How to Set Up Emergency Call Routing for a Contractor

Do this

• List every phone number that appears on Google Business Profile, your website footer, vehicle wraps, invoices, and yard signs.
• For each number, document the carrier (cell carrier, VoIP provider, or business line), the after-hours behavior, and who currently owns the device or extension.
• Place a test call to each number during business hours, after 6pm, at 11pm, and Sunday morning. Record what happens at each step (rings, transfers, menus, voicemail).
• Pull the last 30 days of call-log data from your carrier and identify which numbers actually carry call volume versus which are vanity routes.
• Note any call that ended in voicemail, an abandoned tree, or an unanswered ring at 5 or more rings: those are the dead-end paths the new routing must replace.

Do this

• Pick the top 5 to 10 phrases your real callers use that indicate a true emergency, by trade.
• HVAC urgency triggers: gas smell or gas leak, no heat with kids or elderly in the home below freezing, carbon monoxide alarm, water leaking from the air handler onto a ceiling, full system failure during a heat advisory.
• Plumbing urgency triggers: sewer backup with sewage in the home, active leak you cannot shut off, no water to the building, burst pipe behind a wall, water heater leaking onto a finished floor.
• Electrical urgency triggers: sparking outlet or panel, burning smell from outlet or switch, power out to a medical device, partial outage with breakers tripping repeatedly, smoke from any electrical equipment.
• Roofing urgency triggers: active interior leak during current rain, missing shingles after storm with rain forecast, tree-on-roof or limb-through-roof, ceiling sag, water pooling on flat-roof drains during a storm.
• Appliance-repair urgency triggers: fridge or freezer not cooling with food at risk, oven gas smell, washer leaking onto finished flooring, dryer overheating with a burning smell.
• Write each trigger as a short phrase the AI or live agent can match in transcript ('says "smell gas"', 'says "no heat" + "baby"'), not a clinical category.

Do this

• Mandatory fields: caller name, callback number (formatted dialable), service address, problem description in the caller's own words, urgency tag, time of call, transcript link.
• Trade-specific fields: HVAC adds equipment age and last service date; plumbing adds water-shutoff status; electrical adds 'is power off in the room or whole house'; roofing adds 'is the leak active right now'.
• Decision fields: 'Can the caller shut off water / power / gas?' Yes or No. 'Is anyone unsafe?' Yes or No.
• Cap the SMS to 12 to 16 lines so the tech does not have to scroll on a phone lock screen. Anything longer than 16 lines belongs in a transcript link, not the page.
• Template the format and pin it. Every emergency packet on every page should look identical so the tech recognizes it at a glance, even half-awake.
• Sample format that fits on a single iPhone lock-screen preview: '[URGENT] HVAC no-heat - Jane Smith, 555-867-5309, 1234 Oak St, says furnace not igniting, kids in home, 2yrs old, last serviced 2024. Heat off, gas on. Transcript: bit.ly/xyz. Reply Y to confirm.'

Do this

• Set up a 4-week rotation: tech A nights weeks 1 and 3, tech B nights weeks 2 and 4, owner as escalation for unacknowledged pages.
• Place 3 test calls per shift: one at 6:30pm (just past close), one at 11pm, one at 4am. Use the language from your urgency-trigger list so the system flags them correctly.
• Verify each test call: caller hears the right greeting in your business name, AI completes intake in under 90 seconds, urgency flag fires correctly, SMS arrives within 30 seconds of caller hanging up, tech acknowledges within 5 minutes.
• Test the escalation: do not acknowledge a test page on purpose. Verify the second-tier contact is paged within the configured window (typically 8 to 12 minutes).
• Test the boundary: place a non-urgent call ('quote for a panel upgrade next month'). Verify the system does NOT page anyone and instead emails or queues a morning packet.
• Test holiday and overnight transition: place a call at 11:55pm Saturday and at 12:05am Sunday. Verify the rotation rolls correctly across the date boundary.
• Document every misfire in a small spreadsheet (date, scenario, expected vs actual). Fix the highest-volume misfire first, retest, then move down the list.

Do this

• Define the abandonment formula clearly: (calls answered with no name + address + problem captured) / (calls answered). Calls that disconnected before greeting do not count as abandonments because they never entered the system.
• Pull abandonment rate every Monday for the trailing 7 days. Segment by hour (after-hours vs business hours) and by trade if you run multiple trades.
• Set thresholds: abandonment under 8% means routing is healthy; 8 to 15% means a script or pacing tweak is needed; over 15% means stop and audit.
• Common abandonment causes: greeting too long (cut to under 8 seconds), AI asks address before urgency (flip the order, ask 'are you safe and what's the issue' first), tech does not call back fast enough so callers leave a message and abandon on the next ring.
• Pair the abandonment metric with first-response time (page to acknowledgement) and call-to-dispatch time (call ended to truck rolling for true emergencies). Three numbers reviewed weekly is a coverage dashboard.