Results
Case studies, and how we will measure them
There are no HVAC case studies on this page yet, because no deployment has completed a measurement period. Rather than fill the space with logos or percentages nobody can check, this page publishes the standard the first one will be held to: what gets measured, against what baseline, over what period, and which claims we will not make whatever the numbers say.
The empty state, stated plainly
Nothing here yet, and that is not an oversight
A case study needs a customer, a baseline taken before launch, a measurement period of the same length and season afterwards, and the customer's agreement that the figures are right. The first HVAC deployment to complete that will appear here. In the meantime the honest things we can offer are the engineering record on our work page, and a free analysis of your own call data, which is evidence you produce rather than evidence we assert.
The measures
What a study will report
Every one of these is taken from a named system over stated dates, with the same measure taken before launch so there is something to compare it to.
- Calls received
- The denominator, and the one most often quietly changed between the two periods. Taken from the phone system, not from anyone's memory.
- Calls that reached a person or an agent
- As against ringing out, hitting voicemail, or finding the line engaged. This is the number the whole product is bought to move.
- Calls outside opening hours, separately
- Reported on its own rather than folded into the total. It is a small share of volume and a large share of value, so averaging it away hides the effect.
- Time to answer
- Median rather than mean, because one twenty-minute outlier drags a mean somewhere useless.
- Calls triaged as urgent, and what happened next
- How many were escalated, how fast, and to whom. A system that escalates everything is not triaging, and this is the number that shows it.
- Handovers to a person
- Counted as a feature, not a failure. A rising handover rate on a particular topic is how the next round of tuning gets targeted.
- Appointment requests captured
- Distinct from appointments booked, because at the entry tier the agent captures and your office books.
- Appointments booked
- And, where the integration allows it, how many needed a dispatcher to change them afterwards. A booking that gets moved is not a booking.
- Booking rate from answered calls
- The ratio that shows whether answering more calls actually produced more work, or only produced more conversations.
The method
Five rules the first study is already bound by
- A baseline is recorded before anything is switched on
- From the customer's own call records, over a stated period with stated dates. If there is no baseline there is no case study, because a figure with nothing to compare it to is decoration.
- The measurement period matches the baseline in length and season
- This is the one that matters most in this trade. Comparing January with April produces whatever number you would like it to and means nothing at all, so either the periods are the same season or the comparison is year on year against the same months.
- The customer sees the figures before anyone else does
- Nothing is published that the customer has not read and agreed is accurate. They are the ones who have to live with it being on the internet.
- The method is published alongside the result
- Which system each number came from, over which dates, and what was excluded. A result you cannot argue with is a result nobody should believe.
- Limitations are stated, not omitted
- What the numbers do not show, what else changed during the period, and where a figure is the customer's own attribution rather than something we measured.
Whatever the numbers say
Four claims that will not appear on this page
- Revenue recovered, stated as fact. A booked job is not proof the call would otherwise have been lost. Where revenue appears it is the customer's own attribution, labelled, with their method shown.
- A comparison between different seasons. In this trade that produces whatever figure is wanted and means nothing.
- A percentage without the two raw numbers behind it. A rise from two calls to three is a fifty per cent improvement and is not worth printing.
- A quote we drafted for a customer to approve. If they did not write it, it is not a quote.
Frequently asked questions
Frequently asked questions
- Why does this page have no case studies on it?
- Because no HVAC deployment has completed a measurement period yet, and the alternatives were worse. We could show the software portfolio and let it be mistaken for receptionist results, or publish percentages nobody can check, which is what a great deal of this market does. Publishing the measurement commitment instead means that when the first study appears you can see it was held to a standard set before there was anything to prove.
- What proof do you have that any of this works, then?
- For the receptionist specifically, none of the kind that belongs on a case study page, and you should weigh that. What we can show is the engineering record, which is real software running in production, and fifteen years of the trade experience the triage rules are built from. What we would rather you did is call the demo when it exists, or send a week of call data for a free analysis, because both are evidence you generate yourself rather than evidence we assert.
- Would you name my company in a case study?
- Only with written permission, and never as a condition of anything. Where a customer would rather not be named, the study describes the business instead: the trade, the number of trucks, the market. A study is publishable without a logo on it, and a discount offered in exchange for a testimonial is how testimonials stop meaning anything.
- Will you publish revenue figures?
- Only as the customer's own attribution, labelled as such, with their method stated. The honest position is that a booked job is not proof the call would otherwise have been lost: some callers ring back, some book anyway the next morning, and some were never going to buy. Anyone quoting you a recovered-revenue figure with confidence is either estimating or guessing, and it is worth asking which.
- What happens if a deployment does not work well?
- We would rather learn that from a measurement than from a renewal conversation, which is most of why the baseline exists. A result that shows a small effect is a useful thing to know and, published with its method, a more credible page than another set of large numbers. Whether a given study gets published is the customer's decision, not ours.