Pool Service Software Problems: The 7 That Eat Route Margin, and How to Avoid Them

Why does replacing Skimmer end up in the scope so often?

Because the first conversation is about frustration rather than about money, and the frustration attaches to whatever screen the owner looks at most. Skimmer is on every phone in the fleet, so Skimmer takes the blame for a leak it was never designed to close.

Skimmer is a good route and chemical log app. It records what happened accurately and quickly, which is genuinely hard, and the techs already use it. What it does not do is reconcile what happened against what should have been billed, chase a skip before the customer notices, or answer the phone at nine at night. Those are different jobs and they sit above the field app rather than inside it.

A project scoped as replacement fails twice. The budget goes into rebuilding route sequencing, chemical logging and offline capture that already work, and the techs get a new app to learn, which costs you stops in the first weeks of a season. Then the money that was leaking is still leaking, because nobody built the reconciliation engine.

Scope it as a layer instead. Read dosages, completions and skips out of Skimmer or ServiceTitan, add the revenue capture, write nothing back on day one. That is the $50,000 to $120,000, ten to sixteen week first release in our delivery experience, and it stops a real leak while your field operation keeps working exactly as it does today.

What goes wrong when customers, plans and chemical history are migrated?

The customer list migrates cleanly. The plans do not, and the plans are what the whole build depends on.

In most pool companies the service plan is a price and an implied set of inclusions that live in the owner's head or in a contract PDF from four years ago. One gallon of acid included per visit. Filter cleans twice a year. Salt included but not chlorine. The customer who has been on a legacy rate since before the current pricing existed. The commercial account with a different allowance for summer than for winter. None of that is recorded as structured data anywhere, so a reconciliation engine has nothing to reconcile against.

The failure is starting the build before that work is done, which produces a system that flags every dosage as an overage and gets switched off inside a fortnight. The fix is to make plan definition an explicit phase with the owner in the room: allowance per chemical per plan, per visit or per month, with exceptions recorded as exceptions rather than as separate plans. Expect it to surface accounts nobody could price today.

Chemical history is the easier half. Migrate it in full, because it is the input for spotting pools trending toward green and customers who cancel in month four, and it costs almost nothing to carry.

Why do the Skimmer and ServiceTitan integrations break after launch?

Three ways, and only one of them is technical.

The first is drift in what the field records. Reconciliation depends on techs logging dosages consistently, and a new tech in July logs a bag of shock in a different field or a different unit than the tech he replaced. The engine keeps running and quietly stops finding overages on his route. The defence is a coverage check rather than an error check: alert when a route's dosage logging rate drops against its own history, not just when a record fails validation.

The second is export shape. A vendor changes a column, adds a status value, or alters how a partially completed stop is represented, and the nightly pull keeps succeeding with subtly wrong data. Contract tests that assert the shape of every field, and a hard rule that an unrecognised status stops the batch instead of defaulting to complete, are cheap and they save a billing cycle.

The third is scope creep into writing back. Once the layer proves itself, someone wants the credit to appear in the field app. Write paths into a field system are a different risk profile from read paths, because a bug now changes what the tech sees on a route. Sequence that deliberately, after the read side has run a full season.

What happens when commercial, HOA and multi plan billing rules are not covered?

You get a system that works on residential flat rate accounts and gets bypassed for everything else, which is usually where the margin actually is.

Commercial and homeowner association contracts carry their own terms: negotiated chemical inclusions, service frequency that changes by season, penalty clauses if water chemistry falls outside a range, invoicing to a property manager rather than to the site, and purchase order numbers that must appear on the invoice or it will not be paid. A reconciliation engine built only for residential plans cannot express any of that, so the office keeps handling those accounts manually and the two most valuable parts of the book stay invisible.

The same applies to skips. A residential skip triggers a text and a credit rule. A commercial skip may trigger a contractual notification with a defined remedy window, and getting that wrong is a contract problem rather than a customer service problem.

Model contract terms as first class objects from the start, even if the first release only implements the residential path. Retrofitting a second billing model into an engine that assumed one is close to a rewrite, and mixed residential and commercial operations sit at the upper end of the cost band precisely because of this.

Should you build custom or configure what you already own?

If you run one or two trucks, bill a clean flat rate with no chemical overage, and your phone is answered during business hours, Skimmer, Jobber or Housecall Pro is genuinely enough. Do not spend six figures to save an hour a week. Off the shelf earns its keep for a long time in this industry and most companies should stay there.

Configuration is also the right answer for a narrower set of problems than people assume. If your issue is that invoices go out late, fix the process. If it is that techs skip stops because a route is badly sequenced, re-sequence the route. Neither needs a build.

The build case is specific. You run multiple crews and can feel margin leaking but cannot point to where. You hand reconcile chemical logs against billing every week. You are losing after hours calls and you know it. You have years of history in a system that has never told you anything. Or you are a rollup trying to run one playbook across several acquired brands on four different systems, which is the hardest version of this and the one where a shared layer above the incumbents pays fastest. At that scale the subscription fee is not the cost. The margin the tool cannot see is the cost.

How do hidden costs get into the quote?

Billing rule count is the main one, and it hides because it is invisible until someone writes the rules down. Per plan allowances, commercial versus residential, homeowner association contracts with their own terms, seasonal variations and legacy rates each multiply the reconciliation logic. A single plan residential company sits at the low end of the band. A mixed operation with contract accounts sits meaningfully higher, and the gap is rules rather than screens.

Integration depth is the second. Reading a clean nightly export is one price. Live synchronisation against a field system, with conflict handling when a tech edits a completed stop after the batch ran, is another. Quotes rarely distinguish them, so ask which is priced.

The phone agent scope is the third. Handling a residential green pool call is a bounded problem. Handling commercial callers and property managers, with account lookup, contract terms and escalation rules, is a different build and it usually appears in phase two proposals without a price attached.

The fourth is history cleanup. Years of inconsistent data has to be normalised before it can be mined, and how bad it is determines the number. It is almost always worse than the owner believes.

What separates a build that works from one that fails here?

The builds that work start where the money is and prove it in weeks. Chemical reconciliation, skip detection with credit rules, and same day invoicing, sitting on top of the existing field app. The owner sees a weekly batch of draftable charges with a dollar total on it, and the argument for phase two makes itself. The builds that fail start with a dashboard, because a dashboard is easy to demonstrate and it never once put a charge on a statement.

The second marker is whether a skip is an event or a red dot. A red dot on a screen is what you already have. An event fires a workflow: text the homeowner with the reason and a reschedule, apply your credit rule automatically if the pool is not serviced inside your window, and log the pattern so you can see that one route loses ninety minutes every Friday to bad sequencing. That last part turns a tech problem into a routing fix, which is a different and much cheaper conversation.

The third is whether the developer understands route based field operations at all. Ask them to whiteboard how a logged chemical overage becomes an approved billable line on a statement, including who approves it and what happens when the customer disputes it. If they cannot, they will learn on your dime during your busiest season.

Then ownership, in writing, before kickoff. The chemical readings, customer history and route data are your business asset and the software that reads and bills against them should belong to you outright, not be rented back. A developer who hesitates on that point has told you what the relationship will be.