Why the Facebook Innovation Machine Doesn't Work on Mobile Devices

On mobile, there's less data to work with, and the result is a product that only feels right to Facebook's employees.


Facebook's future depends on succeeding on mobile phones, as questions about Facebook Home at the company's shareholder meeting once again made clear. Whatever Facebook says about the product now, they sold Home hard, and it has not lived up to its billing.

There's a simple, almost mechanical reason that it's hard for Facebook to be as good and user-satisfying on the phone as it is on the desktop: feedback.

On the desktop, Facebook is a machine designed to make itself better. They hoover up data about how you're using every piece of the service. They A/B (or A/X) test every single part of the user experience. Based on all that feedback, they tweak and tweak and tweak. Then, they do it all over again. Disputes can be settled by simply testing alternatives on a small base of users: may the best data win. And they can do all this quickly because on the web, Facebook can change the code any time they damn well please (although mostly once a week on Tuesdays).

Now, think about the mobile world. Apple and Google control distribution. You have to send your app to them and get it approved. You can't send one app to 0.1 percent of your users and then another app to the other 99.9. You can't iterate on your own schedule. In short, they know less and can do less.

Put it all together: the Facebook innovation machine, the most successful engagement machine the web has ever known, does not work in mobile. Some of the company's key success factors are just missing or not possible within the current corporate and technological constraints. And the numbers show that: on the Google Play store, Instagram rates a 4.6, Twitter rates a 4.0. Facebook's apps: the standard mobile app and Home, come in at 3.6 and 3.0. It's not that Facebook is making bad apps, only that their apps are not -- as the website has been -- dominant.

Facebook has to go by feel now. Before a big product release like Home, they mostly can only test on their own employees. And that's a problem. Facebook employees are essentially required to live on Facebook. Unlike many normal people, who spread their presence across many networks -- Pinterest, Tumblr, Twitter, LinkedIn, G+ -- Facebook's employees tend to concentrate their usage within Facebook. When they try something like Facebook Home out, it may be great for them because they mostly want access to Facebook.

I wonder if Facebook employees have actually done *too much* dogfooding, a artful term that means using one's own product. Just to keep the metaphor going: they've developed a palate that doesn't reflect what normal canines might want.

On the web, that didn't matter as much, intuitions could be tested easily and robustly. On mobile, what feels right to Facebook's designers and engineers does matter because there's less data to win arguments with.

For me, that's the best argument for the Instagram acquisition. Kevin Systrom and his team have a great feel for what people want to do on their phones. Can they successfully transfer that phronesis, that "practical wisdom," to Facebook? And if they do, will they be able to retain their independence?