The Field Operations Business

A field operations business sells work that happens at the customer's property. Someone drives a truck there, does the job, and leaves a record behind. Pest control, lawn care, HVAC, plumbing, pool service, cleaning, security installs and last-mile delivery all sell different work, but they share a cost structure: the expensive part isn't the materials or the minutes at the door, it's the miles between doors and the hours a skilled person spends driving them. Every system in this volume exists to put the right person at the right property at the right time, prove the work was done, and get paid for it, while driving as little as possible in between.

This chapter is the map. It covers how these businesses make money, why route density decides whether they make any, how they're organized on the ground, what a customer's life looks like from the first knock to the last visit, and how a platform for them splits into domains that hand work to each other.

Four Business Models#

Field businesses mix four models, and most run more than one. Which one a piece of work belongs to decides how it's sold, scheduled, priced and billed:

Model	Example	Scheduled by	Revenue	The hard problem
Recurring route-based service	Quarterly pest control, weekly pool cleaning, monthly lawn treatments	A cadence in the agreement	Predictable, per visit or per month	Keeping density as customers come and go
On-demand repair	A broken furnace, a leaking pipe	The customer's emergency	Per job, priced on site	Same-day capacity and skills
Installs and projects	A new HVAC system, a security install, termite baiting	A multi-visit plan with phases	Large tickets, often financed	Coordinating crews, parts and permits
Delivery	Appliance delivery, parts, groceries	A promised window	Per drop or per order	Windows, capacity and failed attempts

Recurring service is the strongest of the four, and the one most of this volume is about. The agreement fills the calendar for a year or more, the revenue is predictable enough to value a business on, and each new customer on an existing route costs little to serve. Its weakness is churn: a recurring business shrinks unless it sells faster than it cancels, which is why the same companies tend to run aggressive sales operations (Door-to-Door Sales).

On-demand repair pays more per job and fills nothing in advance. Its systems live or die on same-day dispatch: who has the skills, who's nearest, who can fit it in (Dispatch & the Day of Service).

Installs and projects are sequences: a site inspection, an install that may take a crew several days, follow-up visits, and sometimes a removal at the end of life. They're where service phases matter most.

Delivery shares the routing problem (time windows, vehicle capacity, a driver's shift) without the recurring relationship. Much of Route Optimization applies unchanged.

The models blur at the edges. A recurring pest customer calls about a new problem between visits, and that's an on-demand job on a recurring account. A termite installation is a project that turns into a recurring monitoring agreement. A system that hard-codes one model ends up faking the others; one that models work as agreements made of phases, each phase made of visits handles all four.

Route Density Is the Business#

Every visit has two costs: the time at the property and the time getting there. Only the first produces anything the customer pays for. A technician's productive day is roughly:

revenue per hour = (stops × revenue per stop) / (service time + drive time + everything else)

Most of the levers in that formula are fixed. Revenue per stop is set by the price list; service time is set by the work and, for regulated services, by the product label. The lever a business actually controls is drive time, and drive time is a function of density: how many customers there are per square mile of a route. Two technicians with the same skills and the same price list can differ widely in what they produce, purely by how far apart their stops are.

Density explains choices that look strange from outside:

• Selling is geographic. A new customer three streets from an existing route is worth more than the same customer across town, because their marginal drive time is near zero. Sales teams are assigned territories that match service areas, and canvassing targets streets the routes already cover.
• Price can depend on location. A property far from any route costs more to serve. Businesses either charge for it, refuse it, or cap how far a booking channel may reach (a phone rep may book within a few miles of an existing route, say, but not beyond).
• Cancellations hurt more than their revenue. A cancelled customer in a dense neighborhood leaves a gap the route still drives past.
• Scheduling trades flexibility for density. A recurring visit that may move a week either way can be grouped with its neighbors; one fixed to a date can't. That's why recurring agreements specify a window, not a date (Recurring Schedules).

My rule of thumb: when a feature request trades density for convenience (let customers pick any date, let any rep sell anywhere), the cost is real and shows up in drive time. Make it visible before saying yes.

From Branch to Region#

Field businesses are organized around the place the trucks start from. Names vary (branch, office, service center, depot, location), and this volume uses service center: a physical base with technicians, vehicles, stock and an office team, serving a defined geography.

The service center is the natural unit for almost everything:

• Geography. It serves a set of areas, drawn as polygons, which are split again into zones that individual technicians or teams cover regularly (Territories & Capacity).
• Time. It has a timezone, and its day (when routes start, when the office closes, when reminders go out) runs on local time.
• Licensing. Many regulated trades license the business location as well as the individual, and the service center holds those licenses (Licenses & Permits).
• People. Technicians, office managers and dispatchers belong to one. Their permissions usually stop at its boundary.
• Reporting. Completion rates, revenue and backlog are reported per service center first, then rolled up.

Service centers roll up into regions, and regions into divisions or the whole company. That hierarchy is a reporting and permissions structure, not a tenancy boundary: one company with eighty service centers is one tenant with eighty partitions, and treating each as a separate tenant makes moving a customer, a technician or a whole area between them far harder than it needs to be (Multi-Tenancy covers the general tradeoff).

Boundaries move. A service center opens and takes areas from two neighbors; a region is reorganized. Any record that says which service center owns something needs effective dates, or last year's reports quietly change when this year's map does.

The Life of a Customer#

Follow one recurring customer from the first contact to the end:

1. Lead. A knock on a door, a web form, a referral, an inbound call. The lead is attached to a property before it's attached to a person, because whether the property can be served (and at what cost) decides whether it's worth pursuing.
2. Agreement. The customer chooses a plan, a length and a payment method, and signs. A sale made at the door or in the home comes with a legal window to cancel (Consumer Protection), so a signed agreement isn't yet a customer. The sale usually reserves the first appointment at the same time (Signing the Agreement).
3. Initial service. The first visit is longer and heavier than the rest: an inspection, a full treatment, sometimes equipment installed. It's often priced separately. For many recurring businesses, billing starts when the initial service is completed, not when the agreement is signed (Billing Recurring Service).
4. Recurring service. Visits repeat on the agreement's cadence, each inside a window around its target date. The schedule is generated, not booked by hand.
5. Reservice. When the problem comes back between visits, the customer calls and a technician returns at no charge. A reservice is part of what was sold, so it doesn't bill and doesn't reset the cadence.
6. Renewal or cancellation. At the end of its term the agreement renews (often automatically, which is regulated) or ends. Customers also cancel mid-term, freeze for non-payment, or move away. A cancelled customer at a still-serviceable address is a lead again.

Two things about this lifecycle drive the system design. First, it crosses every team: sales owns the first two steps, operations the middle, finance the money, and customer service the exceptions. Second, it's long: a customer may be active for years, through price changes, plan upgrades, technician turnover and a move across town, and the records have to make sense at every point along the way.

The Domains#

A platform for field operations splits naturally along the lifecycle. These are the domains, what each owns, and what it hands to the others:

Domain	Owns	Hands off
CRM	Accounts, people, properties, contacts, notes, alerts	Who the customer is, to everyone
Sales	Territories, leads, knocks, pricing rules, agreements, reps	A signed agreement and a reserved first visit
Field service	Schedules, appointments, routes, technicians, work orders, service records	Completed visits, with what was done and applied
Billing	Subscriptions, invoices, payments, payment methods, balances	Paid, failed, frozen and cancelled
Communications	Messages, templates, consent, delivery status	Sent, delivered, opted out
Licensing	Jurisdictions, requirements, licenses, who may work where	Compliant or not, per person, role and place
Platform	Identity, roles and permissions, service centers, feature flags, audit	Who may do what, and where

The boundaries follow ownership. Each domain is the only writer of its records, and others learn about changes from events: an agreement completes, billing creates the subscription, CRM records it on the account, and field service turns the reserved slot into a confirmed appointment, each reacting to the same agreement.completed event without calling the others (Events, Service Boundaries).

Some data has to be shared, and it's shared as copies with a clear owner. The status of a subscription belongs to billing, but field service needs it to decide whether to schedule, so field service keeps a read-only projection updated by events. When the two disagree, billing is right and the projection catches up. Platform Architecture covers the mechanics.

The handoffs are where the business loses money, so they're where the design effort goes:

• Sale to first service. A sale whose first visit never gets booked is revenue lost and a customer annoyed before they've had a single service. The reserved slot has to survive the gap between "signed" and "confirmed", including queue lag and a slow signature.
• Service to billing. A completed visit that never bills is free work. A bill for a visit that never happened is a dispute.
• Billing to service. A customer who stopped paying keeps getting serviced unless billing's freeze reaches the schedule. One who paid keeps getting cancelled visits if it reaches it wrongly.
• Any change to communications. A reminder for an appointment that was cancelled yesterday is the most visible bug in the business. Every event that removes or moves an appointment also has to retract what was queued about it.

Seasonality#

Field work follows the weather, and so does every number the business runs on.

• Demand. Insects, lawns and pools are busiest in the warm months; heating repair peaks in the cold ones; snow removal exists only when it snows. Recurring cadences often change with the season: visits closer together in summer, further apart (or skipped) in winter, and a plan's schedule has to encode that (Recurring Schedules).
• Selling. Door-to-door sales in many home services runs as a summer campaign, staffed by seasonal reps who arrive for a few months, often relocated to the markets they sell in. The sales system has seasons as a first-class concept: rosters, territories, goals and pay terms all belong to one (Sales Teams & Commissions).
• Capacity. A summer of selling becomes a backlog of initial services that has to be absorbed by the same technicians already running full routes. Capacity planning is mostly about the weeks when the backlog and the season's peak demand land together.
• Weather events. Rain, wind, heat and storms stop work for a day or a region. A platform needs a way to stand down a service center for a day and move every affected visit, not a dispatcher doing it one appointment at a time (Dispatch & the Day of Service).
• Hours. The work happens when customers are home and the weather allows it. Evenings and weekends are common in the trade, which means the systems that support technicians (and the people who support the systems) can't assume office hours.

Seasonality is also why "the same week last year" is a better comparison than "last week" for almost every operational metric (Operating Metrics).

Who Uses the System#

A field platform serves more kinds of users than most products, each on a different device, in a different place, with a different tolerance for friction:

User	Where	Device	Needs
Technician	In the truck and at the property	Phone or tablet, often offline	Today's route, the property's details, what to apply, how to finish a visit
Sales rep	At the door	Tablet	The map, the price, the contract, a first appointment
Dispatcher or route planner	The service center	Desktop	Every route today and next week, and what's at risk
Office manager	The service center	Desktop	Their center's people, backlog, compliance and money
Customer service rep	A contact center	Desktop plus phone system	The whole account, fast, in the middle of a call
Finance	Headquarters	Desktop	Receivables, failed payments, write-offs, exports
Customer	Anywhere	Their own phone	Their next visit, their bill, a way to reschedule

The technician and the sales rep are the ones to design for first. They're working outdoors, often in bad connectivity, frequently with gloves on or a customer watching, and every extra tap is multiplied by every stop of every day. The desktop users can absorb complexity; the field users can't.

What This Means for the System You Build#

• Agreements made of phases, phases made of visits, so recurring, on-demand, project and delivery work share one model.
• Density as a first-class input to selling, pricing and booking, with distance rules per channel and drive time visible on every route.
• Service centers as partitions, not tenants, rolling up into regions, with effective dates on every ownership boundary.
• The customer lifecycle as explicit states, with billing starting at the initial service and reservices inside the cycle.
• Domains with single owners, handing off through events, with read-only projections where others need the data.
• Retractable side effects: every event that moves or cancels work also cancels what was queued about it.
• Seasons in the data model: cadences, rosters, goals and capacity that change through the year.
• Field users first: offline, few taps, built for the truck and the doorstep.

References#

Reference

• U.S. Bureau of Labor Statistics: Pest Control Workers, work at the client's property, driving service trucks, evening and weekend work, and state licensing
• Martin Fowler: Bounded Context, splitting a large domain into contexts with their own models