Average Cost Per Lead by Industry: Home Service Trades

What a lead actually costs in roofing, HVAC, plumbing, painting and 12 more trades, with published figures and the two numbers that explain the spread.

By Gavin Sevastian · Updated August 4, 2026 · 12 min read

Below are the published cost per lead figures for sixteen home service trades, plus the two numbers underneath them that explain why one trade's leads cost roughly five times another's. The headline is not what most people expect: the trade paying the most per click is not the trade paying the most per lead. Painting has the dearest clicks of all sixteen and still produces leads far cheaper than roofing does. Once you see why, the table stops being trivia and becomes something you can point at your own account. One caution before you scroll: every benchmark on this page is published in US dollars. Every dataset here that states a country is American; two smaller vendor datasets do not state one. There is no comparably documented public Canadian table, and that gap matters more than it sounds.

Cost per lead by industry: the full home services table

Sorted most expensive to cheapest, with the category median pinned at the bottom for reference.

Roofing sits at the top at $228.15 and pools at the bottom at $45.15, a spread of about five times. Notice that the cost per click column does not run in the same order as the cost per lead column, which is the whole point of this page. If you want the detail for one trade, there are separate guides for roofing, plumbing, HVAC, electrical, painting, garage doors and landscaping.

Where these numbers come from, and what they are not

The table comes from LocaliQ's 2025 search ad benchmarks for home services, built from 3,211 US search campaigns running between April 2024 and March 2025, with at least 103 active campaigns behind every category. The important detail is buried in their methodology note: these are medians, not averages. That matters, because a handful of enormous accounts would otherwise drag every figure upward and the table would describe nobody.

There is no pest control category in the source. Sixteen subcategories, and pest control is not one of them. Rather than borrow a number from a blog with no stated sample or period, the honest answer is that the gap exists, and the structural points on this page apply to that trade the same way they apply to the rest. Our pest control lead guide covers the trade itself.

What the sample can and cannot support

The LocaliQ report covers both Google and Microsoft search advertising rather than Google alone. It also reflects campaigns run through one publisher's customer base, not a random sample of every home service advertiser in the United States. Businesses that never advertised, accounts managed elsewhere, and contractors that stopped before the reporting window are outside the sample. That selection does not make the numbers unusable. It limits the claim to what similar active campaigns produced during that period.

The category labels are broad. "Construction and contractors" can contain work with different tickets, sales cycles, and search intent. "Doors and windows" can blend repair and replacement. A single median across that mix cannot tell you whether an emergency service line and a considered installation should share one target. The table supports comparison between broad categories, not a promise about one service inside them.

The conversion definition is another boundary. Publishers can standardize reporting fields more easily than they can standardize what every advertiser counted as a lead. A form submission, a qualified phone call, and a tap on a phone number can all appear in conversion reporting if accounts define them that way. The stated campaign minimum improves the stability of each category median, but it does not make the underlying lead definitions identical.

That is why the source belongs at the beginning of the conversation rather than the end. It gives a documented period, population, currency, and trade spread. Your own account still has to supply a stable lead definition, the split between known-brand and new-customer demand, and the link from enquiries to booked work.

The two sources everybody quotes are the same company

Home services is expensive relative to the wider market. The category median of $90.92 sits well above the $66.69 all-industry cost per lead in the current WordStream Google Ads benchmarks, though those are different reports covering different windows, so read it as a rough position rather than a precise gap.

Worth knowing before you go looking for a second opinion: WordStream says it and LocaliQ are part of the same company, and Gannett acquired WordStream in 2018. Treat them as one publisher rather than independent corroboration, even though the reports above cover different samples and windows. One more thing from the newer report, because it contradicts the line every ad seller repeats: 2026 recorded the first all-industry cost per lead decrease in five years.

Cost per lead comes from two numbers, not one

The formula, and why it beats the table

Cost per lead is what you pay for a click divided by the share of clicks that turn into an enquiry. That is arithmetic, not opinion. It has one consequence that most contractors never get told: there are exactly two ways to end up with an expensive lead, expensive clicks or a low conversion rate, and they have nothing in common. A contractor who knows only his cost per lead knows the symptom and not the disease. If you want the click side on its own, that is covered in what Google Ads actually cost contractors.

One number is set by your competitors. The other is set by you.

This is the asymmetry that makes the formula useful. Click price comes out of an auction against other advertisers and is also influenced by your ad quality. Conversion rate is shaped by the searcher's intent and by things inside your business: whether the search lands on a page about that specific service or on a homepage, whether the phone gets answered and how fast, how much the form asks for, how the page behaves on a phone, and whether the offer is clear.

The half you can move is the half almost everyone ignores. For scale, Unbounce puts the median landing page conversion rate at 6.6% across 464 million visits, and they explain their choice of a median well: the definition of a conversion varies so much from one page to the next that an average stops describing anything.

Why the columns in that table do not divide out

Here is the part most pages quietly hope nobody checks. Each column in the trade table is computed independently across thousands of campaigns, so dividing the published click price by the published conversion rate does not reproduce the published cost per lead. Roofing is the clearest case: $10.70 divided by 3.70% comes to $289, against a published $228.15, an overshoot of about 27%.

The mechanism is still right. Cost per lead really is those two numbers multiplied, exactly, for one account over one window. It is just not exact across a median of medians.

Roofing versus painting: dearer clicks, cheaper leads

Painting pays the most per click and still gets leads for less

Painting carries the highest click price of all sixteen trades at $13.74, about 28% dearer than roofing's $10.70. Its leads come in at $138.38, roughly 39% cheaper than roofing's $228.15. Every bit of that reversal sits in the conversion rate: 10.80% against 3.70%, a gap of nearly three times. Reason from click price alone and you get this exactly backwards.

Both trades are worth reading in their own right: painting leads and roofing leads behave differently at almost every stage, not just in the auction.

The same reversal in the other direction

General contracting has the cheapest clicks in the entire table at $5.31 and the third most expensive leads at $165.67, on a 2.61% conversion rate. Cheapest clicks, near worst leads. Doors and windows is the third case: a mid pack click price of $8.76, a 4.41% conversion rate, and $200.34 per lead. Three trades, one lesson, and it runs against the intuition every time.

LocaliQ's own commentary adds a related note that is worth keeping in mind before you assume high ticket work is always the expensive end: the largest click price increases they observed showed up in lower cost update categories like painting rather than in big ticket categories like pools.

What this means when you compare yourself to another trade

Two contractors comparing cost per lead over a beer are usually comparing two different businesses, not two different levels of competence. A plumber's lead is cheap partly because plumbing emergencies convert. A roofer's is dear partly because a roof is a decision, not a phone call. The only comparisons worth making are against your own trade, and far more usefully, against your own account last quarter with the same definitions in place.

What actually explains the spread: how long the decision takes

Emergency work converts. Considered purchases do not.

A blocked drain at nine at night is one call, now. A roof replacement is three quotes, possibly an insurance claim, possibly financing, a spouse who has to agree, and a season it has to happen in. Deliberation suppresses conversion rate, and a suppressed conversion rate multiplies the click price into a large cost per lead.

The data lines up with that cleanly. The best converting categories are the small, urgent, low consideration jobs: cleaning at 17.65%, window cleaning at 13.58%, handyman at 13.45%. The worst are the big considered ones: general contracting at 2.61%, roofing at 3.70%, doors and windows at 4.41%. A broken spring is an urgent repair, which is part of why garage door leads behave differently from a landscaping project somebody has been thinking about since March.

Seasonality, financing, and the second decision maker

Three multipliers push on the same effect. A trade with a narrow installation window compresses demand into the months when every competitor is bidding at once. A job that needs financing adds a step between the click and the call. A job a couple has to agree on adds a person who was not the one searching. None of these are account problems, and none of them get fixed with a bid change.

A high cost per lead is not evidence of a badly run account

Roofing costing more than painting is a characteristic of the demand, not a failure of the advertising. This is worth being firm about, because "your cost per lead is too high" is the most common opener in a cold agency pitch, and on its own it is worthless. Without the trade attached, and without knowing what share of those leads came from people searching the business by name, it is a sentence that could be said to anybody.

Branded or non-branded: the split that moves the number most

Everything above this line compares one trade against another. This section compares two campaign types inside the same trade, and the gap it opens is bigger than anything in the sixteen-row table.

What the split looks like in published contractor data

The clearest public dataset on this comes from SearchLight Digital, which reports on HVAC and plumbing Google Ads cost per lead across 816 contractors, 8,077 campaigns and $14.88M in spend over a single month in 2026. Their blended cost per lead is $104. Underneath that number, branded search comes in at $34 and non-branded search at $149.

The blended figure describes neither of them. It is an average of two things that behave nothing alike, weighted by however the budget happened to be split that month.

Note these are spend-weighted figures, not medians, so they are measured differently from the trade table further up and should not be blended with it.

Now the line that makes this table worth publishing. The cost per lead gap between branded and non-branded is 4.4 times. The cost per paying customer gap is 7.7 times, $104 against $804. Cost per lead understates the real difference, because the cheaper leads also booked at a higher rate. Any decision made by ranking campaigns on cost per lead alone flips the moment booking rates differ, and booking rates always differ.

The same pattern in roofing and garage doors

It is not an HVAC quirk. The same publisher's roofing benchmark puts branded at $44 against $124 non-branded, and states plainly that branded roofing search is 65% cheaper than non-branded. Their garage door benchmark shows $66 branded against $173 non-branded, a 62% gap. Three trades, three samples, one shape.

There is also a useful like-for-like comparison buried in that roofing report, and it is like-for-like because it comes from one platform, one window and one definition of a lead: non-branded cost per lead of $124 for roofing, $128 for electrical, $149 for HVAC and $183 for plumbing.

Now compare that roofing figure of $124 to the $228.15 in the table at the top of this page. Same trade, wildly different number. Different samples, different windows, different definitions, and one is a median while the other is spend-weighted across branded and non-branded separately. Two credible sources disagreeing by that much is not a scandal. It is the ordinary state of published benchmarks, and it is the strongest argument on this page against treating any single figure as your number.

Branded leads are cheap for a reason, and you cannot buy more of them

Two consequences, both practical.

First, a branded lead comes from somebody who already knew your company name before they searched. That awareness got paid for somewhere else, in trucks and yard signs and ten years of doing good work, and a share of those people would have found you without the ad. Including branded in a headline cost per lead flatters the figure in direct proportion to how well known the business already is. Which means it flatters most where the advertising deserves the least credit.

Second, branded is capacity-limited. In the roofing dataset it was 9% of total spend, and it is capped by the number of people who know your name. You cannot scale a great branded cost per lead by spending more, because the demand is not there to buy. Every additional dollar of growth has to come out of the expensive half.

So if a report lands on your desk showing one cost per lead and no split, that report cannot be answered. Not by you, and not by the person who sent it. The first question in working out whether your agency is doing a good job is what share of those leads came from people typing your business name.

Exclusive or shared: three different things called a lead

A shared lead is a right to compete, not an enquiry

There are at least three assets sold under the word lead, and they are not comparable on unit price.

An exclusive search lead is the enquiry itself. One person, contacting one business, because they clicked one ad. A shared marketplace lead is the right to compete for a job alongside several other contractors who bought the same contact details. A Local Services Ads lead is a third thing again, covered in the next section.

Comparing a shared lead price to an exclusive lead price is comparing a bid packet to a signed contract.

The only comparison that survives is cost per booked job

Close rates on shared leads are structurally lower. You are one of several callers, often not the first, and the homeowner is in a different frame of mind when the third contractor rings than when the first one did. That means a shared lead at a fraction of the price of an exclusive one can still cost more per booked job, and the arithmetic is not hard once you have both close rates.

The trouble is that most contractors track spend by source and close rate nowhere, which is why this argument is nearly always settled on sticker price. If you do one piece of measurement this year, make it close rate by lead source. It is the number that decides which channel is actually cheap.

One documented note on marketplace lead quality, and only this one: in 2023 the FTC ordered HomeAdvisor to pay up to $7.2 million over deceptive claims about the quality and source of the leads it sold to service providers. The FTC said some leads did not match the type of work or geographic area providers had signed up for. That is the full FTC release if you want the detail.

Even exclusive is not always exclusive

This is the part worth reading twice, because it is Google describing its own product in two places that sit oddly together.

On Google's getting started page for Local Services Ads in Canada, the pitch is that customers choose you, and that you only hear from customers who have specifically selected your profile out of all the rest. On Google's page explaining how leads work, the price of a message lead is set partly on whether the customer has contacted other Local Services advertisers.

Both are true at once, and together they say something useful. The lead is self-selected, and the platform's own pricing model assumes the same homeowner may be talking to your competitors right now.

Generalise it and you get the honest version: exclusivity describes who the lead was sold to. It never describes who the homeowner phoned. A person can search once and contact four businesses in eight minutes, and no advertising product prevents that. If you are weighing the two channels against each other, Local Services Ads versus Google Ads is the comparison to run.

Local Services Ads: a different unit, and an invoice that is not the cost

You pay per lead, and the platform decides what a lead is worth

Google's own framing is that with Local Services Ads you pay for valid leads, and that lead prices vary by location, job type, lead type and bidding mode. Published LSA benchmarks covering 888 home services contractors and $6.72M in spend put the blended cost per lead at $53, with electrical at $39 and plumbing at $57. Book rate in that dataset is 43.9%.

Set $53 against the $149 non-branded search figure for the same trades and it looks like a decision that makes itself. It is not, because those are two different products, bought on two different units, at two different points in the search. Our guide to Local Services Ads for contractors covers how the product actually works before you compare anything.

Credits mean the invoice number and the real cost differ

Google assesses leads when the customer first makes contact and does not charge for ones judged invalid or low quality. Charged leads then get reassessed over time, and credits are issued automatically where the model decides one is warranted.

Here is the operational detail that catches people out, straight from Google's page on lead credits: credits are usually applied to your account balance within 30 days, and the original lead charge still appears on your invoice. So the number on the bill and your true cost per lead are two different numbers, by Google's own description, and a month read straight off an invoice overstates what the leads cost.

Two more current limits worth knowing. Automatic lead credits are available in the United States and Canada only. And Google has stopped supporting credits for leads flagged as job type not serviced or geo not serviced, which used to be two of the most commonly disputed categories. If you are still working through the badge itself, that is covered in how to get Google Guaranteed.

Why an LSA cost per lead cannot be set beside a search cost per lead

Three reasons, one line each. Different unit of purchase, because one is billed per lead and the other per click. Different definition of what gets counted, because one is Google's judgment of validity and the other is whatever your account was configured to count. Different position on the page, reaching a different moment in the same search.

The comparison that does hold up is the same one as before: cost per booked job, with the close rate on each source measured separately. Canadian readers should also read the Canada version of Google's getting started article, linked above, since several product details differ here from the American version.

What counts as a lead is a setting, not a fact

One conversion per click, or every conversion

This is the quietest reason two cost per lead figures are not comparable, and almost nobody checks it.

Google Ads lets an advertiser choose, for each conversion action, whether to count every conversion that happens after an interaction or only one. That is a setting somebody picked, not a fact about the business. Two identical companies with identical results can publish cost per lead figures that differ substantially on this choice alone.

One thing to be precise about, because it is widely misunderstood: the setting de-duplicates per ad interaction, not per person. The same homeowner arriving through two separate ad interactions can count twice either way, and the counting option is not person-level de-duplication.

Call length, taps, and the same person twice

The other configuration choices that move the number without anything real changing. Whether a phone call has to last past a minimum duration before it counts. Whether a tap on a phone number counts before anybody speaks. Whether a chat, a brochure download or a booking request counts as a lead. Whether a repeat call from an existing customer gets counted again.

Every one of those is a defensible choice, and different accounts make different ones honestly. None of them are standard. If your call volume looks wrong in either direction, why your Google Ads are not getting calls works through the mechanics.

What to ask before you compare your number to anybody else's

The generalisation, said plainly: nearly every published cost per lead is really cost per conversion, as each advertiser happened to define it. That includes every figure quoted on this page. It is still useful. It is just not a standard unit of measure, and treating it as one is how contractors end up feeling bad about a number that was never comparable.

How to work out your own cost per lead

The calculation, and what belongs in the cost

Total cost divided by leads. The arithmetic is not the hard part. The argument is over what goes in the numerator.

Ad spend alone gives you a number comparable to every benchmark on this page, because that is what all of them measured. Ad spend plus management fee plus landing page and call tracking costs gives you the number that actually matters to your business, because that is what leaves the bank account. Both are legitimate. They are not the same number, and quoting one while thinking about the other is how contractors end up arguing past each other. Pick one, and say which one you picked. If you are weighing the fee side, what Google Ads agencies charge covers the models and what each one includes.

The result is only comparable when the scope is stable: the same cost definition, lead definition, campaign set, brand treatment, and a window with enough outcomes to interpret. Changing any of those can move the reported figure even when account performance did not change.

Split it before you read it

This is where the branded point stops being theory and becomes a job on your desk. If branded and non-branded searches sit inside one campaign, your blended figure is uninterpretable and the fix is structural rather than analytical. You cannot compute your way out of it. Somebody has to separate the campaigns.

There is one thing you can check yourself in about ten minutes. Look at whether searches for your business name are being paid for at all, and what share of your leads came from them. If a meaningful slice of your leads are people who typed your company name into Google, your headline cost per lead is describing your reputation, not your advertising.

Choose a window long enough to mean something

A short window with only a few leads produces a cost per lead that can swing on one event. Comparing two such windows can produce a story about randomness that sounds exactly like a story about performance.

Composition is the other trap. Month over month movement in cost per lead is frequently service mix shifting, or branded share shifting, or a seasonal service waking up, with nothing about the account having changed at all.

What a good cost per lead looks like for your business

Work backwards from job value and close rate

There is no benchmark that answers this, and any page handing you one is guessing. The derivation is short.

Take your average job value for the work these leads produce. Multiply it by your close rate on that specific lead source. That gives you revenue per lead. Whatever share of that revenue you are willing to spend to win the work is your ceiling on cost per lead. Above the ceiling you are buying work at a loss. Below it you should probably be buying more.

Run it in both directions, because the second half is the part nobody says out loud. A $200 lead is cheap against a large job at a good close rate. A $40 lead is ruinous against a small ticket at a poor one. Sticker shock is not analysis.

Working out how much budget that implies is a related but separate question, covered in what a contractor should spend on Google Ads.

Cost per lead versus cost per acquisition

A lead is somebody who raised a hand. An acquisition is somebody who paid you. Between them sits your close rate, which lives in your business and not in the ad account.

The branded and non-branded table further up this page is the cleanest proof of why the distinction matters. In that dataset the cost per lead gap between the two campaign types is 4.4 times, and the cost per customer gap is 7.7 times. Same accounts, same month, and the two metrics disagree about how big the difference is by nearly a factor of two.

Cost per lead is a diagnostic. Cost per booked job is the bottom line. If you only ever look at one, look at the second, and if you are still deciding whether the channel earns its place at all, are Google Ads worth it for contractors works through that question properly.

A benchmark is a sanity check, not a target

Which closes the loop on the table at the top of this page. That table washes out geography, seasonality, service mix inside a trade, account age and budget scale. Within one trade, the difference between January and June, or between a dense metro and a rural service area, routinely moves cost per lead more than the entire gap between two trades in the table.

Hitting the benchmark proves nothing. Missing it badly is a reason to open the account and look. That is the whole of what a benchmark can do for you.

If your cost per lead is too high, fix it in this order

Decompose before you touch anything

The first move is not a change. It is a division.

Pull the click price and the conversion rate separately, for non-branded only, and see which one is out of line with your trade. If the click price is normal and the conversion rate is weak, the first place to look may be the page, the phone, the form, or the intent of the traffic rather than the bid itself.

Diagnosis starts by separating branded from non-branded demand and reading click price beside conversion rate. A high cost per lead can come from expensive auctions, weak conversion, service mix, or the lead definition itself. Those causes call for different decisions, so changing bids from the blended headline number is premature.

The conversion-rate side

What moves it, roughly in order of size.

Whether the click lands on a page about the specific service somebody searched for, or on a homepage that makes them go looking. Whether the phone is answered, and how fast. How much the form asks for before it lets somebody send it. How the page behaves on a phone, which is where most of this traffic is.

The response speed point deserves its own sentence, because it involves no advertising at all. The same enquiry is worth materially more to the business that calls back first than to the one that calls back tomorrow, and every contractor knows this from their own life as a customer. Nothing about the ad changed. The money changed. Several of the other items on that list show up in the common Google Ads mistakes contractors make.

The click-price side, and its limits

What you can genuinely move: which searches you are paying for, what is excluded from targeting, match types, bid strategy, and whether a chunk of your budget is going to searches from people who were never going to hire anybody. That last one is usually the largest single win available on the click side, and it is unglamorous work.

What you cannot move: the fact that every other advertiser in your city helps set the auction price, and that a trade with high job values will always have expensive clicks because the job is worth paying for. Anyone promising to cut your click price in half is describing a different account.

A Canadian contractor has no published benchmark

Every figure on this page is American and in US dollars

Worth being blunt about, because most pages covering this quietly present American figures as universal.

The sixteen-trade table is 3,211 US campaigns. The cross-industry report is 13,474 US campaigns. The roofing and garage door datasets say United States explicitly, and two of the other datasets do not state a country at all. No comparably sourced Canadian per-trade table exists in public circulation.

Canadian figures do circulate on agency blogs. They arrive without a sample size, without a period and without a method, which is why none of them appear on this page. A number with no methodology behind it is not a benchmark. It is a guess with a decimal point.

Read the table for shape, not for level

What may transfer: the broad relationship between a considered purchase and conversion rate, the branded and non-branded gap, and the structural questions on this page. Those are useful hypotheses to test in a Canadian account, not Canadian benchmark levels.

What does not transfer: the levels. Currency is the obvious one. Competitor density and market size are the bigger ones. A Toronto auction has a different set of bidders with different budgets than a Dallas auction does, and a trade that is crowded across the GTA can be nearly uncontested ninety minutes north of it.

What to do instead

The honest answer is that a Canadian contractor's benchmark is his own account, measured over a long enough window, split branded from non-branded, with one written-down definition of a lead.

That is more useful than a table anyway. A trade median tells you where thousands of other businesses landed. Your own account last year tells you whether the thing you changed worked.

Why nobody can hand you your number

The variables that move it most are all account level

Which services inside your trade get the budget. Whether each of those services has a page that matches the search. What is excluded from your targeting. What counts as a lead. What happens in the first hour after one lands. How much of your spend is branded.

A trade average cannot know any of those things. That is not a flaw in the table. That is what an average is.

The number moves when nothing about the ads changed

The season turns. The service mix shifts. A competitor enters the auction, or leaves it. The office is short-staffed for a fortnight and calls go to voicemail. A page loads slowly after a plugin update.

Every one of those moves your cost per lead without a single setting being touched, which is why the number needs watching continuously rather than reading quarterly. By the time a quarterly report shows the movement, the quarter is spent. How long it takes to get a readable signal in the first place is covered in how long Google Ads take to work.

What ongoing management is actually for

The initial build is only one part of the job. Ongoing management is the judgment work of noticing when click price, conversion rate, lead quality, or the definition of a lead has changed, then finding the cause before a long reporting period hides it.

Some contractors do that themselves and do it well. What matters is that somebody owns the recurring review and can connect platform numbers to booked work. If that person is not going to be you, it needs to be somebody. That is the work I do.

What to do with this

Three things decide any cost per lead figure: the trade, whether branded searches are inside it, and what somebody counted as a lead. Get those three straight and the number becomes usable. Leave any one of them vague and it stays a number that cannot be argued with or acted on.

For a benchmark to be useful, the report beside it should define what counted as a lead, separate branded from non-branded demand, and connect paid leads to booked jobs. Without those boundaries, a precise cost per lead can still support the wrong decision.

If you want somebody responsible for that interpretation and for the account changes behind it, that is the job I do.