D&D 5e worldbuilding hierarchy: regions, settlements, POIs, factions

Why flat worldbuilding eventually breaks

The first campaign I ran in earnest, my world was a big dump of names. The town of Brassgate. The Iron Crown faction. The forest of Mordgate. Lord Halrick. The river Veliss. All sitting in a single Google Doc, in the order I thought of them, with no relationship between any of them.

Twenty sessions in, I couldn't remember which town the party had visited where, who lived there, which faction was active in which region. I was rebuilding my world from scratch every prep night because my notes were a wall of text I couldn't navigate.

The fix wasn't writing more lore. It was structure. Specifically, a hierarchy: regions contain settlements, settlements contain points of interest and shops, factions cross-cut all of it. Once I started building worlds that way, prep time dropped by half and the world started feeling like a place instead of a list.

This post is the hierarchy that's been working for me, why each layer matters, and how to use it without it becoming bureaucracy.

The four-layer hierarchy

Here's the model. Top to bottom:

1. Region. A geographic area. Could be a kingdom, a wilderness, a coastline, a mountain range. Big-scale stuff. Examples: the Sunfall Coast, the Ironpeak Mountains, the Verdant Marches.
2. Settlement. A populated place inside a region. Town, city, village, fort, hidden enclave. Examples: the city of Brassgate (in the Sunfall Coast), the dwarven hold of Karaz Drun (in the Ironpeak Mountains).
3. Point of interest (POI). A specific notable location, either inside a settlement or out in the wilderness of a region. The temple of the Pale Star, the abandoned mine, the ruined watchtower, the inn called The Crooked Anchor.
4. Shop. A specific type of POI that buys and sells goods. Blacksmith, alchemist, fence, magic shop. Often inside settlements but sometimes standalone (the traveling caravan, the hermit's cabin where you can buy potions).

That's the place hierarchy. Crossing it sideways:

• Factions. Groups of NPCs with shared goals that operate across regions, settlements, or POIs. The Iron Crown might have offices in three settlements across two regions. Treat factions as their own dimension, not children of any place.
• NPCs. Individuals. Each NPC lives somewhere (one settlement, or roams a region) and may belong to factions. NPCs are leaves in the tree, but they have multiple parents.
• Lore. Histories, religions, events. Lore can attach to anything (a region's history, a settlement's founding myth, a POI's curse, a faction's origin) but doesn't fit cleanly in the place hierarchy itself.

The point of the hierarchy isn't taxonomy. It's that you can navigate any node and see its parents, children, and lateral connections. When the party arrives at the Crooked Anchor, you click through and see: it's in Brassgate, which is in the Sunfall Coast region, run by the innkeeper Tobin Vance, frequented by the Mercer Trading Co, and the players have a +25 reputation with the Mercers. That's seconds, not minutes of digging.

Region: the big-scale layer

A region exists to give settlements and POIs context. The Sunfall Coast feels different from the Ironpeak Mountains because the region defines climate, dominant culture, threats, and travel time.

What I track at the region level:

• One-paragraph description. Climate, geography, dominant cultures, mood. Three to five sentences.
• Major threats. What's dangerous about traveling here? Bandit clans, wild magic, monstrous predators, harsh weather.
• Active factions. Which of the campaign's factions operate in this region? Iron Crown? Mercer Trading Co? Local cult?
• Travel notes. How do you get around? Roads, ferry, dangerous passes, magical means.
• Top three settlements. Which settlements in the region are most likely to come up?

That's it. I keep regions short. The trap is over-detailing the region itself when the actual content lives in the settlements and POIs underneath.

A pattern I use: write a region's "vibe" in one sentence before anything else. The Sunfall Coast's vibe is "broken empire, smugglers, sun-bleached ruins, paranoia." The Ironpeak Mountains is "ancient dwarven holds, undead beneath, surface clans at war." Once the vibe is locked, everything I add downstream pulls toward it. Settlements feel coastal-and-paranoid in the Sunfall Coast and stoic-and-hardened in the Ironpeak. The region does the work of consistency without me having to enforce it manually.

How many regions you need: usually three to six for a campaign. Less than three and the world feels small. More than six and most are scenery the party never visits. Trim aggressively.

Settlement: where most play happens

Most D&D sessions take place in a settlement. The party arrives, they shop, they gossip, they get a quest, they go to the inn, they meet someone, they leave. Settlements are the workhorse layer of your world.

What I track at the settlement level:

• Type and size. Hamlet (50 people), village (200), town (2,000), city (10,000+). This shapes available services. A hamlet doesn't have a wizard's tower.
• Government. Who's in charge? Mayor, council, lord, guild, theocracy, no one in particular.
• One-paragraph mood and recent history. What's it like to walk into? What's the rumor on the streets right now?
• Active factions. Which factions operate here, and at what strength?
• Top three POIs. The places the party will probably interact with first. Inn, temple, market.
• Top three NPCs. Innkeeper, faction contact, person of authority. Names, voices, one-line descriptions.
• Hooks. Two or three things going on right now that could become quests.

Notice I'm not writing 800 words about the settlement's founding history. I'm writing what the party will encounter when they walk in. The history lives in lore documents and gets pulled in if and when it matters.

A useful exercise: when you create a new settlement, write the first thing the party will smell when they enter. The Crooked Anchor in Saltmarsh smells of fish, pipe smoke, and salt. Fishing village. Locking in a sensory detail at the top makes everything that follows easier to populate.

POI: where the action happens

Points of interest are the named places inside or outside settlements. The temple, the ruin, the mine, the inn, the tower, the cave. Anywhere a scene might occur.

What I track at the POI level:

• Type. Inn, temple, ruin, shop, tower, cave, etc.
• Owner or notable inhabitant. Who runs it or who's there?
• Description. What does it look and feel like? Two to four sentences.
• What's interesting? Why might the party come here? What's hidden, what's hooked, what's available?
• Connections. Which factions are tied to it? Which NPCs spend time here?

POIs are where I put the most concrete creative effort. The temple of the Pale Star with the secret room behind the altar. The Crooked Anchor where the smuggler ring meets at midnight. The abandoned watchtower where the warlock has set up shop. These are the locations the players will remember, and that's because each one has something worth remembering.

A heuristic I use: every POI should have one specific detail that isn't generic. "An inn" isn't a POI. "An inn whose innkeeper is a retired adventurer who lost an eye to a beholder and tells the same story every night" is a POI.

Shop: the commerce layer

Shops are a special case of POI because they have mechanical implications. The party buys things at shops. So the data is different.

What I track at the shop level:

• Type. Blacksmith, alchemist, general store, fence, magic shop, etc.
• Owner. Name, race, one personality detail.
• Inventory. What's actually for sale, including prices.
• Faction tie-in. Is this shop aligned with a faction? Does faction reputation affect prices here?
• Hidden inventory. What's in the back room, available only to trusted customers?

The hidden inventory thing is important. A neutral-reputation party gets the front room. An honored-reputation party gets the back room. This creates a real incentive for the party to invest in faction reputation, and it gives you a clean lever for unlocking interesting items as the campaign progresses.

I covered the faction reputation system in detail in another post. The shop hidden-inventory thing is one of the simplest, highest-impact uses of that system.

Factions: the cross-cutting layer

Factions don't fit the place hierarchy because they aren't places. They're networks of NPCs with shared goals that operate across multiple settlements and regions.

I treat factions as their own collection in my notes. Each faction has:

• Name and one-line description.
• Goal. What the faction is trying to achieve.
• Leader and key NPCs. Two or three names with roles.
• Active locations. Which settlements or regions they operate in, and how strongly.
• Friction with other factions. Who they're aligned with, who they hate.
• Reputation track. Where the party currently stands, with notes on key events.

The trick with factions is that they show up in multiple settlements. The Iron Crown has an office in Brassgate, a chapter house in Karaz Drun, and an outpost in the Verdant Marches. So when the party walks into a settlement, you check which factions are present, and you know who the party will be encountering.

If you only had a flat list of NPCs, this would be hard to keep straight. With a faction layer, it's automatic.

Lore: the optional layer

Lore is everything else. The history of the kingdom that fell five hundred years ago. The religious schism that split the priesthood. The myth of the wandering hero who left a sword somewhere.

I keep lore as separate notes that link to whichever layer they're relevant for. The history of Brassgate is a lore note linked to the Brassgate settlement. The myth of the wandering hero is a lore note linked to the region where the hero might have left their sword.

The trap with lore is that it grows. You can spend infinite hours writing lore that never reaches the players. My rule: I only write lore that I expect to come up in play within the next five sessions. Anything beyond that is hypothetical and can wait.

If you write 30 pages of cosmology before the party even leaves the starter town, you're not building a campaign, you're building a novel. Different hobby. (No shade if you genuinely love it. Just be honest about what you're doing.)

How the hierarchy speeds up prep

Here's the payoff. Once the structure is in place, prep nights look different.

Old me: "The party is heading to Brassgate. What's there? Let me search my Google Doc... okay, Brassgate, here are some scattered notes. Who's the inn run by? I think I named someone but I can't find it. Let me check the Sessions log... session 6, Tobin Vance, okay. What faction does he work for? Did I write that down? Let me search..."

That's twenty minutes of digging through unstructured notes to recover information I already wrote.

New me: "The party is heading to Brassgate." Click. Settlement page. Top three POIs, top three NPCs, active factions, recent hooks, all visible. The inn is the Crooked Anchor, run by Tobin Vance, who's a Mercer Trading Co contact, and the rumor right now is that a Veilkeeper agent has been asking about the docks. Five seconds.

The hierarchy isn't fancier. It's just structured. The structure is the value.

How tools handle this

Quick honest comparison.

Notion. You can build the hierarchy with relations between databases. Regions database, Settlements database (with region relation), POIs database (with settlement relation), and so on. It works, but you'll spend significant setup time wiring the relations and building the views.

Obsidian. Markdown files in a folder structure. Vault layout becomes the hierarchy. Linking is fast, and Dataview can query across files. Powerful but requires fluency.

World Anvil. Built for this. Article hierarchies are first-class. Powerful but the UI complexity is real.

LegendKeeper. Map-driven hierarchy. You navigate by clicking on the map. Beautiful for visual thinkers but limited if you don't have or want maps.

Dungeon Diary. I built this hierarchy literally into the data model. Regions, settlements, POIs, shops, factions, NPCs are all separate collections with explicit cross-relations. When the party arrives at a settlement, the page shows everything inside it and everything connected to it. The AI generation also uses this hierarchy, so when you ask for a new shopkeeper in Brassgate, the AI knows the city's existing factions and culture.

The tool you pick is less important than whether you actually use the structure. A working hierarchy in Notion beats an unused one anywhere else.

A starter exercise

If you're starting a campaign and you don't have any of this yet, here's a forty-five minute exercise that will set you up for the next thirty sessions.

Pick a campaign location. The Sword Coast, Faerûn, your homebrew, whatever.

Write down:

• Three to five regions with one-line descriptions and vibes.
• Two settlements per region (so six to ten total). For each, the type, government, and three-line mood.
• Three POIs per settlement (so eighteen to thirty total). For each, the one-line "what's interesting."
• Four to five factions, each with goal, leader, friction with another faction.
• Six to ten NPCs scattered across the settlements with name, faction, and one motivation.

Resist the urge to expand. Forty-five minutes. Then start running sessions. Add detail as the party encounters things, not in advance.

You'll find that this skeleton holds the campaign for the first ten to fifteen sessions, and by then you'll have organic depth in the places the party visited and you can leave the rest as scenery.

The closing thought

Worldbuilding hierarchy isn't about having more lore. It's about being able to find and connect the lore you already have. When the structure is right, the world feels like a place. When the structure is missing, the world feels like a list.

Pick a tool that supports the hierarchy. Set up regions, settlements, POIs, shops, factions, and NPCs as connected layers. Resist over-detailing any one node, especially at the top. Let the world grow where the party touches it.

That's the model. Forty-five minutes to set up, dozens of hours saved over the campaign.