This article argues that incrementality—not platform-reported ROAS—should guide HVAC, plumbing, and electrical marketers because it identifies ads that create truly new calls and booked jobs. Traditional metrics like CTR, impressions, branded search ROAS, and automated campaign performance often overstate impact by capturing demand that would have occurred anyway due to brand, SEO, or weather. The piece outlines practical measurement methods, emphasizing randomized and geo holdout tests, and prescribes tracking unique calls, booked jobs, close rate, average ticket, and margins. Marketers should calculate incremental lift, cost per incremental job, and incremental ROAS, then shift budget to channels and tactics that beat profitability thresholds, using simple geo splits on Google and Meta when possible.

Stop Chasing ROAS Mirages: Incrementality Is What Pays the HVAC/Plumbing/Electrical Bills

TL;DR Incrementality tells you which ads create new calls and booked jobs—not just take credit for demand that was coming anyway. Run simple holdout tests by audience or geo, measure cost per incremental job, then reallocate budget to what moves the needle. Calls, not clicks.

What Incrementality Really Means (Contractor Edition)

Incrementality measures the causal impact of your marketing. Translation: What changed only because your ads existed? If your AC repair calls would’ve happened without ads—because a heatwave hit or your trucks are everywhere—your spend didn’t create value; it took credit.

For home services, the “incremental” unit is the booked job: water heater replacements, drain clears, panel upgrades. If ads don’t increase the number of booked jobs (or profitable revenue) versus a proper control, they’re not working—no matter what your dashboard says.

“Calls, not clicks.” And only the calls that wouldn’t have happened without the ad.

Why Your Dashboards Lie

• CTR, impressions, and platform ROAS look great while stealing credit from brand equity, SEO, or weather-driven demand.
• Branded search often captures people already looking for you. That’s protection, not creation. Useful—just don’t call it growth.
• Automation (PMax, Advantage+, broad match) hunts the lowest-friction conversions, which frequently means demand that already exists.

Incrementality cuts through that. It shows the lift: incremental calls, incremental jobs, incremental revenue. If the lift is flat, you’re buying vanity.

How to Measure Incremental Lift

1) Randomized Holdout (Best for Digital)

Split your audience into test (sees ads) and control (doesn’t). Same period, same conditions. Compare booked jobs and revenue. This is the cleanest read on causal impact.

2) Geo Holdout (Best for Local Service Areas)

Divide service zips or cities into test/control. Run the campaign in test geos only. Watch for spillover and shared media. Great for Search, YouTube, and Meta when your business is location-bound.

3) Synthetic Control/Causal Modeling

When you can’t run true holdouts, use historical data and matched controls to estimate impact. Helpful, but easier to bias. Treat as directional, not gospel.

4) Marketing Mix Modeling (MMM)

Statistical modeling across channels using long-run data. Useful for multi-location brands with enough volume. Overkill for a single-shop contractor; fine for regional groups.

A Simple Incrementality Plan I’d Use

1. Hypothesis: “Non-brand search for ‘AC repair’ increases incremental booked jobs and revenue in Northside versus Southside.”
2. Design:
   • Pick comparable geos (income, housing stock, seasonality). Northside = Test; Southside = Control.
   • Run the campaign only in Test. Keep budgets stable elsewhere. Avoid overlapping media.
   • Lock creative, bids, and landing pages for the test window. Don’t change five things at once.
3. Run long enough: Minimum 4–6 weeks, across a full billing cycle and enough volume (ideally 100+ calls per cell) to beat random noise.
4. Track what matters:
   • Unique call tracking by geo.
   • Booked jobs, close rate, average ticket, and gross margin.
   • Exclude brand terms or measure brand separately.
5. Calculate lift:
   • Incremental jobs = (Jobs in Test − Jobs in Control) adjusted for baseline differences.
   • Cost per incremental job = Ad spend in Test / Incremental jobs.
   • Incremental ROAS = Incremental revenue / Ad spend in Test.
6. Decide: If cost per incremental job is below your target (say, 15–25% of average ticket or aligned to your gross margin), scale. If not, pause or rebuild.

Platform Support (Keep It Simple)

• Google: Geo split tests are practical for Search and YouTube. Conversion Lift exists mostly for YouTube/Display. Build clean brand vs. non-brand guardrails.
• Meta: Conversion Lift and brand lift studies help, but geo split with clear objectives is often faster for contractors.
• TikTok/Amazon: Mixed support. Use geo splits or structured on/off periods and read the phones
  

  Why incrementality is the only metric that proves marketing’s real impact