How to structure firebase data for multiple views - data-structures

Question

I'm new to Firebase and I'm building my first app on it so thought I'd ask if my current plans for the app's data structure make sense.
I've read the Firebase blog posts and several answers on SO which have helped me understand the concept of "optimise for the way the data will be read". However, my data will be read in a few different ways and it feels like I may be over complicating things.
Background
The app is like a directory for businesses in multiple towns (schemes) to promote their upcoming events and offers. I think of the data hierarchy like this:
Scheme: A town (the app has multiple schemes)
Category: A group of businesses around a theme (e.g. shoe shops)
Business: An administrative organisation (handles billing etc). Each business can have multiple locations (shops in different towns).
Location: A shop in a town.
Event: Each location can promote events. An event can be promoted at multiple locations but not necessarily all of a business's locations.
Offer: Similar to an event but a different type of object.
Viewing the data
The app user can view the offer & event data in 5 ways:
specific to a business (e.g. Joe's shoes' offers)
for a scheme (e.g. all offers in a Smalltown)
for the whole app (e.g. all offers anywhere)
in a category in a scheme (e.g. all shoe offers in Smalltown)
in a category in the whole app (e.g. all shoe offers anywhere)
In addition, I need to make sure that an administrator from each business can see/edit all of their business's data via a CMS I'm also building.
My approach
This is the data structure I'm thinking of using:
root {
schemes{
scheme1{
name: "smalltown",
logo: "base64 data",
bgcolor: "#FF0000"
},
scheme2{...}
},
businesses{
business1{
name: "Joe's Shoes",
logo: "base64 data",
locations: {
location1: true,
location3: true,
location15: true
},
address_hq: {
street: "45 Acacia Avenue",
town: "Bigtown",
postcode: "BT1 1JS"
},
contact_hq: {
name: "Joe Simpson",
position: "Owner",
email: "joe#joesshoes.com",
tel: "07123 456789"
},
subscription: {
plan: "Standard",
date_start: "10/10/2015",
date_renewal: "10/10/2016"
},
owner: "james1"
},
business2{...}
},
locations{
location1{
name: "Joe's Shoes",
logo: "base64 data",
scheme: "scheme1",
events: {
event1: true,
event27: true
},
offers: {
offer1: true,
offer6: true
},
business: "business1",
owner: "james1"
},
location2{...}
},
events{
event1{
schemes: {
scheme1: true,
scheme4: true
},
locations{
location1: true,
location21: true
},
categories: {
shoes: true,
footwear: true,
fashion: true
},
business: "business1",
date: "5/5/2016",
title: "The History of Shoes",
description: "A fascinating talk about the way shoes have...",
image: "base64 data",
venue: {
street: "Great Hotel",
town: "Bigtown",
postcode: "BT1 1JS"
},
price: "£10"
},
event2{...}
},
offers{
offer1{
schemes: {
scheme1: true,
scheme4: true
},
locations{
location1: true,
location21: true
},
categories: {
shoes: true,
footwear: true,
fashion: true
},
business: "business1",
date_start: "5/5/2016",
date_end: "5/5/2016",
title: "All children's shoes Half Price",
description: "Get 50% off all children's shoes - just in time for the summer",
image: "base64 data",
},
offer2{...}
}
}
Here's a graphic of similar data structure in case it's easier to read:
My question is whether I need to denormalise the data further (repeat more data in more places) or is there a better way to think about this altogether?
It feels like I'm getting potential complications from having to keep data in sync without the ability to simply read from a single place (e.g. I'll need to use queries and indexes (?) to combine location and event data for scheme-wide event listings).
Any advice on making this data structure more efficient would be great.