[RPG] a plot hook, exactly

You've run into one of the dangers of pre-planning a plot. I'll give some ideas at the end about how to plan campaigns so this doesn't happen as much in the future, but first we have to deal with the current situation. Other answers have dealt nicely with the "stay on the rails" and "take a short detour" options, so I'd like to talk about a third choice:

Take a new path through the bush. Forget the plot you had lined up. You've got interesting people and conflicts already present in the world and waiting in the wings, but you can let go of exactly how you expect it to play out. Go along with the PCs' choices and look for opportunities to introduce the interesting people and ideas you have prepared. Instead of killing off this NPC and negating the PCs' hard work, use the NPC as a gateway to new adventures which will incorporate your ideas in new, interesting ways.

It shouldn't be too hard for, e.g., someone connected with the NPC to kick the bucket in a ghost-inducing way, so the NPC drags the party in for the ride and you get a ghost investigation on the party's terms rather than on your own. Players tend to be more engaged with plots that arise from PC agency than plots which are thrust upon them.

Now lets talk briefly about avoiding this kind of situation in the future. For me, the key lies not in how I plan, but in what I plan. Instead of creating interesting stories to walk my players through, I need to create interesting people, situations, and conflicts which are happening when the game starts.

My favourite kind of game prep is to set up a complex set of NPC/faction/world interactions and then watch my players roll through them like a lopsided bowling ball. This way, my players can engage with a world and have their choices matter because they're interacting with dynamic processes. As the party acts and makes choices, I'm free to have the world react: NPCs change their plans based on PC action, natural events occur when it's most dramatic, and so forth.

An RPG story is about the PCs, so I like to give them a chance to really make the world sit up and notice their choices. The best way I've found to do this is to avoid planning stories that hinge on the players making certain choices.

You're right, these three things are key to a true West Marches campaign:

1) There was no regular time: every session was scheduled by the players on the fly.

2) There was no regular party: each game had different players drawn from a pool of around 10-14 people.

3) There was no regular plot: The players decided where to go and what to do. It was a sandbox game in the sense that’s now used to describe video games like Grand Theft Auto, minus the missions. There was no mysterious old man sending them on quests. No overarching plot, just an overarching environment.

Ben Robbins, whose blog you quoted, should know, he popularised the concept!

There is a bit more to it though...

Implied, but not explicit in those first three points:

1. Every game session begins and ends at the same point of origin (normally a home town). This means:

   • Every session can be entirely self contained. The new session will always represent a completely new foray, out from the base, into the wilderness, even if (coincidentally) the party is exactly the same as last time, and want to resume a similar pursuit.
   • Players' characters don't have to spontaneously appear in narratively unsatisfying ways when needed.
   • In later adventures, as travel times increase away from that point of origin, the act of travelling itself is more likely to be handwaved / skipped over.

2. The players decide where to go and what to do in advance.

   • Yes, it's a sandbox, but the players must decide what they would like to do in advance of the session. Normally, as part of booking the DM's time for the session, the players would also state what they wanted to do in that session. This way the DM can prepare in advance, without needing to prepare the whole world, or improvise large peices of content.

Outside of those first three points:

1. Session reports are always shared

   • As all PCs are assumed to spend their downtime in the same town, word gets round about what happens on each adventure. Players are encouraged to write up session reports and distribute them to each other. It's a living world, the same goblin warlord cannot be killed by two different groups, so the second group that wanted to try needs to know if the first succeeded or not.
   • New quest hooks can be picked up by any group of players.

2. There is a shared world map, that's potentially unreliable

   • All initial objectives and later objectives that are discovered are marked on a shared map, which players can use to suggest places they want to explore.

   • The initial map is produced in-game by a character and is only as reliable as that characters map-making abilities / trustworthiness. It is later edited by the players who may also make mistakes. This means it's possible to get lost, if the map is wrong (and the players can subsequently correct it).

3. Competition between players is actively encouraged

   • Jealousy is considered to be a useful motivational force in getting sessions booked and games actually played. If anyone else can pick up from the interesting place where you last left off, or someone else has discovered something exciting, then that motivates you to prioritise organising your next session. Especially when magic items are on the line.

4. Content is loosely tiered

   • Players started at a low level and would meet on average higher levels of danger the further they ventured outside of town. This meant players could largely assess whether a threat was likely to be appropriate or not.
   • Significantly stronger threats in low level areas were normally well sign posted. Having pockets of more difficut enemies made the world more exciting, more diverse and incentivized PC's returning to earlier explored areas later, when they were stronger.

Matt Colville made a great video that covers all of this in detail, if you don't fancy reading all of Ben Robbins' blogs.

However, when someone says they are running a "West Marches style" campaign, they may often mean something much looser than this:

The only conclusive way to know what someone else really means when they say they're running a "West Marches style" campaign is to ask them.

That said, at a bare minimum, it is likely that these things are implied:

1. They don't have a regular group of players, but pull from a larger pool each session.

2. Each session will be entirely self contained.

3. There's probably a focus on exploration.