The June LSA Benchmark Just Dropped: $62.56 Per Lead, $302.33 To Buy A Customer, 7.78x ROAS. Google Just Rewrote The Policy Book On July 6. HVAC Median CPL $59, Electrical $49.26, Plumbing $71. Competitor-Brand LSAs Are Out. Responsiveness Is Now A Ranking Signal. Rebenchmark Against Your Trade, Not The Average.

Two things hit your Google Ads account this week that most trade owners are going to miss. First, the biggest LSA benchmark report of the year dropped July 2 with numbers that redraw what “a good cost per lead” means. Second, Google’s new LSA policies went live July 6. If you run a plumbing, HVAC, roofing, or electrical shop and you have not read either one, you are about to burn budget you did not need to burn.

Here is what you actually need to know.

The benchmark: $62.56 CPL, $302.33 to buy a customer

Jon Torrey’s Data-Driven Trades LSA benchmark for June 2026 covers 945 home services businesses and $13.1 million in LSA spend tracked through SearchLight. It is the largest public LSA data set in the trades, and this month he changed the format. Instead of averages, he now publishes ranges.

The headline numbers:

• $62.56 cost per unique lead across the sample — effectively flat vs May’s $63.29, up from $55.08 in March.
• $302.33 cost per paying customer, down from $315.52 in May.
• Average spend per business jumped 43% month-over-month to $13,844 — peak-season scaling — and Google’s pricing did not punish it. That is the story.
• Median business spent $6,070. Middle 80% of the sample spent between $700 and $33,000. If your budget is not $13,800, you are not behind.

Torrey’s line: “A $75 CPL isn’t broken. It’s inside the normal range.” The middle 50% of businesses paid between $46 and $76 per unique lead. The middle 80% paid $33 to $88. If you have been beating yourself up over $70 leads, stop.

The trade breakdown that changes your plan

This is where it gets useful. Torrey’s June numbers by trade:

• HVAC: CPL range $36-$85, median $59. Cost per paying customer $129-$716, median $281. Highest average ticket in the sample at $2,857.65. Median closed ROAS 8.1x.
• Plumbing: CPL range $43-$95, median $71. Cost per paying customer $170-$591, median $319. Highest share of unique leads that convert to paying customers. Median closed ROAS 6.2x.
• Electrical: CPL average $49.26 — the least expensive trade in the sample. Sample grew from 21 accounts in May to 93 in June, and book rate lifted from 34% to 39.6%. Median closed ROAS 6.1x. Torrey calls it “the most compelling ticket-to-CAC ratio if volume can be achieved.”
• Roofing: $90.42 CPL from 20 accounts — directional, small sample.

The point Torrey is hammering: “These are fundamentally different unit economics by trade.” Benchmark yourself against your trade, not the $62.56 headline number. Otherwise you will misread your own P&L.

The policy update that went live July 6

Same week, Google shipped a full LSA policy refresh. The policies are now called “Local Services Ads requirements.” Google says it is mostly wording and readability, but two things matter for revenue:

1. LSAs cannot target competitors’ branded searches. Google’s clarified language: if you want to appear on a competitor brand query, use Search or Performance Max, not LSA. If your agency has been running LSA visibility off competitor names, it stops now.

2. Requirements align to the Google Verified badge system. In October 2025 Google retired the Google Guaranteed, Google Screened, and License Verified badges and replaced them with a single blue “Google Verified” checkmark. The money-back guarantee is gone. Location3 and Coalmarch both confirmed it. The July 6 requirements refresh brings the policy language in line with what the badge actually promises today: identity and licensure verification, not consumer reimbursement.

If you sold the guarantee to homeowners as a trust signal, your sales script is out of date.

The responsiveness signal Google is quietly punishing

Buried inside the algorithm updates over the last six months: real-time responsiveness scores now materially affect LSA visibility. Miss the call or answer slow, and Google cycles your profile out of top placements. That change dovetails perfectly with the AI-receptionist stack we covered yesterday — xAI Grok Voice Agent Builder at $0.05/min, Google’s own AI-powered calling in AI Pro. Google is turning “does this shop pick up” into a ranking factor.

Match that to Torrey’s data. Book rate in the middle 80% of businesses runs 24% to 55%. The bottom of that range is voicemail rot. The top is a shop with an AI agent, a live dispatcher, or both, and a call-flow that answers in under 30 seconds.

What every trade owner should do this week

1. Rebenchmark against your trade range, not the average. If you run HVAC and your CPL is $75, you are at the 75th percentile — not broken. If you run electrical and your CPL is $75, you are getting fleeced. Pull the last 90 days of LSA data and score yourself against the ranges above.

2. Kill any LSA campaign targeting competitor brand names. Effective July 6. Move that spend to Performance Max or standard Search with dedicated competitor-brand negatives on your own campaigns.

3. Fix your response speed before you fix your bids. Google is grading you on how fast you answer. Put a human on the first 15 seconds and an AI receptionist on seconds 15-30. Then check LSA impressions and cost per paying customer against Torrey’s $302.33 baseline in 30 days.

4. Fill out the Google survey after every lead. Location3 and industry practitioners confirm: unfilled surveys mean lower-quality leads and higher CPLs. It is a five-second click that Google uses to price your placement.

The bottom line

The LSA market did not get harder in June. It got clearer. $62.56 blended CPL, $302.33 to buy a paying customer, and 7.78x aggregate closed ROAS across nearly $13M in real spend. The trades with the best unit economics right now are electrical and HVAC. The shops falling behind are the ones that missed the July 6 policy shift and the responsiveness signal underneath it. Fix the phone. Fix the bids. Fix the benchmark you compare yourself to. Then run the numbers again in 30 days.