Calculate the cost of Zanus AI for an HVAC business from a written supplier quote and a reviewed sample of your own calls. The useful comparison includes implementation, telephone services, staff handoffs and record correction. It also distinguishes a booking recorded by the system from a visit the office can actually deliver. A low cost per answered minute is not enough to establish that the complete phone process costs less.
This guide gives you a worksheet and a method for collecting its inputs. It does not quote a Zanus subscription price, estimate typical HVAC wages or predict a saving. Where your business has not measured an input, leave it unresolved and identify who will supply it. An explicit unknown is easier to resolve than an invented assumption hidden inside a persuasive return-on-investment figure.
Define the phone process you are comparing
Choose a business boundary before opening a spreadsheet. An inbound repair enquiry might begin when a homeowner calls and end when the service coordinator accepts the request. A replacement enquiry might need an estimator to review the property and agree a survey. A maintenance reminder might end with a confirmed response against an existing appointment. Those are different units of work, even when they share the same business number.
Write down the start event, the required information and the final responsible person for each route. Include the point at which the business makes a commitment to the customer. A transcript, a callback task and a confirmed visit should have separate labels. This prevents a comparison in which the manual process is charged for arranging a visit while the proposed AI process is credited merely for answering the telephone.
For an illustrative pilot, use three routes: a new repair request, an existing customer asking about a visit and a homeowner requesting a replacement estimate. Keep them separate in the worksheet. You can combine their costs later using the actual mixture observed in your sample, rather than assuming that every conversation has the same duration and outcome.
Establish an honest current baseline
Record how the office handles those routes today. Include who answers, who takes over and who enters the information into the system the business uses for work. Some calls may interrupt a dispatcher or technician rather than a dedicated receptionist. That time still matters, but it should be recorded against the actual role performing it.
Use a representative, clearly bounded observation period. Note closures, unusual weather, a temporary staffing gap or an advertising campaign that may affect the sample. A busy period can be useful evidence for peak capacity, but it should not silently become the assumed pattern for the whole year. Keep the dates and circumstances beside the figures so another person can assess the comparison.
Separate observed time from a staff estimate. If the team cannot measure every interruption, ask them to log a small set of examples and label the remaining estimate. This gives you a starting point without pretending the baseline is more precise than the evidence supports.
Use cost categories that do not overlap
| Category | Input to obtain | Evidence owner |
|---|---|---|
| Supplier subscription | Scope, term, included use and exclusions | Buyer reviewing the proposal |
| Telephone services | Number, call, transfer and related charges | Telephony account owner |
| Implementation | Configuration, source preparation and testing work | Implementation lead |
| Office handling | Staff time by role and route | Office manager |
| Correction and reconciliation | Time spent resolving incomplete or wrong records | Service coordinator |
| Ongoing maintenance | Knowledge updates, training and periodic review | Named system owner |
Give each activity one home in the worksheet. If a coordinator reviews a transcript while arranging a callback, do not charge the same minutes as both transcript review and callback preparation unless they are genuinely separate activities. Similarly, determine whether a supplier's implementation fee already includes the configuration work you are considering as an additional internal cost.
Read the supplier proposal as an operating specification
Ask for the proposed module scope, service term, billing unit and included usage in writing. Establish what happens when usage exceeds an allowance or a required feature falls outside the selected package. Do not copy a general market rate into a cell labelled Zanus AI. The signed proposal and applicable terms should provide the commercial inputs for the proposed configuration.
The official Zanus AI Phone page describes telephone answering and departmental handoffs. The Auto-Call page describes outbound call orders and their results. Use those product descriptions to identify which workflows you want demonstrated, then confirm how the supplier prices that exact scope. A product page and a customer-specific commercial agreement answer different questions.
Record exclusions alongside inclusions. Number arrangements, connected systems, source preparation, unusual configuration work and ongoing support may need their own clarification. An unresolved item should remain visible in the purchasing decision rather than being treated as free because the first quotation did not mention it.
Calculate staff handling by role
For each route, multiply the observed staff handling time by the business's approved loaded cost for the relevant role. Use a consistent unit. If the time is recorded in minutes and the role cost is hourly, convert the minutes to hours before applying the rate. Keep the original time observations so the calculation can be checked independently.
A loaded labour input is a costing assumption chosen by the business, not a universal wage benchmark. Ask the responsible finance or management colleague which elements the comparison should include. Explain the treatment of salaried staff, contractors and work that interrupts another responsibility. The objective is a consistent decision model, not a claim that every saved minute immediately becomes cash.
Separate time released from expenditure removed. If the same office team remains in place, the practical benefit may be additional capacity or fewer interruptions. Do not label that benefit a payroll saving unless the business has a specific, justified change to expenditure. This distinction is especially important when a proposal assumes a receptionist or dispatcher disappears from the cost base.
Track the complete handoff
In a controlled repair enquiry, follow the conversation until the coordinator has enough information to act. Record the service address, the caller's description, the proposed next step and the person responsible. If the conversation is transferred, include the staff time after acceptance as well as any later record completion. The transfer itself is only one event in the process.
Repeat the test when the intended recipient is unavailable. Inspect what the caller hears, what record is created and who monitors the unresolved request. A message without a responsible person can appear inexpensive because no immediate staff action occurred, while merely moving the workload into a less visible queue.
Keep urgent-contact handling within the business's approved instructions and the responsible team's process. The costing exercise should not introduce a technical diagnosis, an improvised safety script or an unsupported promise of dispatch. Price the demonstrated administrative route, including the human action it requires.
Use a sample ledger for each call
| Field | Reason to keep it |
|---|---|
| Sample reference and date | Recover the evidence used in the calculation |
| Request type | Compare similar customer needs |
| Conversation outcome | Distinguish answered, captured, transferred and completed work |
| Staff actions | Count review, clarification and customer follow-up |
| Record destination | Check where the usable enquiry ended up |
| Remaining question | Identify work still outstanding |
The ledger does not need to reproduce every word of the conversation. It should point to the evidence and make the cost categories recoverable. Use test records or appropriately controlled business information, and follow the organisation's access and retention requirements. The comparison should not create a second uncontrolled archive of customer conversations.
Build the recurring-cost equation
Start with the recurring supplier charges and telephone services applicable to the observation period. Add staff handling, review, correction and the operating work needed to keep the setup current. Divide that total by the number of reviewed enquiries that reached the defined completed state. Keep unsuccessful attempts and unresolved requests visible beside the denominator.
Written symbolically, recurring cost per usable enquiry equals recurring system cost plus applicable telephone cost plus staff operating cost, divided by usable enquiries. The symbols do not make the input assumptions disappear. Keep a source reference, measurement period and uncertainty note beside every input so a colleague can reproduce the result.
For an illustrative comparison, use the same set of test requests with the current process and the proposed process. Do not compare a carefully selected AI demonstration with every difficult call in the office's baseline. Where the routes differ, explain the difference rather than forcing the results into one headline figure.
Account for implementation without hiding it
Implementation can involve preparing approved answers, confirming service areas, checking calendar rules, mapping records, training staff and reviewing test outcomes. Assign responsibility for each item. Ask which work the supplier performs, which work remains with the business and which decisions cannot be delegated because they concern the actual service commitment.
Keep one-time effort separate from recurring effort. You may decide to spread implementation cost over a chosen planning period, but show that assumption explicitly. Present the original implementation amount and the chosen allocation period alongside the resulting monthly equivalent. Otherwise a favourable-looking monthly comparison can conceal an important initial commitment.
Include the work of returning to the previous arrangement if the pilot is unsuccessful. The business should know who controls the number, which records need to be recovered and what staff need in order to resume handling calls. These are implementation questions to demonstrate, not claims that a particular supplier offers a specific migration or rollback feature.
Test the assumptions most likely to change the result
Longer conversations, repeated callbacks and more staff handoffs can change the result even when the advertised subscription stays the same. Model those inputs separately using your observed range or clearly labelled planning assumptions. Show the effect on staff capacity and usable enquiries as well as the effect on the bill.
Do not assume that a transfer is a failure or that an AI-only conversation is a success. An appropriate handoff can be the correct result for a technical or commercial question. An apparently complete conversation can still create the wrong address, wrong visit category or an unsupported arrival promise. Inspect the business result before assigning a cost advantage.
Keep uncertain opportunity value outside the core operating-cost total. A missed enquiry may represent potential future work, but the value and likelihood require separate evidence. Present any opportunity scenario as a scenario, with its own assumptions, rather than combining speculative revenue with observed expenditure and calling the sum a measured saving.
Reconcile a supplier bill with the sample ledger
Before extending the cost model to a larger rollout, reconcile a controlled period of usage with the supplier's bill or account usage statement. Identify which line represents the telephone service, which represents AI use and which represents another service. Ask the account owner to explain any amount that cannot be tied to a billing rule or a known piece of work.
Use the actual billing units. A conversation's spoken duration may not be identical to every duration appearing on a telephone statement, especially when a transfer or another call segment is involved. The worksheet should describe the billing treatment confirmed for the proposal rather than assuming that a single duration explains the complete charge. Retain the supplier's explanation so the office can repeat the reconciliation later.
Inspect unsuccessful attempts as well as completed conversations. A request may be retried, reach voicemail or fail before the intended next step. Record what the applicable terms say about each event, then compare that explanation with the controlled usage record. If the account cannot reconcile an event reliably, keep the corresponding cost unresolved instead of extrapolating a rate from a successful demonstration.
Cost a change to the approved business information
Use a simple maintenance scenario before accepting the ongoing-cost estimate. For example, the HVAC business changes its service area or temporarily closes a particular appointment calendar. Identify who approves the change, where the authoritative information is updated and which test confirms that callers receive the new answer. Record the staff effort required to complete that sequence.
Then inspect a request already waiting in an office queue. A customer who enquired before the change may still need a response under an earlier arrangement. Ask whether staff can recover the relevant context and decide what should happen next. Updating an answer source is not necessarily the end of the office work created by a policy change.
Keep this maintenance scenario separate from the initial configuration charge. It provides evidence about the recurring responsibility of operating the system. Assign a named owner and a review trigger, such as a changed service policy, staff contact or booking rule. The resulting estimate should explain the work the business expects to perform, without inventing a fixed number of maintenance hours for every HVAC company.
Review the evidence with the people doing the work
Ask a service coordinator, an estimator and the account owner to review the same sample records. The coordinator can identify incomplete intake, the estimator can identify ambiguous replacement enquiries and the account owner can reconcile charges. Disagreement is useful evidence about the process, especially when a record appears complete to one reviewer but leaves another person unable to act.
Document changes made during the pilot. If you adjust the approved answers or the routing policy, retain the version associated with each sample. Otherwise the final result can mix several different configurations without showing which one produced the usable outcome. A reproducible comparison is more useful than an optimistic average built from incompatible trials.
Finish with a decision record. State which routes are supported by the evidence, which questions remain unresolved and what commercial commitments have been confirmed. You may choose a restricted rollout, further configuration work or a different approach for particular calls. The decision should follow the demonstrated workflow and the business's actual costs, with a named owner for the next review.
Official sources and related guides
This independent Sysevo guide uses the following official documentation. The workflow and trial scenarios are evaluation proposals, not reported customer results or a hands-on benchmark.
- Zanus AI official product documentation
- Zanus AI official supporting guide
- GoHighLevel Voice AI for HVAC: Test Delays Before Launch
- Zanus AI HVAC Pilot: Score Repair and Estimate Call Outcomes
- Zanus AI for home services
Compare the same requirements with the relevant Sysevo product, Voice Agents and CRM. Review Sysevo plans and pricing or book a demo to discuss the proposed setup. Sysevo sells its own business software and is not affiliated with the competitors discussed here.


