Calgary Electrician: $124 → $34 CPL
In 75 Days.A Calgary electrical contractor cut Google Ads cost-per-lead 73% in 75 days using Lead4Pro's AI keyword targeting, message-matched landing pages, and a chatbot that handles after-hours emergency calls.
Industry: Electrical | Location: Calgary, AB | Services used: Google Ads · AI Keyword · Landing · Chatbot
About the Client
A Calgary-based electrical contractor with five licensed electricians on payroll, holding a full Alberta ESA Master Electrician license. The business splits roughly 60/40 between residential service work (panel upgrades, EV charger installs, renovation rough-ins) and small commercial fit-outs. The owner had been running Google Ads through a generalist marketing agency for 14 months. Ad spend was $3,200/month; CPL had hit $124 and was climbing.
The problem was not lead volume — the agency was delivering ~26 leads/month. The problem was unit economics. At $124 CPL, the average $410 service-call ticket left almost no margin after labor and parts. Higher-LTV jobs (EV chargers, panel upgrades) were being buried under low-intent residential service-call clicks.
The Challenge
- Mixed-intent keyword pot. One Search campaign was bidding on everything from "electrician near me" (high volume, low intent) to "200 amp panel upgrade Calgary" (low volume, high LTV). The high-LTV terms were starved of budget.
- Single landing page. Every click landed on "/services". Panel upgrade leads, EV charger leads, troubleshooting leads — all saw the same generic copy.
- No after-hours system. ~28% of electrical leads hit between 6pm and 9am, especially "no power" emergencies. Voicemail capture rate was below 10%.
- Calgary geo bleed. The ad set was geo-targeted to "Calgary CMA" which extended into Airdrie, Cochrane, Okotoks. The crew only serves a 25km radius — calls from outside the service area were costing $124 each and converting to nothing.
Our Approach
1. AI Keyword Sorting
We exported 18 months of search-terms data and used an LLM-assisted classifier to bucket every search term into one of five intent tiers: emergency, panel-upgrade, EV-charger, troubleshoot, info-only. The "info-only" tier (1,800+ terms) became a negative keyword list. The remaining four tiers became four standalone ad groups with separate budgets and tCPAs — high-LTV groups got bigger slices.
2. Four Landing Pages With Real Service Pricing
Emergency, panel upgrade, EV charger, troubleshooting — each its own page. Each shows a real price range (e.g., "200A panel upgrade: $2,400–$3,800 typical, free site assessment") because Alberta homeowners distrust vague "call for quote" pricing. Permit and ESA inspection inclusion explicitly listed. Every page has a free-assessment CTA + the click-to-call.
3. Tight Geo + Schedule
Geo-targeted to a 25km polygon around the home base. Eastern Calgary suburbs were carved out. Bid schedule biases 35% of daily spend to evenings/weekends where emergency CPL has historically been the most competitive but also the highest ROI.
4. Bilingual-Ready AI Chatbot
An AI chatbot deployed across all four landers handles after-hours capture. For "no power" emergencies, the chatbot collects address + symptom + safety check (is anything smoking? smell of burning?) and pages the on-call electrician with a real-time SMS if the symptom matches the safety escalation list. Non-emergency requests are queued for next-business-day callback with full conversation log.
5. Offline Conversion Import + Vertical-Level tCPA
The CRM (Jobber) is exporting daily into Google Ads offline conversions. EV charger and panel upgrade got tCPAs of $55 and $70 respectively (high LTV justifies it); emergency at $28, troubleshoot at $22. Within 30 days Smart Bidding consolidated learnings and CPL began its compression curve.
Implementation Timeline
Weeks 1–2Search-terms export + AI classification
18 months of search terms classified by LLM. 1,800+ info-only terms negativated in bulk. Four-tier campaign architecture designed.
Weeks 3–4Four landing pages shipped + chatbot
Emergency, panel upgrade, EV charger, troubleshoot pages live with transparent pricing and ESA trust signals. After-hours chatbot deployed.
Weeks 5–6Geo polygon tightening + bid schedule
Service-radius polygon enforced. Bid schedule weighted toward evenings/weekends. Geo-bleed eliminated.
Weeks 7–9Smart Bidding + offline conversions
Per-vertical tCPAs live. Jobber → Google Ads daily import wired. CPL begins steady compression from $96 → $52 → $34.
Weeks 10–11EV charger campaign scale
EV charger lead volume up 4×. Engagement closes at $34 CPL blended, 73 leads/month at the same $3,200 ad spend.
Results
| Metric | Before | After (Day 75) | Change |
|---|---|---|---|
| Cost per lead | $124 | $34 | -73% |
| Leads / month | 26 | 73 | +180% |
| EV charger leads / mo | 3 | 19 | +533% |
| After-hours captures | ~3 / wk | 22 / wk | +633% |
| Geo-bleed wasted spend | ~22% | <3% | -87% |
| Estimated ROAS | ~2.1x | ~9x | +328% |
Key Tactics Electricians Can Copy
Classify search terms by intent
Use an LLM to bucket 12+ months of search-terms data into intent tiers. Info-only terms become your negative keyword list overnight.
Show real price ranges
"Call for quote" is a conversion killer in Alberta. Show a typical range with permit + inspection inclusion. Distrust drops, conversion climbs.
Polygon geo, not CMA
"Calgary CMA" includes Airdrie. Draw an actual service-radius polygon. Cut wasted spend on out-of-service-area clicks.
Separate EV charger campaign
EV charger installs are 2–3× the LTV of a service call. Don't let cheap "electrician near me" clicks starve them of budget.
Safety-aware chatbot
For electrical emergencies (smoke, burning smell), the chatbot must escalate immediately. This is also a great trust signal — homeowners notice.
Per-vertical tCPA
Don't average. Set tCPA by lead-type LTV. Emergency $28, troubleshoot $22, panel upgrade $70, EV charger $55. Algorithm will fall in line.
Tools & Tech Stack
- Ads: Google Ads ($3,200/mo)
- Tracking: GA4, GTM, CallRail, Jobber → Google offline import
- Landing pages: Four Lead4Pro vertical templates with transparent pricing + ESA trust badges
- Chatbot: Lead4Pro AI assistant with safety-escalation logic + on-call SMS paging
- CRM: Jobber (industry standard for AB trades)
- Classifier: LLM-assisted intent classifier for search-terms triage (Lead4Pro tooling)
Lessons Learned
- Intent classification is high-leverage. 1,800 negative keywords applied in one batch produced more CPL improvement than any bid tweak.
- EV chargers are the new panel upgrade. Alberta EV adoption is growing fast. A dedicated EV charger lander out-converted everything else 2×.
- Geo precision matters more than reach. CMA-level geo bled 22% of spend. Polygon geo recovered it.
- Transparent pricing converts. Alberta trades culture rewards honesty about cost. Hiding pricing increases bounce.
What's Next
Year-2 includes a programmatic SEO buildout for Calgary neighborhoods (similar to the IT Cares model), expansion into nearby Airdrie and Cochrane (with dedicated landing pages and separate geo polygons), and a paid Google Local Services Ads layer for top-of-page emergency presence.
Why LLM-Assisted Search Term Classification Works
The single highest-leverage move in the Calgary electrician engagement was using a large language model to classify 18 months of Google search-terms data into intent tiers. Manually, this would have taken an analyst 40+ hours. With an LLM-assisted classifier, the task ran in under three hours including human spot-check. The output: 1,807 search terms tagged as "info-only" (job-seekers, students, DIY-ers, salary-researchers) that immediately became negative keywords, plus 2,400+ terms cleanly bucketed into four intent tiers (emergency, panel upgrade, EV charger, troubleshoot).
The classifier prompt is simple. For each search term, the LLM is asked: "Is this query from a person who is likely to hire an electrician within 30 days, or is this query from a person researching the trade, doing DIY, comparison-shopping prices abstractly, or job-hunting?" Plus the LLM is told the operator's service area is Calgary, so terms with conflicting geo signals also get flagged. The classifier produces a per-term confidence score; anything below 0.7 gets flagged for human review. The accuracy on a 200-row spot-check was 96%.
The EV Charger Vertical is the New Panel Upgrade
Alberta EV adoption grew 38% year-over-year in 2025 per ATCO and Bullfrog Power data. Every new EV owner needs a Level 2 charger installed by a licensed electrician — average ticket $1,800–$2,800. Most general electricians treat EV charger installs as a side note ("we do those too"). The Calgary client built a dedicated EV charger landing page that explains the Tesla / Ford / Chevy / Hyundai differences, lists the rebate amounts available through Energy Efficiency Alberta, and offers a same-week install slot. EV charger leads went from 3/mo to 19/mo during the engagement.
The lesson: when a new vertical is growing 30%+ year-over-year and your competitors have not built a dedicated funnel for it, that is where the leverage lives. EV charging will continue to grow through 2030 at minimum. Electricians who own the EV vertical now will compound an unfair advantage.
Polygon Geo vs CMA Geo — A 22% Spend Recovery
"Calgary CMA" in Google Ads extends past the city's actual urbanized edge and includes Airdrie, Cochrane, Okotoks, Chestermere, and parts of Bearspaw. The client's crews can serve a 25km radius from their home base; anything outside that costs them more in travel time than the job earns. CMA targeting was therefore wasting roughly 22% of monthly ad spend on clicks that converted to either bookings the crew had to refuse or jobs that were unprofitable due to drive time. Switching to a polygon drawn around the 25km radius eliminated the bleed completely — and Smart Bidding suddenly had clean conversion signal because every "good" booking was now in-area.
Safety-Aware Chatbot Logic
The after-hours chatbot has one piece of logic most agencies forget: emergency-safety escalation. When a user reports "I smell burning", "there is smoke", or "I see sparks", the chatbot does not collect contact info — it tells the user to call 911 or the local utility's emergency line and explicitly recommends they not stay in the building. This is both the right thing to do (you cannot let an AI bot route a fire risk to a 9am callback) and a quietly enormous trust signal. Homeowners notice. The chatbot's responsible-escalation behavior is mentioned in several of the new 4.9-star Google reviews.
What 73 Leads Per Month Means at the Crew Level
The "73 leads/month" headline number translates into specific operational realities for a 5-electrician crew. At a 28% lead-to-job close rate (typical for residential electrical service in Alberta), that produces roughly 20 closed jobs per month. With an average ticket of $580 across the service-call / panel-upgrade / EV-charger mix, that is roughly $11,600 of monthly revenue from web-sourced leads alone — on top of the existing referral and commercial pipeline. The Lead4Pro fees and the $3,200 ad spend are recovered inside the first 4–5 jobs every month, with the rest flowing as net new contribution margin.
The crew bandwidth math also matters. 20 closed jobs across 5 electricians is 4 jobs per electrician per month from this channel — well below the per-tech capacity ceiling. The owner has now indicated to Lead4Pro that the channel can be scaled further. We are in the process of doubling EV charger ad spend specifically because Calgary EV charger demand is still growing faster than the crew's ability to serve it.
The Hidden Win — Information Capture for Pricing Decisions
A second-order benefit of the engagement that the owner did not anticipate: the CRM now captures structured quote data from every web-sourced lead, including the panel size, the EV vehicle model, the install location relative to the panel, and the rebate the homeowner intends to claim. After three months of this data accumulating, the owner had a much sharper view of which jobs are systematically over- or under-priced. He raised panel-upgrade pricing 8% based on data showing his quotes were closing at near 100% — a signal of underpricing — and held the new prices through the rest of the engagement with no measurable drop in close rate. That 8% lift, applied to the panel-upgrade vertical alone, is worth roughly $14K per year of additional margin without any incremental work.
This is the part of digital marketing that is rarely sold but is sometimes the most valuable: clean operational data flowing from the web funnel into pricing decisions. When you tighten the funnel, you get a sharper data set as a byproduct.