Window Cleaning Business Automation in 2026: Pane Counts, Routes, and Recurring Contracts

Window cleaning is the most quotable trade there is, and most companies still do not quote on the phone

Almost every service trade has a legitimate excuse for refusing to price over the phone. Plumbing hides behind walls. Roofing needs a ladder. HVAC needs a load calculation. Window cleaning has no such excuse. The price is a function of how many panes there are and how high off the ground they sit, and the customer can count their own windows while standing in their kitchen.

That means the companies still answering "we'd have to come look at it" are losing jobs to companies that say "for a two-story house with about thirty panes, exterior only, that's $265, and I can do Thursday morning." The second answer wins even when the first company does better work, because the customer was making a small, low-risk decision and just wanted it made. As of August 2026, that reflex — deferring an answerable price to a site visit — is still the most common self-inflicted wound in the trade.

The window cleaning operations that scale past two vans have generally done three things: separated the phone from the squeegee, turned pane counting into a scripted intake, and stopped rebuilding their recurring commercial routes by hand every month. This guide covers the whole stack — answering, quoting, add-ons, access method, routing, seasonality, commercial contracts, billing, and reviews — mapped to the Run with Jarvis platform.

Why window cleaning loses calls in a specific pattern

The work is physically incompatible with answering. A technician on a ladder at the second story, or running a water-fed pole with both hands overhead, cannot take a call. Unlike a trade where someone can step out of a crawlspace, there is no safe way to interrupt.

The tickets feel small enough to lose. A $185 residential exterior job feels like a rounding error until you notice you are losing four of them a week, and each one of those customers had gutters, screens, and a twice-a-year recurring interest behind it.

Demand is impulse-driven and short-lived. Nobody has a window cleaning emergency. The customer looked at their own glass on a sunny morning, got annoyed, and called two or three companies. That impulse has a half-life measured in hours, which is the decay described in the speed-to-lead guide. A callback the next afternoon frequently reaches someone who already booked.

The spring spike is brutally concentrated. The first genuinely warm, dry, bright week produces more inbound volume than the preceding two months combined, because everyone's windows became visible on the same day. A single person answering one call at a time drops most of it. The pattern is predictable even though the exact week is not.

Layer 1: An AI receptionist that answers while the crew is on a pole

The foundation is KeyBot, the AI phone agent. It answers on the first ring in English or Spanish, at any hour, runs your pane-counting intake, quotes from your own price rules where quoting is safe, books the job, and texts a confirmation.

Parallel answering is the property that matters most here. A bright Saturday in April does not produce calls politely spaced out. It produces nine calls in two hours. One person handles one and loses eight. The AI holds all nine simultaneously and every caller gets a number and a slot.

The second property is that it never gets bored of the interview. Window cleaning intake is repetitive by design — pane count, story count, interior or exterior, screens, tracks, sills, hard water spotting, roof access, pets, gate codes. A human running that thirty times a day starts trimming questions by mid-afternoon, and the trimmed question is always the one that changes the price or the duration.

A message-taking answering service solves the wrong half of the problem, because it hands you a name and a number without any of the fields that make a quote possible. A phone menu is worse: an impulse caller does not press three.

Layer 2: Quoting by pane count and story height

Residential window cleaning pricing is close to arithmetic. The intake needs four things.

Pane count. Ask the customer to count windows, and then clarify what a pane means in your pricing — whether a double-hung counts as one or two, whether grids and storm windows change the number, whether French doors and transoms are counted. Customers undercount by habit, so a well-built script asks room by room for anything above a certain size rather than asking for one total.

Story height. Ground floor, second story, and anything above that are different labor and different risk. Second-story exterior is where ladder work or a water-fed pole enters, and third story is where the conversation changes entirely.

Scope: interior, exterior, or both. Exterior-only is the fast, high-margin job. Interior work means moving furniture, protecting floors, someone being home, and a longer duration per pane. Both is not simply the sum of the two prices, because the crew is already on site — and the intake should present the combined price as the natural upgrade rather than waiting for the customer to ask.

Condition. New construction with stucco or paint overspray, hard water etching, or years of neglect are not standard-rate jobs. A caller who says "they've never been cleaned since we bought the house in 2019" has just told you this is a first-clean at a premium, not a maintenance clean.

Two guardrails on automated quoting:

Quote only where quoting is safe. Rope access and high-rise work, atriums and vaulted interior glass, restoration and stain removal, historic leaded glass, and anything requiring lift equipment should route to a site estimate. A number that has to be revised upward on arrival costs more trust than the call was worth.

Capture the inputs even when you do not quote. If the call ends in an estimate visit, the estimator should arrive already knowing the pane count, story height, and stated problem. That is the difference between a fifteen-minute estimate and a forty-minute one, which across a week is most of a day of capacity. The general pattern is in how an AI receptionist quotes from your price list.

Layer 3: Add-ons are the whole margin story

Window cleaning has an unusually strong add-on economy, and almost all of it is lost when the intake does not ask.

• Screens. Removing, cleaning, and reinstalling screens roughly doubles the time per opening. It must be a separate priced line, not an assumed inclusion, or you will lose an hour you never charged for.
• Tracks and sills. Detailing tracks is the add-on customers notice most, because a clean pane above a filthy track looks unfinished. It is also slow, so it needs its own price and its own duration.
• Hard water and mineral stain removal. This is a restoration service with its own chemistry and its own failure modes. Never quote it blind, and never fold it into a standard rate.
• Gutter cleaning. The single most natural cross-sell in this trade, because the crew is already on a ladder at roof level.
• Solar panel cleaning. Same access, same equipment, entirely separate service line, and increasingly common on the same roofs.
• Skylights, chandeliers, and mirrors on interior jobs.
• Post-construction first cleans, which are premium-priced and often come from builders rather than homeowners.

The reason this is an automation topic and not a sales-training topic is that the ask has to happen every time. A technician who remembers to mention gutters on one out of five jobs is producing a random ticket. An intake script that asks about gutters, screens, tracks, and solar on every single call produces a higher average ticket with no additional marketing spend. If you want the rule stated in plain language rather than built as a form, that is the sort of thing the Jarvis brain exists to handle.

Layer 4: Access method drives crew size and duration

Two houses with identical pane counts can be very different jobs depending on how you reach the glass.

Ladder work needs stable footing, clearance, and often two people for safety on second-story work. Landscaping, decks, air conditioning units, and steep grade all interfere with ladder placement, and none of that is visible from a pane count.

Water-fed pole work is faster and safer from the ground but has a height ceiling, needs a water source and sometimes purified water handling, and is unsuitable for heavy staining that needs contact scrubbing.

Rope access and lift work are a different business entirely, with different certification, insurance, and pricing. Those should never be quoted by phone.

The practical implication is that your intake must capture the physical conditions, not just the counts: height, obstructions, whether there is a spigot, whether there is a locked gate, whether there is a dog, whether anyone needs to be home, and where the van can park. Those fields ride along with the appointment in GetTimePad, so the crew is not calling the office from the driveway to ask basic questions.

Duration modeling matters just as much. A thirty-pane exterior-only ground-floor job is a couple of hours. The same pane count with interiors, screens, and tracks is most of a day. A calendar that treats both as a generic "window cleaning" appointment will happily triple-book a crew with three small jobs that are collectively eleven hours. The booking mechanics are covered in how AI appointment booking works.

Layer 5: Route density, the quiet profit lever

Window cleaning has one of the worst drive-time-to-revenue ratios in the trades if you are careless. A $185 job twenty-five minutes in the wrong direction is close to break-even after drive, setup, teardown, and drive back.

Density is built at two moments.

At booking. When a flexible residential customer calls, the system should offer the days a crew is already working near them first. Most residential window cleaning customers genuinely do not care whether it is Tuesday or Thursday. Steering them into an existing cluster is worth more than any after-the-fact routing optimization, because once the stops are scattered the day is already lost. This only works when answering and scheduling are the same system — an answering service that cannot see tomorrow's route cannot steer anyone into it.

At dispatch. IntelliDrive sequences the day with GPS routing, tracks the crew, and sends automatic arrival texts. Those texts eliminate an entire category of zero-revenue inbound call: the customer asking when you are coming. For an operation running five or six stops a day, that is a real chunk of phone volume gone. IntelliDrive is the CRM, dispatch, and invoicing layer.

The nearly identical density logic in an adjacent exterior trade is worth reading in the pressure washing automation guide, since many window cleaning companies run both services off the same route.

Layer 6: Recurring commercial work is where the actual company lives

Residential window cleaning pays this month's bills. Commercial route work builds an asset.

Storefronts on weekly or biweekly cycles, restaurants, medical and dental offices, car dealerships with enormous glass frontage, office buildings, and HOA common areas are recurring, scheduled, and priced on contract rather than negotiated per visit. The revenue is predictable, the routes are dense by construction, and the customer is not price-shopping every time.

Commercial work rewards administrative discipline and punishes sloppiness in ways residential does not. A property manager with eleven locations cares about consistency, documented completion, invoices that arrive on schedule, and a vendor who does not need chasing. If your operation runs on a personal cell phone and a paper route sheet, you cannot hold that account at scale even if your glass work is excellent.

The stack handles it with:

• Recurring schedules that generate the visits without anyone rebuilding a weekly storefront run every month.
• Dispatch records documenting arrival and completion per location, which is what a multi-site manager actually wants to see.
• Contract billing on a cadence rather than an invoice per visit, with QuickBooks sync behind it so the bookkeeping is not a separate evening job.
• A customer portal so the manager can look up their own history instead of emailing you for it.
• A CRM record per location, not per company, because a chain's eleven storefronts have eleven different access notes, alarm situations, and contact people. That multi-site dispatch pattern is covered in CRM and dispatch software for multi-tech operations.

One scheduling constraint most tools handle badly: a lot of commercial glass is cleaned before opening or after close. If your calendar treats 5 AM to 8 AM or 9 PM to midnight as after-hours rather than billable capacity, and your crews split between early commercial routes and daytime residential, both shifts need to exist in one schedule so nobody gets dispatched to an 8 AM residential job after a 4 AM storefront run.

What each plan gives a window cleaning company

Every plan is month-to-month with zero setup fees, unlimited users, and no per-call fees.

The arithmetic, on your own numbers

Assume a two-van operation books 95 jobs a month at a $215 blended average across residential exteriors, interior-and-exterior jobs, and storefront route stops. That is $20,425. Assume the phone rings 240 times a month and, because both crews are on glass all day, 70 calls go unanswered.

Assume conservatively that only 8 of those 70 would have closed. At $215 that is $1,720 in recovered monthly revenue. Core at $500 plus roughly 220 overage minutes at $0.45, which is $99, comes to $599. Recovered revenue is nearly three times the cost before counting the add-on capture, the route density, or the recurring billing.

Then run the version that reflects April. If your first warm month produces 420 calls and you currently answer forty percent of them, the recovered revenue in that single month exceeds several months of subscription. Seasonality is why this trade tends to see a return faster than steadier trades.

The minute math. A window cleaning quote call is short — the interview is a handful of counts, so three to four minutes is typical. At three and a half minutes, Core's 500 included minutes cover roughly 140 calls, with the remainder at $0.45. A company taking 240 calls a month runs about 340 overage minutes, roughly $153. If your volume routinely runs past about 1,100 minutes a month — which a spring peak can easily do — Pro's 1,000 minutes at a $0.40 overage becomes the cheaper structure, and it adds attribution. The full framework is in the AI receptionist ROI guide.

When attribution starts mattering

Window cleaning is an advertising-heavy category, and call volume is a deceptive metric because broad cleaning search terms attract a lot of non-buyers — people wanting gutter guards installed, people wanting a car detailed, people wanting a quote on replacement windows rather than clean ones.

Pro adds Dynamic Number Insertion, so each call ties to the campaign, keyword, and ad that produced it, and booked jobs feed back to Google Ads so bidding optimizes toward revenue rather than ring count. Transcription and lead scoring tell you whether a keyword is producing thirty-pane houses or people asking your hourly rate. Judge campaigns on cost per booked job, never on raw calls.

Reviews are the entire residential funnel

First-time residential window cleaning customers have no prior relationship and no referral source. They search, they scan star ratings and review counts, they call the top two or three. Review volume is effectively your top of funnel.

Review automation asks right after a completed job, while the customer is standing in a room that suddenly looks brighter. That timing is the whole trick, and it never happens consistently when it depends on someone remembering. The approach is in how to get more customer reviews.

An adjacent recurring-service stack worth comparing is the carpet cleaning automation guide, because the residential intake and rebooking cadence rhyme closely with window work.

For general small business planning references, the Small Business Administration publishes material at sba.gov. For anything involving employees on ladders, your state occupational safety requirements will shape your operation more than any national resource.

A sensible rollout order

Answering and phone quoting first. That is where the recoverable revenue is, and it changes your day immediately because the phone stops competing with the pole.

Then booking with honest durations, so interiors, screens, and tracks stop wrecking the schedule.

Then dispatch, routing, and arrival texts, which tighten density and remove status calls.

Then recurring commercial schedules and contract billing, which is the piece that lets you take on multi-site accounts without adding office staff.

Then review automation.

Only after all of that, if you are advertising, move up for attribution or campaign management. Measuring ad performance precisely while still missing seventy calls a month only tells you how much you are losing.

The short version

Window cleaning is the most quotable trade in the field, and most companies still refuse to quote. That is the gap.

Answer every call in parallel. Quote from real per-pane and per-story rules. Ask about screens, tracks, gutters, and solar on every single call. Model duration honestly by scope. Cluster the route at the moment of booking. Let recurring storefront and HOA routes generate and bill themselves.

If you want that modeled against your own pane pricing, call volume, and commercial-to-residential mix, reach out or review the plans on the pricing page.