When a startup goes away, either through acquisition, or any other reason, and I find the site dormant, with a friendly goodbye message, or just gone, I usually just remove the tag from them in my primary system, and they are gone from my research with the next publishing of the project.
When Parse announced they are turning off the lights, and Kimono Labs said they were going away after an "acquisition" by Palantir, I started putting more thought into what I was going to do at the time of deprecation. I want to be able to capture more of these stories. I actually have a lot of data, content, notes, and analysis on these companies that are going away, and I need to do a better job making it all accessible.
To support this effort I am beginning to add a dead pool page to each of my research areas, where I will simply re-tag any company or organization that is present in my research, when they shut off the lights—providing a single place where we can see our recent history.
I find it valuable to have this recent history, right at my fingertips, when I am trying to understand where each of my research areas are headed. Without the actual companies to reference, I am reliant on my own memory, which isn't always the best. In the tech space, I feel like we are good at focusing on the future, and we easily forget the past, even what just happened last month.
I am always amazed at which startups just go away, without any notice. Many of them do not have customers, but some of them do, and many of them have asked me for my thoughts on the space and on their road maps. Part of having a deprecation page is also to inform newcomers that the road is littered with corpses of those who came before them, but to also provide me with a constant reminder that when I am having these conversations with new players that I put pressure on them to have a better strategy for when things go wrong—not everyone wins in this game.