It is tempting to search for the one magic move that will make everything better. A new VP of Sales. A new database layer in your tech stack. A new brand for your company. Moving everything to the cloud. More capital in the business.
But it is rarely one thing that a business needs to succeed. It is often a little bit of everything.
Back in the early days of Twitter, we could not keep the website and API up. We would hire advisors and they would recommend something new and we would try it and we would still go down. It was terribly frustrating and threatened the business.
During this period of instability, Twitter purchased a search engine called Summize. Summize was a small team of engineers, most of whom had come out of AOL.
After we cut the deal to acquire Summize, I asked Jack Dorsey, who was running Twitter at the time, how we planned to integrate the Summize team. He looked at me and said “we are not going to integrate them, they are going to integrate us.” And Jack made Greg Pass, Summize’s engineering leader, Twitter’s engineering leader.
It was interesting to watch Greg and the Summize team tackle the “fail whale.” Instead of searching for a magic solution, they instrumented the entire system and just started rebuilding every part that was about to break.
It was a slow and steady approach. It took time. But within six months (or thereabouts), we had a much more stable system. And after about a year of this approach, we had mostly said goodbye to the fail whale.
Grinding isn’t very satisfying. It is hard to stand up in front of everyone and say “we are going to fix things around here bit by bit with a lot of hard work.” Big flashy moves are an easier sell most of the time. But they don’t work nearly as well and are prone to complete and abject failure.
If given a choice between a flashy operator or a grinder, I will take a grinder every time. It is a much higher percentage bet. It requires faith and patience and the results are sometimes hard to see. But if you look at the results from grinding it out over a long enough time frame, you can see the power of that approach.
