Lonta Research · Evidence review
Home services marketing benchmarks can describe a click, an answered call, a homeowner's preference or a paid job. Those are different outcomes. Compare the original definitions before you compare your shop.
Search advertising, phone handling and customer research. Each number keeps its sample and its limits.
All Lonta research and tools · For local weather patterns, explore the HVAC seasonality calendar.
What the review establishes
Search advertising
LocaliQ's air conditioning figure is $127.74 per ad conversion. The report summarizes campaign medians across two search platforms. It does not measure collected jobs or your dispatch costs.
Inspect the benchmark →Phone handling
Invoca reports 52% for all calls and 65% for calls longer than 15 seconds. That 13-point gap is between filtered populations in one report, not evidence of an operational improvement.
See the denominator change →AI discovery
BrightLocal measures reported use of AI for local recommendations. Invoca measures the lead share of answered ChatGPT-referred calls. Neither establishes that AI supplies 45% of contractor jobs.
Check the survey scope →The evidence library
21 of 21 published observations. Several rows can come from one study.
Search advertising
Air conditioning installation & repair
Not cost per booked or collected job.
3,211 campaigns overall; at least 103 per category. Exact category counts are not published.
April 1, 2024–March 21, 2025
Google and Microsoft search ads combined; campaign medians, not national market averages or sold-job costs.
Read the original · LocaliQLocate: Metric tables and About the data.
Search advertising
Heating & furnaces
Keep this category separate from air conditioning.
3,211 campaigns overall; at least 103 per category. Exact category counts are not published.
April 1, 2024–March 21, 2025
Google and Microsoft search ads combined; campaign medians, not national market averages or sold-job costs.
Read the original · LocaliQLocate: Metric tables and About the data.
Search advertising
Plumbing
The matching heating value does not establish matching job economics.
3,211 campaigns overall; at least 103 per category. Exact category counts are not published.
April 1, 2024–March 21, 2025
Google and Microsoft search ads combined; campaign medians, not national market averages or sold-job costs.
Read the original · LocaliQLocate: Metric tables and About the data.
Search advertising
Electricians & electrical contractors
Mixed repair and project demand is not separated here.
3,211 campaigns overall; at least 103 per category. Exact category counts are not published.
April 1, 2024–March 21, 2025
Google and Microsoft search ads combined; campaign medians, not national market averages or sold-job costs.
Read the original · LocaliQLocate: Metric tables and About the data.
Search advertising
Air conditioning installation & repair
Do not divide independent medians to reconstruct a pooled CPL.
3,211 campaigns overall; at least 103 per category. Exact category counts are not published.
April 1, 2024–March 21, 2025
Google and Microsoft search ads combined; campaign medians, not national market averages or sold-job costs.
Read the original · LocaliQLocate: Metric tables and About the data.
Search advertising
Air conditioning installation & repair
Tracked conversion, not a paid invoice.
3,211 campaigns overall; at least 103 per category. Exact category counts are not published.
April 1, 2024–March 21, 2025
Google and Microsoft search ads combined; campaign medians, not national market averages or sold-job costs.
Read the original · LocaliQLocate: Metric tables and About the data.
Calls and response
Home services
A person answered; automated engagement is not the same outcome.
Home services call count not stated. The broader, ten-industry dataset exceeds 70 million calls; that is not this segment's sample.
Measurement dates not stated; report published July 2026
Customer-base averages. Segment counts, dispersion and averaging weights are not disclosed. AI referrals have very low reported volume.
Read the original · InvocaLocate: Call Answer Rates; Phone Lead and Conversion Rates; Methodology.
Calls and response
Home services
Short calls are excluded; this is not an improvement over the all-call rate.
Home services call count not stated. The broader, ten-industry dataset exceeds 70 million calls; that is not this segment's sample.
Measurement dates not stated; report published July 2026
Customer-base averages. Segment counts, dispersion and averaging weights are not disclosed. AI referrals have very low reported volume.
Read the original · InvocaLocate: Call Answer Rates; Phone Lead and Conversion Rates; Methodology.
Calls and response
Home services
Another duration filter on the same source dataset, not an independent study.
Home services call count not stated. The broader, ten-industry dataset exceeds 70 million calls; that is not this segment's sample.
Measurement dates not stated; report published July 2026
Customer-base averages. Segment counts, dispersion and averaging weights are not disclosed. AI referrals have very low reported volume.
Read the original · InvocaLocate: Call Answer Rates; Phone Lead and Conversion Rates; Methodology.
Calls and response
Home services
Use the detailed section's denominator, not the shortened research-highlight wording.
Home services call count not stated. The broader, ten-industry dataset exceeds 70 million calls; that is not this segment's sample.
Measurement dates not stated; report published July 2026
Customer-base averages. Segment counts, dispersion and averaging weights are not disclosed. AI referrals have very low reported volume.
Read the original · InvocaLocate: Call Answer Rates; Phone Lead and Conversion Rates; Methodology.
Calls and response
Home services
Conversion on a call is not collected job revenue.
Home services call count not stated. The broader, ten-industry dataset exceeds 70 million calls; that is not this segment's sample.
Measurement dates not stated; report published July 2026
Customer-base averages. Segment counts, dispersion and averaging weights are not disclosed. AI referrals have very low reported volume.
Read the original · InvocaLocate: Call Answer Rates; Phone Lead and Conversion Rates; Methodology.
Calls and response
ChatGPT-referred home services calls
Referral volume is described as very low; exact channel count is not published.
Home services call count not stated. The broader, ten-industry dataset exceeds 70 million calls; that is not this segment's sample.
Measurement dates not stated; report published July 2026
Customer-base averages. Segment counts, dispersion and averaging weights are not disclosed. AI referrals have very low reported volume.
Read the original · InvocaLocate: Call Answer Rates; Phone Lead and Conversion Rates; Methodology.
Calls and response
ChatGPT-referred home services calls
A small channel cohort; no evidence of its share of all contractor bookings.
Home services call count not stated. The broader, ten-industry dataset exceeds 70 million calls; that is not this segment's sample.
Measurement dates not stated; report published July 2026
Customer-base averages. Segment counts, dispersion and averaging weights are not disclosed. AI referrals have very low reported volume.
Read the original · InvocaLocate: Call Answer Rates; Phone Lead and Conversion Rates; Methodology.
Calls and response
Small businesses across 58 industries
Headline rounding; includes voicemail and no response. Not a current HVAC-only rate.
85 businesses, 58 industries. Number of calls not stated.
30 days; calendar dates not stated; published January 18, 2016
Historical mixed-industry sample; voicemail is included in the unattended-call headline.
Read the original · 411 LocalsLocate: Paragraphs beginning We monitored and While 37.8%.
Calls and response
Web-form leads at six companies
An odds comparison, not 21 times the sales, bookings or revenue.
More than 15,000 leads and 100,000 call attempts at six companies.
Three years; calendar bounds not stated; presented October 16, 2007
Qualification definitions varied by company. Close ratios were not studied; this is not a contractor sales-lift experiment.
Read the original · InsideSales.com / James OldroydLocate: PDF pages 1, 2 and 5; retrieved copy carries a July 2008 print date.
Customer research
US adults
Across local businesses; not a contractor purchase rate.
1,002 US adults via SurveyMonkey; question-level counts not stated.
Fieldwork dates not stated; 2026 report edition
Self-reported behavior across local businesses; not an HVAC customer cohort, traffic share or measured booking rate.
Read the original · BrightLocalLocate: Review-reading and recommendation sections; Methodology.
Customer research
US adults
Self-reported platform use, not Google's share of traffic.
1,002 US adults via SurveyMonkey; question-level counts not stated.
Fieldwork dates not stated; 2026 report edition
Self-reported behavior across local businesses; not an HVAC customer cohort, traffic share or measured booking rate.
Read the original · BrightLocalLocate: Review-reading and recommendation sections; Methodology.
Customer research
US adults
Any reported use is not 45% of searches, leads or purchases.
1,002 US adults via SurveyMonkey; question-level counts not stated.
Fieldwork dates not stated; 2026 report edition
Self-reported behavior across local businesses; not an HVAC customer cohort, traffic share or measured booking rate.
Read the original · BrightLocalLocate: Review-reading and recommendation sections; Methodology.
Customer research
US homeowner decision-makers
A preference, not an 80% increase in conversions.
1,040 homeowners via SurveyMonkey Audience; demographically balanced. Question-level counts not stated.
October 23, 2025
Stated expectations and preferences, not observed conversion lift. Extracted from the publisher's public summary, not the full questionnaire.
Read the original · Housecall ProLocate: Digital-first service; AI inquiries; Methodology.
Customer research
US homeowner decision-makers
Reported expectation, not the effect of redesigning a site.
1,040 homeowners via SurveyMonkey Audience; demographically balanced. Question-level counts not stated.
October 23, 2025
Stated expectations and preferences, not observed conversion lift. Extracted from the publisher's public summary, not the full questionnaire.
Read the original · Housecall ProLocate: Digital-first service; AI inquiries; Methodology.
Customer research
US homeowner decision-makers
Comfort with a service interaction differs from AI search usage.
1,040 homeowners via SurveyMonkey Audience; demographically balanced. Question-level counts not stated.
October 23, 2025
Stated expectations and preferences, not observed conversion lift. Extracted from the publisher's public summary, not the full questionnaire.
Read the original · Housecall ProLocate: Digital-first service; AI inquiries; Methodology.
Before you compare
Choose two observations to inspect their definitions. This check does not calculate a pooled average or decide statistical compatibility.
First observation
52%
Second observation
65%
One report, three denominators
Person answer rates in Invoca's 2026 home services report. These are conditional views of the same source dataset, not three independent studies or a time trend.
Source: Invoca, Call Answer Rates. Home services call count and fieldwork dates are not stated. The 13-percentage-point gap is 65 − 52.
To compare your office, first match the call-length filter and what counts as an answer. A live person, voicemail and an automated agent are different outcomes. A callback later in the day is also different from answering the initial call.
The older 62% unattended-call headline from 411 Locals includes voicemail and comes from a mixed-industry sample published in 2016. Its presence here explains a commonly repeated number; it is not a current HVAC target.
LocaliQ's air conditioning tables report a median click cost of $9.68, a median conversion rate of 6.56%, and a median lead cost of $127.74. Dividing $9.68 by 0.0656 gives $147.56, not the published lead figure.
That arithmetic does not demonstrate an error in the report. The median of campaign-level ratios is not generally the ratio of two independently reported medians. These summaries do not reconstruct a single representative campaign.
Compare your own cost per conversion using your actual spend and conversion count, after matching channel, trade, tracking rules and time window. For the amount your work can afford, use the calculator that includes dispatch and unsuccessful visits. A competitor benchmark and your advertising ceiling answer different questions.
Source: LocaliQ's 2025 tables and methodology. The page was updated in 2026; the disclosed campaign window remains April 2024–March 2025.
We started with the cost, call-response and customer-research claims in our earlier statistics source audit. On October 9, 2026, we followed those topics to accessible documents from their original publishers and checked the values against the detailed metric and methodology sections.
This is a targeted review of six purposively selected studies, not an exhaustive search of every home services report. The older call and web-lead studies are included because their figures remain widely cited, not because they describe today's contractor market. The library contains selected observations, not every statistic in each report.
Statistical pooling needs a justified common quantity and appropriate information about precision or counts. The Cochrane Handbook's discussion of meta-analysis explains why study differences and uncertainty matter. We use that general statistical principle; this marketing review is not a Cochrane review.
Here, a call-share average, a campaign median, a reported odds comparison and a survey preference are not interchangeable. Even two percentages can use different events and denominators. Adding campaigns, calls and people would give a large number with no valid analytical meaning.
The original datasets and full questionnaires were not independently audited. All six studies were published or co-produced by commercial vendors; their samples do not establish a census of US contractors. Some reports omit dates, subgroup counts, uncertainty or averaging weights. We retain those gaps in the downloads and avoid confidence intervals, causal sales-lift claims and claims of national representativeness.
No client records or synthetic observations are included. Lonta created the extraction, definition checks and analysis; the numerical findings belong to the cited publishers. This review has no paid placements or affiliate links.
Every source was opened and checked on October 9, 2026. The measurement period stays separate from that check date.
LocaliQ · 2025
Design: Observed advertising campaigns.
Population: US home services search campaigns; LocaliQ customers.
Sample: 3,211 campaigns overall; at least 103 per category. Exact category counts are not published.
Window: April 1, 2024–March 21, 2025.
Commercial interest: Sells advertising services.
Limit: Google and Microsoft search ads combined; campaign medians, not national market averages or sold-job costs.
Find the extraction in: Metric tables and About the data.
Invoca · 2026
Design: Observed platform calls.
Population: Invoca home services customers across nine sub-industries.
Sample: Home services call count not stated. The broader, ten-industry dataset exceeds 70 million calls; that is not this segment's sample.
Window: Measurement dates not stated; report published July 2026.
Commercial interest: Sells call analytics and AI agents.
Limit: Customer-base averages. Segment counts, dispersion and averaging weights are not disclosed. AI referrals have very low reported volume.
Find the extraction in: Call Answer Rates; Phone Lead and Conversion Rates; Methodology.
411 Locals · 2016
Design: Observed business calls.
Population: 85 small businesses across 58 industries; geography not specified.
Sample: 85 businesses, 58 industries. Number of calls not stated.
Window: 30 days; calendar dates not stated; published January 18, 2016.
Commercial interest: Sells local SEO services.
Limit: Historical mixed-industry sample; voicemail is included in the unattended-call headline.
Find the extraction in: Paragraphs beginning We monitored and While 37.8%.
InsideSales.com / James Oldroyd · 2007
Design: Observational web-lead analysis.
Population: Web-form leads at six companies using InsideSales.com.
Sample: More than 15,000 leads and 100,000 call attempts at six companies.
Window: Three years; calendar bounds not stated; presented October 16, 2007.
Commercial interest: Co-produced by a lead-response software vendor.
Limit: Qualification definitions varied by company. Close ratios were not studied; this is not a contractor sales-lift experiment.
Find the extraction in: PDF pages 1, 2 and 5; retrieved copy carries a July 2008 print date.
BrightLocal · 2026
Design: Consumer survey.
Population: US adults considering local businesses across industries.
Sample: 1,002 US adults via SurveyMonkey; question-level counts not stated.
Window: Fieldwork dates not stated; 2026 report edition.
Commercial interest: Sells local SEO and review software.
Limit: Self-reported behavior across local businesses; not an HVAC customer cohort, traffic share or measured booking rate.
Find the extraction in: Review-reading and recommendation sections; Methodology.
Housecall Pro · 2025
Design: Homeowner survey.
Population: US homeowners who make household maintenance decisions.
Sample: 1,040 homeowners via SurveyMonkey Audience; demographically balanced. Question-level counts not stated.
Window: October 23, 2025.
Commercial interest: Sells home services operating software.
Limit: Stated expectations and preferences, not observed conversion lift. Extracted from the publisher's public summary, not the full questionnaire.
Find the extraction in: Digital-first service; AI inquiries; Methodology.
The complete extraction is free and requires no account. CSV and JSON contain the same 21 observations, original source links, sampling information and limitations.
Suggested citation
Lonta AI. Home Services Marketing Benchmarks: Evidence Review. Version 1.0. Sources checked October 9, 2026. lonta.io/research/home-services-marketing-benchmarks/
Cite Lonta for the comparison and method, and the original publisher for a numeric finding. Preserve its period, denominator and limit when quoting a value. Original publishers retain rights to their reports and graphics.
This is a targeted evidence review. A meta-analysis statistically combines compatible results from separate studies. The definitions, populations and uncertainty available here do not support a pooled estimate, so the observations remain separate.
LocaliQ's 2025 edition reports a $127.74 campaign median for air conditioning installation and repair. Its campaigns ran in 2024–2025 and include Google and Microsoft search ads. That gives context for an ad conversion, not an affordable price for your own sold work.
Campaigns, businesses, calls, web leads and survey respondents are different units. Reports also contain several observations from one dataset. Adding their headline sample sizes would not create a valid sample for a pooled marketing benchmark.
Download the extraction and cite this Lonta review for its comparison and method. Cite the original publisher for each numeric finding. The files keep source links, dates, definitions and limitations beside the values; original publishers retain rights to their reports and graphics.
For practical follow-through, read how to separate branded searches from new-customer jobs and how to measure the cost of missed calls.
Lonta runs marketing for HVAC, plumbing and electrical companies. The free written strategy starts with your shop's service area and measurement gaps.
