Custom Window Cleaning Software: When to Build vs. Buy

Your Jobber account has 12 custom fields hacked together just to track pane types. You built a separate Google Sheet for commercial billing because Jobber rolls invoices up to a location, not to the parent account. Your dispatcher spends two hours every morning manually sorting jobs into crew routes. And last month, when Crew 3 was out sick for a week, 40 recurring jobs slipped off the calendar entirely.

You are not using Jobber wrong. You have outgrown it.

This is the point where some operators buy a more expensive tool, hoping it solves the same problems with a shinier interface. Others start asking whether custom window cleaning software would actually pay for itself. This article is for the second group.

According to IBISWorld's window cleaning services industry analysis, the US window cleaning industry generates over $2.5 billion in annual revenue. The majority of that sits with operators running 5 to 30 crews, precisely the size where off-the-shelf tools start creating more friction than they remove.

RaftLabs has built recurring scheduling and field service platforms for operators at this scale. Here is what to build, what it costs, and how to know if it is the right call for your business.

Cost and timeline at a glance

The gap between MVP and full platform is almost entirely commercial contract management and franchise architecture. If you do not have either, the MVP covers the majority of operational needs.

Jobber, Housecall Pro, Service Fusion, and Workiz vs. custom window cleaning software

This is the section that actually matters. Most operators ask the wrong question: "Which SaaS tool is best?" The right question is: "Which problems can SaaS tools not model, no matter which one I pick?"

Here is an honest breakdown of each tool and where they stop working.

Jobber ($69-$349/month) is the most commonly used window cleaning business software for operators up to about 8 crews. It handles recurring jobs, basic routing with a map view, invoicing, and a client hub. The scheduling and invoicing are genuinely good. Where it breaks: commercial account hierarchies (you cannot bill a parent company across 12 locations without manual aggregation), multi-crew route optimization (the route view shows jobs on a map but does not calculate an optimal sequence for multiple crews simultaneously), and pane-level pricing granularity. At 8 or more crews running 10 to 14 stops each, the manual dispatch work compounds daily.

Housecall Pro ($65-$215/month) is the most accessible window cleaning service software for residential-heavy operators. The mobile experience is clean and crews adopt it fast. It breaks earlier than Jobber on commercial work. If more than 20% of your revenue is commercial, you will be managing contract details outside the tool within months. It also lacks route optimization across crews.

Service Fusion ($165-$349/month) goes deeper on dispatch and fleet management. It is a better fit for operators with mixed commercial and residential books. The routing is better than Jobber and Housecall Pro, but still does not handle the Vehicle Routing Problem correctly at 6 or more crews. The reporting is stronger. It gets expensive fast as you add users.

Workiz ($45-$225/month) is newer and has better communication tools: built-in calling, SMS, and a client-facing portal. Good fit for operators competing on customer experience. Breaks on the same commercial contract and multi-crew optimization problems as the others.

The pattern: all four tools are well-built for the operator segment they were designed for. None of them were designed for:

• Commercial accounts with 10 or more locations under one billing parent

• True multi-crew VRP route optimization across 6 or more simultaneous crews

• Franchise networks where each territory needs isolated data with shared reporting

• Pane-type pricing that recalculates automatically when a crew updates pane count on-site

If your operation hits two or more of those conditions, you are spending 5 to 10 hours per week in manual workarounds. That is the real cost of staying on generic window cleaning scheduling software.

Who actually builds custom window cleaning management software

Custom window cleaning management software is not for every operator. Here is who it is actually for.

The multi-location commercial operator. You have 15 to 40 commercial accounts, each with 2 to 20 locations. You need a master contract per client, a frequency schedule per location, separate access and contact info per site, and a single invoice that rolls up to the parent. None of the four tools above model this cleanly. You are currently managing it across spreadsheets, email threads, and Jobber custom fields. The time cost is real and it scales linearly with every new account you win.

The franchise with 3 or more territories. Each franchisee needs their own scheduling, crew management, invoicing, and customer database. You need cross-territory reporting for royalty calculations and performance benchmarks. You also need each franchisee to be unable to see each other's customer data. No off-the-shelf window cleaning business software does multi-tenant franchise architecture. You either run separate accounts (no shared reporting) or one account (no data isolation). Custom solves this.

The scaling residential operator. You have 300 or more recurring residential customers. You are running 8 to 12 crews. Your dispatcher is spending two hours every morning manually sequencing jobs. A custom build with proper route optimization reduces that to 15 minutes of review and approval. At scale, the fuel savings and additional jobs per day from optimized routing alone can recover the build cost within 18 months.

The operator building a defensible business, not just a service company. You want proprietary software as a competitive moat: customer portals with before/after photo history, branded crew apps, and invoicing that feels like a product, not a PDF. This is a positioning decision as much as an operational one.

What window cleaning management software needs to do (V1, V2, V3)

V1 (MVP): Core operations

This is the minimum to replace your current stack and stop the manual workarounds.

Recurring job scheduler. Every customer has a frequency rule: monthly, every six weeks, quarterly, annually. The scheduler generates jobs automatically from the rule, 30 to 90 days out. You never manually create a recurring job. When a customer pauses, the system stops generating jobs and resumes from the right point in the cycle when they return. When a customer cancels, all pending and future generated jobs disappear. When frequency changes, the system recalculates forward correctly.

This is the hardest engineering problem in the whole build. More on why below.

Pane-based quote engine. Price is a function of pane count, pane type (standard, French, skylight, solar panel), floor level (ground, second, third or higher), and interior vs. exterior. Build a configurable pricing matrix, not a lookup table. When a crew visits a property to do an initial quote and updates the pane count from the mobile app, the job reprices automatically. For commercial accounts, pricing is stored on the contract and bypasses the matrix.

Multi-crew route optimizer. Feed it job locations, crew start and end depots, time windows, and job duration estimates. It returns the optimal route sequence for each crew. Run it nightly for the next day's schedule. Let dispatchers lock specific stops when a customer has a hard arrival window. Locked stops are excluded from re-optimization. This is a Vehicle Routing Problem. The correct solver is OR-Tools (open source, from Google). A nearest-neighbor greedy algorithm is not route optimization.

Crew mobile app. Today's job list in sequence. Navigation deep-link to each address. Property access instructions, gate codes, preferences. Before and after photo upload per job. Completion confirmation. Works offline with sync when connectivity returns.

Customer portal. Upcoming scheduled visits, invoice history, online payment, and the option to pause or cancel service. That is genuinely all most customers want.

Cost: $90,000-$160,000. Timeline: 10-14 weeks.

V2: Commercial layer

Add this when commercial accounts represent more than 30% of your revenue.

Multi-location contract management. Parent account with child locations. Each location has its own frequency schedule, access instructions, and site contact. Invoices consolidate to the parent. Contract pricing overrides the standard matrix per account.

Chemical and equipment tracking. Log chemical usage per crew vehicle per job. Track supply inventory across vehicles. Flag when a vehicle needs restocking. Useful for compliance documentation on commercial properties.

Before/after photo documentation. Store photos per visit, not per property. Photos link to the job record: property ID, visit ID, before and after sets. Display the most recent visit photos on the property screen. Customers can view their complete photo history in the portal. Keep photos for at least two years for dispute resolution.

Additional cost for V2: $40,000-$80,000.

V3: Franchise and multi-market

Add this when you are licensing the model to franchise operators or expanding into multiple markets with separate P&Ls.

Multi-tenant architecture. Each franchisee gets an isolated environment: their own customers, crews, jobs, and billing. The franchisor sees cross-territory reporting: total revenue by territory, job completion rates, crew utilization, and royalty calculation inputs.

White-label configuration. Each territory can have its own branding on the crew app and customer portal.

Additional cost for V3: $60,000-$100,000.

Where custom window cleaning software projects fail

Custom builds in field service fail in two predictable places.

The recurring job model is bolted on after the fact.

As Gene Marks, small business technology columnist for The Guardian and Forbes who covers field service software extensively, has noted: "The number one reason field service software projects fails is that the recurring job model is bolted on after the core scheduling system is built. Retrofitting frequency rules onto a poorly modeled job table costs three times what building it correctly from the start would have."

This is exactly right. The recurring job logic has to be designed correctly before you write a single line of scheduling code. The data model question is: do you store a job as a standalone record, or as an instance of a recurring rule? The answer is the latter, always. When you store jobs as standalone records and try to add recurring logic later, you end up with duplicated state, inconsistent cancellation behavior, and pauses that break the customer's ongoing cycle.

Route optimization that is not actually optimization.

A nearest-neighbor algorithm picks the closest next stop at each step. It is simple to build. It produces routes that are 20 to 40% worse than a proper VRP solution on a typical 12-stop crew day. That gap translates to 45 to 90 minutes of extra drive time per crew per day. At 8 crews, that is 6 to 12 hours of paid labor and fuel cost every single day.

Operators who build a home-grown routing algorithm and call it "route optimization" do not know it is underperforming until they compare it to OR-Tools output side by side. By then, the system is in production. Google's OR-Tools VRP solver documentation explains this clearly: the Vehicle Routing Problem is NP-hard, and a good solver finds routes within 2 to 5% of optimal in seconds. That is the correct baseline.

According to a 2023 ABI Research field service management report, companies that implement route optimization in field service reduce fuel costs by an average of 10 to 15% and increase jobs completed per crew per day by 8 to 12%. At scale, those numbers recover build cost quickly.

How RaftLabs builds window cleaning software

We start with the data model, not the features. Before writing scheduling code, we diagram the recurring rule structure, the job state machine, and the commercial account hierarchy. These three decisions determine whether the system works at 500 customers or breaks at 200.

Our typical engagement at this scope:

Weeks 1-2: Discovery and architecture. We map your current operation: how recurring jobs are created today, how commercial accounts are billed, how dispatchers build routes, what the crew needs on-site. We produce a data model and a scope document with phased deliverables.

Weeks 3-10: MVP build. Recurring scheduler, route optimizer, quote engine, crew app, customer portal. Weekly builds deployed to a staging environment. You test with real job data.

Weeks 11-14: QA, refinement, and production migration. Migrate existing customer and property records. Run parallel operations for two weeks before full cutover.

We use PostgreSQL for the core data (jobs, customers, properties, contracts, routes), OR-Tools for route optimization, React for the web dashboard and customer portal, and React Native or a progressive web app for the crew mobile app. We do not over-engineer the stack. The goal is a system your team can use on day one and your next developer can maintain without archaeology.

If you are at 8 or more crews, managing commercial multi-location accounts, or running a franchise that generic window cleaning business software cannot model, here is what the first 90 days with RaftLabs looks like: a 30-minute discovery call to confirm the scope, two weeks of architecture and scoping, and a working MVP in staging by week 12. No retainer, no strategy decks. Scope, build, ship.

Tell us what you're running today and where it's breaking down.