Multi-Channel Capture, Triage & Instant Response
Every enquiry (inbound call, missed call, web form, chat widget, Google Business message, Facebook or Instagram message, text to the office number) lands as one customer record, is triaged by urgency rather than by arrival order, and receives a real answer within minutes on the channel the customer used, at any hour and at any call volume.
- Tier
- Foundational
- Build order
- 1st of 7 for HVAC
- Shape of it
- 8 steps, 1 decision point, 1 escalation rule, 1 handoff to a person, 1 stop rule
- At launch
- Runs supervised, with you approving what goes out
The problem this solves
HVAC demand is not evenly distributed and never has been. On the first hot weekend the phone rings in bursts, the CSRs are already on calls, and the overflow goes to voicemail that nobody leaves. Meanwhile the web form, the chat widget and the Google Business inbox are filling with the same problem from different people, and nothing connects them.
Worse, they all get treated the same. A no-cool call at a house with an infant in it queues behind a filter question, because the queue is ordered by whoever happened to be answered first rather than by what the job actually is.
How it works, step by step
Every wait, threshold and branch below is a value we set with you during the build, against your stages and your language. None of it is a default we impose.
- What starts it
- It decides something
- A timing rule you set
- It hands over to a person
- It escalates to a person
- A stop rule, so nothing closes itself
Step 1Trigger
A customer calls, misses, fills a form, or messages on any connected channel.
Step 2
Whatever the channel, one customer record is created, or matched to an existing one, so an existing member is never treated as a stranger.
Step 3
The record is stamped with the source, the address, the reported symptom, and the exact time it arrived.
Step 4Branch
The message is classified by urgency against a triage ladder agreed with the owner:
Most urgentLeast urgent- 1Safety or property protection (gas smell, water, electrical)
- 2No heat, with a vulnerable occupant or freezing conditions
- 3No cool, with a medical vulnerability or extreme indoor temperature
- 4Complete system failure
- 5Reduced performance
- 6Tune-up, comfort complaint, or non-urgent maintenance
Step 5Timing
Within minutes a reply goes out on the channel they used, naming the symptom they reported rather than opening with a generic greeting, and stating the diagnostic fee and the arrival window policy before the customer has to ask.
Step 6A person
Anything on the top two rungs is not answered by an automation. It is flagged immediately, routed to the on-call technician or the dispatcher, and the customer is told a person is being reached now.
Step 7Escalation
If a flagged safety call has not been acknowledged inside the agreed window, it escalates: dispatcher, then service manager, then owner.
Step 8Stop rule
Every enquiry that has not reached an outcome stays on a list until it does. Nothing closes itself silently, and a call that was abandoned before it was answered is on that list too, with its number.
How it gets built
Built inside what you already run
- ServiceTitan
- Housecall Pro
- Jobber
- FieldEdge
- or whatever your office already runs on
Nothing to log into and nothing to license. If a system needs a record your platform does not hold, we add the field to your platform rather than starting a second one beside it.
This is the actual build order, in the phases its own steps fall into. It runs in supervised mode first, with you approving what goes out, until you are happy with the tone.
- 1
Map
Inventory every channel an enquiry can arrive through, including the ones nobody officially owns: the owner's mobile, the old tracking number on a vehicle wrap, the Google Business chat.
Nothing is designed until we have looked at what already exists, including the parts nobody officially owns.
- 2
Build
Connect each channel to a single inbox and confirm a test enquiry from each creates one correctly attributed record with a lead source on it.
Built inside the software you already run, against your stages and your language.
- 3
Build
Build the matching rule against address and phone number so existing customers and members are recognised rather than duplicated.
- 4
Agree
Write the triage ladder with the owner and the service manager, and get the top two rungs signed off in writing. This is a safety review, not a copywriting exercise.
The thresholds, the wording and the names are yours. We write them down with you and get the consequential ones signed off.
- 5
Build
Build the classifier and test it against deliberately ambiguous messages ("it's making a noise and it smells funny"). Err toward escalation every time.
- 6
Agree
Set the escalation windows and name the person at each level, including who covers when that person is on a roof.
- 7
Build
Connect booking to live capacity with real job durations, and reserve capacity for emergencies rather than letting the day fill.
- 8
Prove
Run in supervised mode first, with every outbound message reviewed before it sends, until the owner signs off on the tone.
It runs with a person approving what goes out until you are happy with the tone. Nothing sends unreviewed on day one.
What changes after it goes live
How it runs today
HVAC demand is not evenly distributed and never has been. On the first hot weekend the phone rings in bursts, the CSRs are already on calls, and the overflow goes to voicemail that nobody leaves. Meanwhile the web form, the chat widget and the Google Business inbox are filling with the same problem from different people, and nothing connects them.
After this one is live
Enquiries stop being ordered by luck. The 9pm no-heat call and the 11am filter question stop being the same event. The CSR opens one list instead of six inboxes, and the list is already sorted by what the job is. Safety calls reach a person faster than they did when a human was triaging, because the routing no longer waits for somebody to be free.
The abandoned calls become visible for the first time, which is usually the single most uncomfortable and most useful thing this system does.
How to measure whether it worked
Your arithmeticRun with your numbers, not ours
Run this with the contractor's own numbers. Pull last peak month's inbound contacts, count how many were abandoned or never returned the same day, and apply the company's own booking rate and average service ticket. Then apply the replacement conversion rate to the share that would have been a changeout. That figure is the recurring cost of the current response gap, and it is demand already paid for.
We agree the baseline before anything is built, and we do not take credit for things that were going to happen anyway. There is no figure on this page claiming what we have produced for somebody else, because there is no verified figure to publish.
What we will not do
This is from the same delivery document as everything above it. It is on the page because a supplier who has not thought about it will not tell you, and you would find out later.
Every system here sends automated messages to consumers in the United States, which puts all of them inside the TCPA and inside carrier A2P 10DLC registration. These are two different things and both apply. Registration is a carrier trust requirement; it does not create consent and does not make an unlawful message lawful. Unregistered application traffic gets filtered or blocked.
Classify every system before it sends. A service message is narrowly connected to an existing job: appointment confirmation, technician on the way, part arrived, invoice. A marketing message promotes something: a tune-up offer, a replacement promotion, a membership push, an estimate reactivation sequence. The standards of consent are different, and a promotional line inside a service template converts the whole message into marketing.
The minimum a compliant build has to do:
- Consent recorded per number with the date, the source, the exact wording shown, the sending brand, and whether it covers service, marketing, or both. Marketing needs prior express written consent with an unchecked opt-in, and consent cannot be a condition of purchase.
- A2P 10DLC registration of the brand, the numbers, and a campaign use case that matches the traffic actually sent. An appointment-reminder campaign does not quietly become a seasonal promotion campaign.
- Quiet hours of 8am to 9pm in the recipient's local time, not the office's, with queued sends held until the window opens. Apply the same gate to service messages by default and handle genuine emergencies as a named exception.
- Opt-out handling for STOP, UNSUBSCRIBE, CANCEL, END, QUIT, REVOKE and OPT OUT, and for free text like "stop texting me". Suppress immediately in practice rather than to the outer limit of the rule, cancel messages already queued rather than only future ones, and propagate the suppression to every sending number, system and platform the contractor controls.
- Replies routed to a monitored human queue, always. A reply that says the heat is off is not a marketing response.
- Human review before any sequence that would materially pressure a prospect, which in this playbook means System 4.
- Suppression reconciled across systems, so a customer cannot opt out in the field service platform and keep receiving messages from the marketing tool.
- Purchased or scraped lists are not messaged. Consent has to name the actual sender.
Nothing in this document is legal advice. State mini-TCPA statutes, carrier policy, and the exact dialling or sending technology can each add requirements, and the rules in this area have moved more than once recently. Every template that states a commercial term, a price, or a guarantee requires review by the contractor's own counsel before it goes live.
Nothing here is legal advice. Rules in this area have moved more than once recently, and every template that states a commercial term or a guarantee goes to your own counsel before it goes live.
Seven systems forHVAC.
We build one at a time and prove it moved before starting the next. The tiers are the dependency order, not a price list.
Tier 1Foundational
Nothing arrives late or unowned. These come first because everything above them assumes they are true.
- 1Multi-Channel Capture, Triage & Instant Responseyou are hereFastest visible result, and it is the one that survives the first heat wave, which is when the owner decides whether any of this was real.
- 2Dispatch Confirmation, Arrival Protection & Technician BriefProtects the capacity and the truck rolls already paid for.
Tier 2Growth
The recoverable money. These work the pools the foundational systems have made visible for the first time.
Tier 3Flagship
One connected system end to end, plus what the owner reads on a Monday. Only once the pieces are proven individually.
- 5The Complete Service-to-Install SystemThe flagship, once the pieces are proven individually.
- 6Service Agreement Renewal, Fulfilment & ReactivationCompounds everything above, and depends on the equipment data System 2 started collecting.
- 7The Owner's Morning BriefOnly meaningful once there are systems to report on.
The tiers are the dependency order for HVAC, not a price list. Most firms do not start at the first one, because the order is a default and the call is where it gets changed.
A note on sequencing this trade
Do not build all seven at once.
Each stage is proven against its agreed measure before the next begins.
Season note. Build Tier 1 in the shoulder season and let it run through one peak before starting Tier 2. A contractor who sees System 1 hold on the first hot Saturday will build the rest without being sold again.
Back to the HVAC overview for the stage map and where these fit.
Is this the oneyou need first?
Often it is not. On the call we look at what is actually costing you most right now, which is frequently a different system from the one that brought you to this page. If there is nothing worth building yet, we will say so.