How to Automate a Pool Service Business (and When to Hire Instead)

A pool route generates office work whether or not anyone is paid to do it. At forty accounts it fits into the evenings. At a hundred and twenty it does not, and owners tend to find out the same way: an invoice run that slips a week, a customer who was never told their visit moved, a card that failed in March and went unnoticed until June.

There are two ways out, and most of the pages written about this will pick one for you. Hire someone, or automate. The useful answer is that they are not alternatives. Software removes some office work outright, makes some of it faster without removing it, and does not touch the rest at all. Working out which of those three buckets each task sits in is the entire decision — and it is the part the vendor pages skip, because all of them end at a signup button.

What the office work on a pool route actually is

Before you can decide what to automate you have to name the work, and most owners have never written it down. It does not feel like a list of tasks. It feels like one undifferentiated hour after the last stop. Written down, it is strikingly consistent from one route to the next.

Seven jobs. Notice that they are not the same kind of work. Two of them are pure repetition against data you already hold. Three of them are repetition wrapped around a decision. Two of them are almost entirely decision. That difference is what decides whether software helps.

What software removes, what it only speeds up, and what it never touches

This is the table worth arguing with. Every row is a judgement about the category of work, not a claim about any particular product — any competent modern platform lands in roughly the same places, because the limits are in the nature of the task rather than in the software.

The pattern in the first column is the finding. Everything software removes outright is billing: producing the invoice, collecting the money, retrying the card, and telling the customer it happened. That is not a coincidence. Those are the tasks where the right action is fully determined by data the system already has — who was serviced, at what rate, on what card. There is no judgement in any of them, which is exactly why a machine can finish them. If you only ever automate one thing, automate billing and payments.

The middle column is where owners get disappointed, because it is sold as the first column. Route optimisation does not build your route; it proposes an order you then correct for the things it cannot know — the customer who is home on Wednesdays, the dog, the gate that is locked before eight. Collections software does not collect; it tells you who to call and drafts the message. These are real savings, often large ones, but they are savings in minutes per task rather than tasks removed. You still need somebody who looks.

The jobs that never automate, and why they are the expensive ones

The last column is short and it is where the actual cost of running an office sits. Deciding whether to credit a customer who says the pool was green when they got home. Telling a long-standing account that their rate is going up. Working out whether a technician who has had three complaints in a month needs training or needs replacing. Taking the call from the customer whose liner was damaged and who wants to know what you intend to do about it.

These resist automation for the same reason they are stressful: each one is a decision with money and a relationship attached, made on incomplete information. No amount of software changes that. What software does change is how much of your week is left for them. An owner spending nine hours on invoicing is not spending those hours on the rate conversation that would have been worth more.

The arithmetic, in hours rather than salaries

Owners usually frame this as a salary decision, which makes it feel bigger than it is and delays it by a year. Frame it in hours instead. You are not choosing between a subscription and a full-time hire; you are choosing what to do with a specific number of hours a week.

Start by measuring. For two weeks, write down what the office work actually takes — not what you think it takes. Then put a rate on it. The U.S. Bureau of Labor Statistics puts the median wage for general office clerks at $21.64 an hour as of May 2025, which is the rate you would be paying somebody else to do it.

Fully loaded cost

The real hourly cost of an employee once payroll taxes, workers' compensation and the employer share of any benefits are added to the wage. It is always higher than the wage itself, and in pool service the workers' compensation component varies enough between states that the only reliable figure is a quote from your own carrier.

Ten hours a week at that wage is a little over $11,000 a year before the loaded cost is added, and a part-time hire is the realistic comparison for most routes rather than a salaried one. Set that against what software costs over the same year and the decision usually stops being close. What it does not do is disappear: the hours software removes are the billing hours, and the hours it leaves behind are the ones you were least looking forward to.

The threshold is a pattern, not a pool count

Owners usually try to settle this by asking other owners how many pools they had before they hired office help, and the answers are useless because the question is wrong. The same hundred pools generate very different amounts of admin depending on how you bill, how many customers pay automatically, and whether your schedule survives contact with a week of rain. Two routes of identical size can be four hours a week apart on office load alone.

The distance between those two columns is where nearly all the recoverable hours live, and none of it depends on route size. That is why the honest version of the threshold question is not how many pools you have, but which of those two columns you are in. An owner in the left-hand column at sixty pools has more to gain from automating than an owner in the right-hand column at two hundred — who may genuinely need a person next, because the easy hours are already gone.

There is a second reason the pattern matters more than the count. Office load does not grow smoothly with pools; it grows in steps, and the steps are triggered by things other than size. Taking on your first commercial account, your first HOA, or your first property manager adds billing complexity that fifteen more residential pools would not. So does a second technician, because the moment two people are on the road the schedule has to be legible to somebody other than you.

Hiring an office person can raise your software bill, on some platforms

This is the part nobody writes about, and it catches owners who do both: you hire the office help and discover the hire has a monthly cost on top of their wage, because your platform charges per seat. It is worth knowing before you choose either. The figures below were read on each company's own pricing page on 2026-10-23.

The order to automate in

Order matters more than product choice, because each stage makes the next one easier and automating out of order wastes the effort. Billing first, always: it is the only stage that pays for itself immediately and the only one where the manual version loses you money rather than just time.

Reporting comes last because reports built on data nobody has cleaned are worse than no reports, since you act on them. The phone comes fourth rather than first because answering calls automatically only helps once there is a schedule and a customer record for the answer to land in. And once the first four are running, the weekly numbers that run a route arrive without anybody assembling them.

1. 1

   Clean the customer list before anything else

   Remove the accounts you lost, mark the seasonal closures, and fix the rates that are wrong. Automated billing against a wrong list sends wrong invoices faster than you did.

2. 2

   Run one cycle in parallel

   Generate the invoices in the new system and compare them line by line against the ones you were going to send, before you send either. Discrepancies at this stage are cheap; the same discrepancies after sending cost you a phone call each.

3. 3

   Move customers onto automatic payment in cohorts

   Start with the accounts that already pay by card without complaint. Leave the two who will object until the process is working, so their objection is the only thing you are handling that week.

4. 4

   Only then turn on the reminder sequence

   Automatic retries and reminders on a book still full of wrong balances will chase people who do not owe you anything, which costs more goodwill than the arrears were worth.

When hiring is the right answer anyway

Sometimes it is, and a page that could not say so would not be worth reading. The signals are about the composition of the work rather than its volume. If what is overflowing is the last column of that table, no subscription will help you.

If you land on hiring, the arithmetic for a first office hire works the same way as it does for a first technician: the question is not whether you can afford the wage this month, but whether the work is steady enough to still be there in four months. Office work on a growing route is.

If you hire, automate the billing first anyway

This is the part that makes the two options less of a fork than they look. A new office hire inherits whatever system exists on the day they start. Hire onto a manual process and you have bought a person to do manual work — at a wage, indefinitely, and with the added cost that the process now lives in two heads instead of one. They will learn your habits, including the ones that were workarounds, and in six months those workarounds will be the documented procedure.

Automate the first column before the hire starts and the same person spends their week on the third column instead: the collections calls that need a relationship, the rate conversations, the commercial accounts, the customer who is upset and needs somebody to actually listen. That is work worth $21.64 an hour and more. Invoicing is not, whoever does it.

What Pool Runs does here, and what it does not

We are built around the first column. Recurring invoices generate against who was actually serviced that month, automatic payment and card retries run without anybody watching, and customers are told when a visit moves. Our pricing counts properties serviced rather than people, so office staff and technicians are unlimited at no extra cost — which means the hire-or-automate decision above is not also a pricing decision here.

What we do not do is the last column, and no honest vendor claims otherwise. We will not decide whether to credit an unhappy customer, have the rate conversation, or manage a technician who is not working out. We also will not pretend route optimisation removes route planning: it proposes an order and you correct it. If the billing hours are the ones you want back, take a look at how it works — and if the hours you want back are the judgement ones, hire, and do it before the season turns.