Google's reputation for discontinuing products is unmatched. There's even a site dedicated to it (KilledByGoogle). We all loved Google Reader, but it shut down almost ten years ago. Experimentation is at the core of Google's DNA, and we should expect experiments to fail, products to evolve, and features to change.
Yet, Google can't shake this meme, and it's starting to do real damage to users and the Google brand. So why can't Google shake this problem?
- Consumer strategy collateral damage in enterprise sales. Letting a thousand flowers bloom works well in consumer applications, but a short (or nonexistent) deprecation policy for enterprise products doesn't work. Google Cloud might have a reasonable deprecation policy, but it might not matter if the meme is strong enough.
- Corporate dysfunction – e.g., messaging applications. Lack of a coherent overall strategy, misalignment with the internal organization (does it belong in Gsuite? Consumer apps? Android?). Conway's law ("shipping your org chart") applies.
- The moat for search is already strong enough – Commoditizing your complement is a strategy that Google uses to ensure that Google Search is protected. Any roads that lead to search (browser, mail, maps, mobile, docs) must be commoditized and owned by Google. My uneducated guess is that many consumer products (like Reader) don't sufficiently move the needle on search. As a result, finding the right moat-builders is becoming harder and harder.
- The transition to a more financially focused company – when Ruth Porat joined Google in 2015, she brought a culture of cost-cutting and financial discipline. It's no coincidence that this lined up with co-founder Larry Page moving on from CEO. This meant highly used products that weren't defending the moat or monetizing were forced to monetize or shut down.
- Experiments can't work at Google-scale. A/B testing works better at scale, but full-on product experimentation is worse. Google experiments acquire millions of users, regardless of if they are good, bad, successful, or not. Startups can fail because they aren't at scale and suffer minor reputational damage (when they fail, they usually die).
What should Google do? Launch projects under different brands (Alphabet). Separate enterprise from consumer more clearly (Gsuite/Google Cloud). It would be the wrong idea to stop experimenting completely – that's how companies lose their advantage.
Disclaimer: I used to work at Google but have no insider knowledge of the answer to this question.