The only sane approach is open data. Leave everything available for export so some other feature company can pick up where you left off.
You could venture into an insane realm where startups enter into a Startup Insurance™ league. When a company gets acquihired, instead of shutting down the product, a separate foundation runs the product (or figures out a way to open all the data for export).
In a sense the promise is a "non-credible threat" in game-theoretic terms. It would attract users if they carried it out, but because acquisition happens after the user-attraction stage they don't have an incentive to carry it out. This means they won't rationally carry it out, which means they don't attract rational users. This situation can be avoided if they "tie their own hands".