A Lens on What Google is (and is not) Good At

There’s a long essay I wrote internally at google that cites things I can’t share publicly. Here’s a heavily redacted summary:

First, some jargon:

• Capability: a raw technological ability to perform a task. For example: an ML model that can estimate the cost of car repair from a photo of the crash.

• Product: wrapping that capability in a tool that fulfills a customer’s need. For example: shrinking the ML model from before and deploying it inside a polished mobile app where you can take photos of a crash that you are in and the app tells you how much your repair will cost.

• Business: providing that app to enough customers so that you build a proprietary dataset of crash data and photos, offering car insurance, and paying out the cost of repair immediately, and with this newfound ability offer a lower rate because you can better predict accident costs.

Businesses contain one or more products. Products contain one or more capabilities.

So What?

This table explains most of where google’s successes and struggles come from:

It’s a simplification, so of course you can likely come up with a counter example here or there.

Stating explicit examples is impolitic. I recommend Googling how many of google’s major successes had products that existed before and got acquired. Google scaled them out or invented capabilities to scale the products and businesses.

There are also strong implications as to whether Google should invent new products, or instead focus on scaling capabilities and look to acquire new products as they grow.