You have a short window to reverse a decline before it becomes a death spiral.
In a death spiral the activity drops, members visit less frequently, so activity drops further etc...You reverse a death spiral by starting again, back in the inception phase. That takes a lot of time, few organizations succeed in reviving a community in the death spiral.
Data is important here. You must measure the growth, engagement, and
Many community managers miss this window of opportunity. They don't monitor the key metrics of growth, engagement, and sense of community to identify and resolve potential issues.
For example, if you're tracking growth you might identify that the number of new members is beginning to dip. You can identify specifically why this is the case. Are less people visiting the site or completing the registration page? Are people that complete the registration page failing to make a contribution?
Using your data you can design an intervention to change this. For example, optimizing the newcomer to regular conversion funnel, initiating events to entice more members to join, or undertaking a WOM drive.
If you're tracking engagement, you might notice that a smaller number of individuals are accounting for an ever-larger number of discussions initiated.
The newcomers are crowded out. This is bad, over the long-term, activity rates will plateau and then dip. You can then design specific newcomers areas to get started, give the old-timers their own place or special roles within the community such as welcoming new members), or create a simple guide for newcomers to participate in the community.
These are just two of many possible examples.
Data is such a useful asset in the community development process. You have to set aside time to both collect and analyze it. Don't leave it several months. Collect it monthly and then adjust your activities accordingly.
I'm willing to bet that most communities that entered the community death spiral didn't track any data, at all.