Context Theory Get your growth audit

Industry

Where home services enquiries are lost before the van is booked

Home and home improvement runs $8.33 a click against a $5.42 average. Most of that spend never reaches a booked job.

Home services is the only one of these industries where winning the enquiry and being able to serve it are separate problems. Everywhere else the constraint is somebody's attention; here it is a technician-hour and a vehicle, both of which are finite and already scheduled. That changes what a response system is for. A provider who answers faster without fixing dispatch converts more enquiries into bookings they then have to move or cancel, which costs more standing in a neighbourhood than the missed call would have. It is also the only one where the customer routinely contacts several providers in the same minute rather than in sequence, so the contest is decided by who can confirm a name and a window first — not by who is best.

Home services has the most expensive click of any category on this site except law — $8.33 against a $5.42 all-industry average — and the shortest window in which that click is worth anything. Someone standing in front of a failed water heater is not evaluating providers; they are working down a results page until somebody confirms a name and an arrival time. Everything the business spent to appear on that page is settled in the next few minutes, usually without the losing providers ever learning they were in a contest.

That is why voicemail is more expensive here than anywhere else. In most industries a message asking for a callback is a delay. In this one it is a decline, because the caller's next action is already dialling the next number, and by the time the callback happens somebody else's van is on the way. It is also why answering in this industry cannot depend on a person being available: at the small-operator end, the person who would answer is the person under the sink.

The constraint that makes this industry genuinely different, though, is on the other side of the booking. Everywhere else the scarce resource is attention. Here it is a technician-hour and a vehicle, both already scheduled. A response system that improves speed without knowing what the dispatch board looks like will book work the business cannot serve, and a cancellation call costs more in a neighbourhood business than the enquiry was worth. Speed and capacity have to be solved together or the improvement shows up as broken promises.

The two-customer problem underneath all of this is the most common configuration error found here. The emergency customer buys an arrival time, decides in minutes and barely checks price. The planned customer — a replacement system, a new roof — collects quotes over weeks and chooses the provider who explained the decision best. One intake script cannot serve both. Run the emergency script on a planned job and it reads as pressure; run the planned script on an emergency and somebody else has already been booked while you were qualifying.

Two things sit outside the business's control and have to be designed around rather than solved. Permits and inspections put a municipal schedule on the critical path between finishing the work and getting paid, which means the follow-up sequence has to survive a gap nobody owns. And demand arrives in bursts — a freeze, a heatwave, a storm — where the phone rings more in an afternoon than in the previous fortnight. Staffing for that peak is uneconomic, which leaves systems as the only available answer to the highest-intent enquiries the business will see all year.

The most recoverable money in most of these businesses is not in the missed calls at all. It is in the estimates that were given and never chased. A quote for planned work sits between a visit that has already been paid for in time and fuel and a decision that takes weeks, and in most operations nothing follows it up because the person who wrote it is on another job. That pipeline is already qualified, already visited and already priced, and it is almost never in any report.

The first useful move is the same one it is everywhere and it costs an afternoon: call your own number during working hours, call it again at seven in the evening, submit your own web form, and tap the call button on your own map listing. Time all four. In most home services businesses at least one of those paths arrives nowhere, and it is very often the one attached to the most expensive spend in the company.

Who decides, and what they are deciding with.

The decision-maker is the owner-operator, and in most of these businesses that person was a technician first. They read a marketing proposal the way they would read a quote for a tool: what does it do, what does it cost, and what happens when it breaks. They are unusually resistant to anything that cannot be explained mechanically, and unusually receptive to anything framed as capacity — because capacity is the language the whole business already runs in.

Two different customers wear the same clothes here. The emergency customer decides in minutes, contacts several providers simultaneously rather than sequentially, barely checks price, and buys availability — they are purchasing an arrival time, not a service. The planned customer, replacing a system or re-roofing, collects several quotes over weeks, compares them on paper, and is influenced heavily by whichever provider explained the decision most clearly rather than by whichever was cheapest. The same intake script cannot serve both: run the emergency script on a planned job and it reads as pressure selling, run the planned script on an emergency and the customer has already booked somebody else while you were qualifying them.

Where acquisition comes from, and where each one leaks.

Acquisition channels
Channel What it costs you, beyond the money
The phone, ringing nowStill the dominant inbound path for anything urgent, and the one least likely to be instrumented. A call that goes to voicemail during a job is not a deferred enquiry in this category; the caller is already dialling the next number on the results page.
Google Business Profile and the map packProximity, review count and recency decide placement, and the enquiry arrives as a tap-to-call rather than a form. It is the highest-intent channel in the industry and the one least likely to appear in any acquisition report, because nothing writes it down.
Lead marketplacesAngi, Thumbtack, Networx and HomeAdvisor sell the same enquiry to several contractors by design. The economics only work for whoever responds first, which means the platform is effectively charging every provider for the privilege of losing a race.
Paid search$8.33 a click against a $5.42 all-industry average, and the intent is frequently an emergency, which is the most valuable and least patient traffic there is. A click this expensive landing on a page with a form and no phone number is the most common configuration error in the category.
Repeat customers and the streetNeighbourhood word of mouth, a truck seen in a driveway, the sticker left on the water heater. It converts at rates no paid channel approaches and is almost never asked for deliberately, so it grows or shrinks without anybody deciding.

What fast means here.

The window here is the length of a phone call. Someone standing in front of a failure works down the results page and stops at the first provider who confirms a name and an arrival window — so the contest is usually over inside a few minutes, and the providers who lose it are never told they were in it. That is true whether or not anybody picks up, which is what makes voicemail so expensive in this category specifically: a message asking for a callback competes with a technician who is already on the way. The planned side of the business runs on a completely different clock, and applying the emergency target to it produces a quoting process that feels like pressure.

The operational bottlenecks.

Why the acquisition problem persists
Constraint The mechanism
The people who could answer have their hands inside a wallIn a small operation the technician, the dispatcher and the owner are frequently the same person, and the hours when enquiries arrive are exactly the hours when that person is unreachable. This is not an attention problem that discipline fixes; it is a physical one, and it is why answering in this industry has to happen without a human being available.
Dispatch capacity, not demand, is the real ceilingA booking that cannot be served is worse than an enquiry that was never answered, because it consumes a slot, generates an expectation and then produces a cancellation call. Any response system built here has to know what the board looks like before it promises a window, which is a harder integration than the response itself.
Permits and inspections put a third party on the critical pathA large share of this work cannot be completed until a municipal inspector has been and gone, and their schedule belongs to neither party. Revenue recognition, final payment and the review request all sit behind that visit, which is why the follow-up sequence in this industry has to survive a gap it does not control.
Reviews are the acquisition channel and nobody owns askingPlacement in the map pack moves with review count and recency, so review generation is not reputation management here — it is the acquisition system. It is also the task most reliably dropped, because the moment to ask is the moment the technician is packing the van to get to the next job.
Demand arrives in bursts that exceed any staffing planA heatwave, a freeze, a storm. Weeks of ordinary volume are followed by days where the phone rings more in an afternoon than in the preceding fortnight, and the enquiries lost in those hours are the highest-intent ones the business will see all year. Staffing for the peak is uneconomic, which leaves systems as the only available answer.

The software this industry runs on.

Common stack, by category
Category Commonly used Where it leaks
Field service managementServiceTitan, Housecall Pro, Jobber, FieldEdge, Service FusionBuilt around the job, which begins after somebody has already agreed to buy. The enquiry that never became a job is outside the system entirely, so the reporting describes the work that happened and is silent about the work that was lost.
Call handling and trackingCallRail, RingCentral, answering services, voice systemsAnswering and capturing are treated as the same job and are not. A message with a name and a number, taken while the board is full, is a callback promise nobody has scheduled and the caller has already stopped waiting for.
Lead marketplacesAngi, Thumbtack, Networx, HomeAdvisorEach one notifies in its own app, and the enquiry is simultaneously with competitors. Response time is measured per platform, if at all, and the provider is paying for leads whose value decays faster than any of the notifications are designed to move.
Reviews and local presenceGoogle Business Profile, Podium, NiceJob, BirdeyeThe request has to be made at the end of a job by somebody already late for the next one. Automation here is straightforward and the reason it is so often absent is that nobody owns the step, not that the tooling is missing.
Estimating and invoicingQuickBooks, Housecall Pro, ServiceTitan estimatingThe quote for a planned job sits between a visit and a decision that takes weeks, and nothing chases it. Unaccepted estimates are the largest recoverable pipeline in most of these businesses and the least likely to be in any report.
Dispatch and routingNative scheduling in the field service platform, Google Maps, whiteboardsThe board is the true constraint on the business and it is almost never visible to whatever is answering the phone. That gap is where an over-promised window becomes a cancellation.

Your own numbers.

The cost per click is set to the published home and home improvement average, so the click figure below is what a benchmark budget buys in this category rather than in an average one. Revenue is yours to enter.

Marketing budget benchmark

Benchmark monthly marketing budget, lower bound
$2,500
Arithmetic
Step Working Result
Revenue a month $1,000,000 ÷ 12 $83,333
Lower bound — 3% of revenue $83,333 × 0.03 $2,500
Upper bound — 5% of revenue $83,333 × 0.05 $4,167
Clicks the lower bound buys, at your average CPC $2,500 ÷ $8.33 300

This assumes

  1. will not tell you what to spend, because a benchmark describes what comparable businesses allocate rather than what your market requires of you

    will return the lower bound of the published band as the headline, and show the upper bound directly beneath it so the range stays visible

  2. will not assume the budget goes to paid search, or that it is spent competently once it is allocated

    will convert the lower bound into clicks at your industry's average cost per click, so a share of revenue arrives as something purchasable

  3. will not adjust for margin, growth stage, or how much of your revenue is repeat business rather than newly acquired

    will state plainly that a low-margin business and a high-margin one cannot sensibly spend the same share of revenue, and that the band is a starting position rather than an answer

2026 SMB marketing budget survey · a $1M business ≈ $2,500–$4,200/mo · verified

The full version, with the reasoning behind every assumption in it, is at Marketing budget benchmark.

Where automation has actually been adopted.

This industry has adopted automated call handling faster than any other on this site, and for a straightforwardly physical reason: the alternative was voicemail. Answering services and voice systems that capture the job type, the address and the urgency are now common at the small-operator end, and scheduling and routing optimisation are standard inside the field service platforms. What has not been adopted is anything that touches the technician mid-job, because a person wearing gloves on a roof is not going to interact with software, and every attempt to make them has failed on contact with the work. The visible failure mode is specific to this category too: a system that books a job the board cannot serve. Speed without capacity awareness produces cancellations, and a cancellation in a neighbourhood business is more expensive than the enquiry was worth.


The five systems, applied here.

These are built and operated under a retainer. Which of them applies to a particular business, and in which order, is what the audit establishes.

What this industry changes the answer to.

10 questions on this site whose answer is different here — what is regulated, what may be automated, and what the buyer is actually deciding on. See them all.

By market.

1 market documented — licensing, local terms, local resources. See them all.

Questions this raises.

We already use an answering service. Is this the same thing?

An answering service solves availability and stops there. It takes a name and a number while the board is full, which produces a callback promise nobody has scheduled to a caller who has already stopped waiting. The question this addresses is what happens between the call being captured and a window being confirmed.

Won't answering faster just book work we cannot service?

It will, if speed is built without capacity. That is the specific failure this industry produces and it is why the dispatch board has to be visible to whatever is answering. A confirmed window the business can keep is the deliverable; a fast reply that generates a cancellation call is worse than the missed call was.

Most of our work comes from repeat customers and referrals. Does this apply?

It applies to the part of that you are not asking for. Map pack placement moves with review count and recency, so review generation is an acquisition system rather than reputation management, and it is the step most reliably dropped because the moment to ask is the moment the technician is loading the van.

What about the quotes we give that go nowhere?

That is usually the largest recoverable pipeline in the business and the least likely to be in any report. The work has already been done — the visit, the measurement, the pricing — and in most operations nothing chases it because the person who wrote it is on another job by the next morning.

Does any of this survive a storm week?

That is the week it is for. Demand in this category arrives in bursts that no staffing plan can economically absorb, and the enquiries lost in those hours are the highest-intent ones the business sees all year. The point of a system is that it behaves identically at ordinary volume and at four times ordinary volume.

METHOD

Every figure below carries its source and the date it was verified. Nothing on this page is asserted.

The numbers on this page.

Datapoints
What Value Specific to
Home & home improvement CPC$8.33This page
All-industry average search CPC$5.42Category-wide
Firms that never responded to a web enquiry at all23%Category-wide
Close rate — response under 5 minutes vs over 24 hours32% vs 12%Category-wide
Customers who buy from the first responder78%Category-wide

LocaliQ / WordStream Search Advertising Benchmarks 2026 · Google + Microsoft Ads, 20 industries · Apr 2025–Mar 2026 · verified

Oldroyd, McElheran & Elkington, "The Short Life of Online Sales Leads", Harvard Business Review (March 2011) · 1.25M inbound leads across 2,241 US firms · verified

Optifai speed-to-lead benchmark · n=939 companies · Q2 2025–Q1 2026 · verified

Multi-study aggregate · verified

What is specific to this page.

Evidence
Kind Claim Check it against
ResponseAn enquiry in this category is routinely placed with several providers in the same minute rather than in sequence, so the response contest is decided before any provider knows it has started. The losing providers are never told they lost, which is why the problem persists in businesses that believe they answer everything.The lead-distribution model published by Angi, Thumbtack and HomeAdvisor, which sends one homeowner enquiry to several contractors by design
WorkflowWinning an enquiry and being able to serve it are separate constrained problems: the ceiling is a technician-hour and a vehicle, both already scheduled, rather than anybody's attention. A response system that does not read the dispatch board will book work the business then has to cancel.The dispatch board in ServiceTitan, Housecall Pro or Jobber, where capacity is modelled as a scheduled resource rather than as a queue
LicensingWho may perform the work is gated by trade licences issued per state and frequently per municipality, and most states require the licence number to appear in advertising. A marketing programme built without the licence data attached is a compliance problem before it is a performance one.State contractor licensing boards, and the licence-number-in-advertising requirement most of them publish
RegulationA large share of this work cannot be completed or invoiced until a municipal inspector has attended, putting a third party with its own schedule on the critical path between finishing a job and being paid for it. Follow-up sequences here have to survive a gap neither party controls.Municipal building department permit and inspection schedules
Buying behaviourTwo customers with opposite behaviour arrive through the same channels: the emergency buyer decides in minutes, contacts several providers at once and purchases an arrival time rather than a service, while the replacement buyer collects quotes over weeks and chooses on the clarity of the explanation.The multi-quote comparison flow that every roof and HVAC replacement runs through, against the same firm's same-day call log
SoftwareThe system of record is a field service platform built around the job, which begins only after somebody has agreed to buy. The enquiry that never became a job sits outside it entirely, so the reporting is accurate about work performed and structurally silent about work lost.The data model of ServiceTitan, Jobber or FieldEdge, where the job — not the enquiry — is the root record

Each row would be wrong on another industry's page. Where a sourced figure exists it is in the table above instead; these are the constraints that shape the work and do not happen to be numbers.