumblr in this regard. Upload or email a couple of photos to Posterous and it automatically creates an HTML slideshow. Upload multiple images to Tumblr, and you’re stuck with a Flash slideshow, which is fine… if you have Flash installed on your computer.1 If you don’t, it’s totally useless. Granted, most people do have Flash installed, but having the option of creating HTML slideshows would be the best of both: those who don’t have an ideological or technical dislike of Flash could leave as-is; those who do could set their blog to use the HTML slideshow option. That way it’s an invisible change, and you take care of the technical issue for those who care about it without impacting those who don’t.

HTML 5 Video and Audio

This is tied into the issue above, and I know it’s a technical and ideological minefield. Even so, it is worth implementing, and with even the most primitive of browser-sniffing technologies, it should be seamless to the reader: those with H.264-compatible browsers get served their video using the HTML 5 <video> tag; those who don’t get the Flash version. (There is, obviously, the issue of transcoding video, but I suspect Tumblr’s existing Flash videos are H.264 served in a Flash container. Even if they’re not, they are still transcoding videos into H.264 for mobile visitors, so transcoding overhead isn’t a good excuse not to implement it.) And, again, if you implement it as an explicit opt-in like I proposed with the slideshows, it’s transparent to the large number of people who don’t care about the issue.

Limited (Or Not) File-Hosting

This one is also pretty simple: let people host smallish files on Tumblr’s servers, appearing to come from their domain if they have one, or their Tumblr subdomain if they don’t. Restricting the feature to paying users only has two benefits: it almost completely eliminates the chances of people attempting to use it Tumblr as a host for illegal files, and it’s a significant “value-added” feature for some users. (It also means people are paying for a service that adds more strain to Tumblr’s service.)

With on-site file-hosting, those who develop Safari Extensions, for example, don’t need to resort to using their iDisk, or Sourceforge, or Dropbox, or [insert preferred filehost here] — they simply upload the file to Tumblr’s servers and can serve up the serve them from their own site. (The same also applies to those who archive the inane writings of cybernetic yogi messiahs).

Proper Structure

Tumblr excels at immediacy, but falls flat on history: its archive pages are visually impressive but not terribly useful — geared more to image and video than text. Your only other option is to wade through pages of posts to try and find what you need.2

Give those who want it the option of structuring their blogs better: let them have sidebars with posts by month, and categories, and a more obvious method of looking for tags within only their own blog. Make it easier for me to set up my blog to let my readers explore it properly. Proper structure is maybe anathema to Tumblr’s hip, “microblogging” philosophy, but it is very important to some of its users. Without more control over the structure of your blog, Tumblr won’t be the platform of choice for really serious bloggers; they’ll continue to use Moveable Type and Wordpress.

Granular Control of Notes

I’m ambivalent about Notes. One one hand, I like them when they’re used for conversation — they handily link together a chain or posts that directly respond to one another (see my posts of Twitter for Mac’s UI overhaul). As for “Likes”, I’m pretty disinterested in those. Sure, it’s nice to know that you posted something and a bunch of people like it, but, really, it’s more noise than a genuine contribution to the discussion.

Most people probably want Likes to show on their blog, or at least don’t mind having them there, so removing them totally isn’t a realistic option. (They’re also quite an important part of Tumblr’s identity as a platform, so I can imagine that would be another reason not to remove them.) The solution would seem to be offering granularity of control: let those who care about it select what kind of Notes show up on their blog. I can live without the ego boost, so I’d leave Reblogs visible, but turn off Likes.

Innovate More

Tumblr seriously trails Posterous in its rate of feature addition. Whilst there’s a lot to be said for not adding the kitchen sink (and it’s something of a personal philosophy for Tumblr’s former CTO Marco Arment), Tumblr sometimes feels like it’s resting on its laurels. Not everything that Posterous adds is necessarily something that all, or even most, of its users want, but what counts is that they’re constantly searching for ways to improve their platform.

By contrast, Tumblr’s Staff blog feels like a self-indulgent celebration of their own success, and leaves one with the feeling that Tumblr’s staff are more focussed on profiling users than relentlessly improving their platform. Whether that’s true or not isn’t really the point: it’s the feeling I get, and it’s shared by others that I know who use Tumblr. (Unfollowing the Staff blog is one of the first things they and I did.)

Even if you follow Google’s model and launch new features as beta, restricting them to customers who specifically opt in, it’s still worth doing. Anyone who opts in will be happy to have new facilities to play with, and — since they’re launched in beta — will be more understanding if, in the cold light of day, they don’t make the cut.

Track- and Pingbacks

Tumblr posts integrate beautifully with other Tumblr posts, but when it comes to integrating outside blogs, there’s absolutely nothing. The platform doesn’t support track- or pingbacks, and any conversation you may try to strike up on a topic is effectively restricted to being within Tumblr’s own network, unless the poster to which you’re replying to subscribes to your feed. This problem could easily be fixed by implementing either trackbacks, pingbacks, or both.

It appears that there are ways you can do this yourself: for the bloody-minded, there’s Mathematical Poetics’s manual trackback method. For some reason, it doesn’t display correctly on his site, but looking at the page source reveals his simplified example command: curl -d url="postURL" trackbackURL. You have to paste or type that into Terminal as a single line, but otherwise, it’s relatively straight-forward to use. (I’ve tried it, but it doesn’t seem to have worked.)

For the more technically intrepid, there’s Evidence of Control’s Perl script for Tumblr track-, link-, and refbacks. I couldn’t manage to work out how to use it, so can’t verify whether or not it actually works.

Regardless of the efficacy of either of these two options, they’re both much, much less convenient than having baked-in support, and they don’t change the fact that, for everyone else, there’s nothing.

À La Carte

Not everyone wants to eat a whole pie, so sometimes it’s best to sell it in slices. With this in mind, perhaps the best idea would be an à la carte billing system, letting people build their own package from an array of options. HD video, for example, is not something I’m terribly interested in, so I’d rather not pay for it; however, I would pay for custom permalinks (goodbye /post/1234567890/!), limited file-hosting, control of my Notes, and better structure.

Set the price relatively low: $2.99 per month for four or five advanced features, 50¢ per feature after that. Pricing it low makes the margin lower, but also makes the uptake higher: $2.99 is impulse-buy (AKA “coffee money”) for most people, so there’s little incentive not to sign up. What you may lose in margin, you more than compensate for in volume, and most blogs (this one included) will cost a small fraction of that $2.99 to host each month.

Implement the things geeks care about, and then let them pick and mix. To the best of my knowledge, noöne else has ever done this, but it seems like the perfect idea: you only pay for what you actually want and use, users are happy, and Tumblr gets a revenue stream.

Final Thoughts

A lot of the above is pretty brutal in terms of criticising Tumblr as both a platform and an organisation; in a lot of cases, I’m probably harder on them than they deserve. It’s easy to be an armchair CEO and tell people that what you want is what they should be doing. Of course, there is value in doing so, and — as I said in “Addendum to (Dis)orientation” — and I genuinely believe that improvement comes only through listening to the feedback of your users.

The fact remains that, in spite of all the rough edges and non-ideal aspects of Tumblr, I continue to use it as my chosen blogging platform. I do so because, 99% of the time, I can either live with its shortcomings or work around them; the other 1% of the time, I can accept them. But, in order to convert me from freeloader to customer, Tumblr needs to make changes, and needs to improve, and most of all, they need to give me a reason to pay. The contents of this post are just some ways in which I think they can do this. Whether they’re technically feasible, or cost-effective, I can only guess.


  1. Which, inevitably, I don’t. The performance improvement in day-to-day browsing, even compared to when I used ClickToFlash, makes the slight inconvenience worthwhile, but a story for another post. 

  2. The times I’ve used it, I’ve found Tumblr’s search function to be woefully broken: searching for the term fringe sometimes returns two results, sometimes none — even though I have three posts on Fringe’s woeful typography