Googlers love to discuss and debate things within the company. As a rule of thumb, the internal discussion is civil and respectful, but can be passionate. I may take a few of my favorite internal posts that I wrote, tweak them a bit, and publish them here just so I can refer to them more easily down the road.
Here’s something I wrote in 2012:
Recently Google has been shooting for more “magical” moments, and that’s a noble goal. But I think our first priority has to be “dial tone” moments. On a basic level, that can be as simple as uptime and reliability–if Google Voice or Google Music or some other product is too flaky, people won’t use it.
But we should also strive for dial tone moments in terms of consistency or utility or latency. Here’s a simple example. Saturday night in Mountain View I wanted to check how hot it would get on Sunday. I got three different answers from Google. Google Now claimed the high would be 88 degrees. Doing a mobile search for [weather mountain view] predicted a high of 82 degrees. And the News & Weather widget predicted a high of 81. I got those results all within a minute of each other.
If we can’t tell the user whether it’s going to be closer to 80 or closer to 90 tomorrow, how can we expect users to trust us for magic moments?
I think a lot of Google’s reputation or “brand” comes from being a useful, functional tool. Magical moments are fine (great even), but overall the most important thing Google needs is to nail “dial tone” moments, where people just assume that Google will always be up, always be fast, and always get them exactly what they need. If we can do better, great, but we have to nail those dial tone moments.
At Charlie’s, the main cafe at Google’s Mountain View headquarters, there’s a door that uses an electric eye to automatically open as you walk up. When the automatic door is working, it’s a smoother experience than walking 10-15 extra feet and pulling open the regular doors yourself.
Unfortunately, the automatic door seems to be broken about 5-10% of the time. Either someone forgets to unlock it when the cafe opens up, or it’s broken, or it takes too long to open. Then you’re left standing and feeling silly waiting for a door that might or might not open. I noticed that over time, lots of people stopped using that door and went straight to the regular doors. It took 3-4 extra seconds to walk over to the regular doors, but they always consistently worked. The automatic door was optimizing for magical moments when it needed to optimize for dial tone moments. The automatic door had to walk before it could run.
What does this mean for you or your company? Look for dial tone moments or rough edges where you could improve. I had one colleague at Google who I swear could walk from his car to his office and find six different things that needed to be improved. There’s stuff all around you that could be much better if you just lower your annoyance threshold enough to notice it.
Just as an example, I’ll pick on tivo.com because I love TiVo. One annoyance: when you tell tivo.com to keep you signed in for 45 days, and then come back a week later, you have to sign in again. Another annoyance: you can only search for shows happening within the next two weeks or so.
I would consider the first annoyance to be a dial tone issue: it shouldn’t be hard to stay signed in to a website. That’s a very fixable issue. In fact, when I come to tivo.com, the message at the top knows my name even while another part of the page is telling me that I have to sign in:

The second annoyance–only searching two weeks out–might easily be a multi-month project or even impossible, depending on how and where TiVo gets their data from. But you’d get almost as much goodwill from fixing the easy, dial tone issue as you would from fixing the really hard problem.
If you’re not using (“dogfooding”) your product every day and looking daily for annoyances, snags, or rough spots, you’re missing out. If you run a website, pay attention to the uptime and speed of your site. Monitor your vital metrics and trigger an alert if something goes seriously out of whack. Your product should probably be reliable before you shoot for flashiness.