3.68x ROAS on Unbranded HVAC Google Ads

June and July 2026 Google Ads Benchmarks

👋 Hey, Jon here!

There’s been a lot of talk lately about how to properly interpret Google Ads performance (and for good reason!). I’ve normally broken these articles down by comparing Unbranded and Branded campaigns (still true for this issue), but I’m reframing this newsletter's approach to focus on you, the operator.

When you review your Google Ads performance, personally or through a vendor partner, it’s critical you ask for the performance to be split between Branded and Unbranded campaigns (sample questions are provided at the bottom of this article) 👇

Closed ROAS for Google Ads came in around 7x across 606 HVAC businesses on our platform this quarter (May - July).

Impressive.

But strip out the brand-name searches, and the ROAS comes in at 3.68x (July, unbranded search).

Same accounts, very different story.

“Blended” combines two very different jobs your ad dollars do, and once you can tell them apart, you’ll read your own report like a pro.

The two jobs: capturing demand you already have (someone searching your company name) and generating demand you don’t (someone searching “AC repair near me” who’s never heard of you).

Both show up as “Google Ads” but are served in meaningfully different ways.

Across May, June, and July ($28.4M in HVAC Google Ads spend across ~600 businesses), branded search accounted for 10% of spend but 41% of revenue.

That’s not a knock on anyone running these accounts; branded search should look amazing.

It just means the blended number is flattering, and it’s fair to ask for both branded and unbranded Google Ads performance reports for your business.

This edition shows you how to separate the two, using four campaign types: branded search, unbranded search, Performance Max, and awareness (YouTube/Reach/Remarketing) — across May, June, and July-to-date.

---

June, Broken Apart (HVAC · 606 businesses)

Source: SearchLight

Branded Search

Branded returned 27x and books at 66% at $36 a lead.

This is expected behavior: someone typing “[Your Company] AC repair” is likely high intent, and so long as lead handling and booking are dialed in on your end, you're bound to see very strong performance.

The conversation about how much to invest in Branded spend on Google Ads is unique to your business - you may feel the need to spend more to protect your backyard.

I’m not sharing this data as a verdict on ‘spend on branded’ or ‘don’t spend on branded’; just make sure you know how much is allocated to branded and have a strategic conversation about it.

When Google Ads performance is included in the blended number, it pulls the average CPL down to a comfortable $107 and lifts total ROAS to 7x.

Nothing wrong with that as long as you’re able to break the performance down further and aren’t unprofitable (within your tolerance) on unbranded campaigns.

---

Unbranded ROAS Performance

Here’s June with branded and awareness set aside — just the unbranded search that actually finds new customers:

• 🚨 Cost per lead: $184.56 — not $107

• ➡️ ROAS: 4.06x — not 7.03x

• ⬇️ Half of the businesses in this sample ran below the average: the median business sees $174 CPL and 2.58x ROAS on unbranded

When you’re deciding whether to add budget to Google Ads, the question isn’t “what’s my blended ROAS” — it’s “what does the next new customer cost, and what do they return?”

For HVAC in June that’s a ~$185 lead at ~4x.

Still profitable at HVAC tickets, but a very different math than 7x implies.

Budget against the blend and you’ll over-assume; budget against unbranded and you’ll size spend to reality.

Note that unbranded campaigns drove a 3.68x in July vs. 4.06x in June.

---

The Pattern Holds All Quarter

Across May, June and July, Branded was consistently ~10% of spend and ~40% of revenue, and the blended number always runs about 1.8x the unbranded reality:

Source: SearchLight

Spend ramped hard into cooling season — daily budgets jumped ~35% from May to June and held through July.

Through the 28th, unbranded search is pacing right in its usual range: ~$177 a lead at 3.68x.

---

A CPL “Drop” That Isn’t One

One more thing to watch out for, because it trips people up.

Blended CPL reads $107 in May, $107 in June, then $79 in July, which looks like a 26% win. It isn’t. A wave of YouTube/Reach view-based “leads” landed in July’s awareness bucket and pulled the blended average down; the real search cost per lead barely moved ($185 → $185 → $177).

Before you celebrate a CPL drop — or panic over a spike — ask what changed in the mix, not just the average.

---

PMax: Useful, But Read It Separately

Performance Max held near 12% of spend at ~7x all quarter — between branded and unbranded.

It’s worth running, but Google blends brand, shopping, and prospecting inside it and won’t fully break it out, so treat its ROAS as a ceiling and track it on its own line rather than folding it into “search.”

---

Questions That Make You the Sharpest Person in the Marketing Meeting

You don’t need to run the campaigns to read them well. Bring these to your next review — whoever manages your ads will have the answers, and asking makes you a better partner:

• “Can we see ROAS and CPL split by branded search, unbranded search, and PMax — not just blended?”

• “What’s our unbranded ROAS — the return on campaigns finding customers who don’t already know us?” (In SearchLight, you can also filter to new and existing customers based on customer profile created date).

• “How much of our leads and revenue come from brand-name searches vs. non-brand?”

• “For PMax, roughly how much is brand/shopping vs. true prospecting?”

• “When our blended CPL moves, what changed in the mix underneath it?”

That’s the whole SearchLight idea: we’re a neutral third party with no stake in the channels we measure, so we can show you the splits that make you a smarter operator and make the great work your team is doing easier to see clearly.

Want us to run this branded / unbranded / PMax split on your account against this ~600-business HVAC benchmark?

Reply to this email or grab time → searchlightdigital.io/book-a-demo

---

Methodology: SearchLight platform, Google Ads only, May–July 2026 (July through the 28th), drawn from ~1,067 Google Ads accounts and classified to HVAC (~600/month). Service (HVAC vs. plumbing/electrical/roofing) is assigned from the campaign name; campaigns without a service token inherit their account’s dominant trade. Branded vs. unbranded is determined from campaign naming conventions (”brand/branded/dba” vs. “nonbrand/non-brand/nb” or generic service terms) plus human review, with Performance Max and awareness (YouTube/Reach/Display) broken out separately. Awareness is excluded from CPL/ROAS lines because it reports view-based “leads.” Revenue = closed revenue on SearchLight’s closest-sale model. Pooled figures are spend-weighted; medians describe the typical business. Directional, not absolute.