The metric you need to keep in mind here is revenue per employee. Keeping one product up and running necessarily means withholding engineers from another. As an extreme example, ads engineers make Google many times their salary, benefits, and ephemeral costs, and even non-revenue-generating services like Photos and Assistant indirectly drive traffic to the money makers.

Meanwhile, consider Reader. A product with only a couple million users, mostly pointy-headed techie types, that doesn’t drive business to any money-making effort, with no alignment with any stategic effort, that could be reimplemented by a particularly talented and hardworking high school senior, staffed by about a dozen engineers. That product will be the first on the chopping block, if only to encourage those engineering resources to find more profitable employment.