7 min Read
The Calls You Miss After Five
Most service businesses answer well during office hours and lose everything else. What voice AI intake actually catches, and what it does not.
Nikhil Sharma
Key takeaways
- The enquiries that arrive outside office hours are not lower quality, they are the ones from people who were working during yours
- Voicemail is worse than no answer because it costs the caller effort and returns nothing, which is why most people simply hang up and call the next listing
- Voice AI earns its place on repetitive qualification, not on conversation, and the moment it tries to be impressive it starts being annoying
- The measurable outcome is booked work from outside business hours, not calls answered
Every service business I have looked at has the same shape in its call data, and almost none of them have looked at it.
Calls between nine and five get answered, handled and mostly converted. Calls outside that window go to voicemail. A small number leave a message. The rest call the next business on the list.
The assumption underneath the way most firms staff their phones is that the after-hours calls are lower quality. In every operation where I have actually pulled the numbers, the opposite was closer to true. The people calling at 7pm are calling then because they were working during your office hours. That is not a signal of a tyre kicker.
Voicemail is worse than nothing
This is the part worth sitting with. Voicemail asks the caller to spend effort, gives them nothing back, and offers no indication of when or whether anyone will respond. It is a worse experience than a phone that simply rings out, because at least a ringing phone lets someone move on immediately without feeling they have done something.
Most people do not leave a message. They hang up and call the next listing. Your voicemail is not a safety net, it is a slow way of losing the enquiry while feeling like you have a system.
What actually needs to happen at 7pm
Not a conversation. Four things:
- Acknowledge fast. Someone answered. The enquiry is real and has been received.
- Qualify. Is this work this business does, and is this person in a position to buy it.
- Capture. Name, contact, the specifics of what they need, in a form that means nobody has to call back for basics.
- Commit. Either book a time, or state clearly when a person will be in touch. Vagueness here undoes the other three.
That is a short, highly repetitive script with a decision tree behind it. It is exactly what a voice agent is good at, and it is nothing like the free-flowing conversation the demos tend to show off.
The temptation that ruins deployments
Once a voice agent is working, there is enormous pull toward making it more impressive. Let it answer questions about pricing. Let it explain the service. Let it handle objections.
Resist all of it. Every capability you add is a new way for it to be confidently wrong to a stranger about your business, at a time when nobody is watching. The narrow version that qualifies and books is boring, reliable, and produces revenue. The impressive version produces a story your competitor tells about you.
The best-behaved system I have deployed does four things and refuses the fifth, out loud, with a callback offer.
Where it lands is the whole build
An agent that qualifies beautifully and drops the result into an email nobody reads has moved the problem, not solved it. The output has to arrive somewhere the business already looks: the CRM, the booking calendar, the job board, the case management system.
This is consistently where these projects cost more than people expect, and it is also the only part that determines whether the thing gets used in month three. The conversation is a commodity. The wiring is the product.
What to measure
One number: work booked from enquiries that arrived outside business hours.
Not calls handled, not minutes saved, not deflection rate. Those all go up whether or not the system is helping. Booked work from hours you previously did not cover is the only figure that tells you whether the thing paid for itself.
Run it for a quarter. If that number is flat, the problem is qualification or handoff, and no amount of tuning the voice will fix either.
When not to bother
If your after-hours volume is two calls a week, this is a solution looking for a problem. Put your mobile number on the site and go home.
If enquiries arrive constantly outside your coverage, and each one is worth meaningful money, then the only question is how carefully it gets built. That part I would rather scope properly than guess at, which is what an MVP Roadmap is for.
FAQ
Quick answers to the most common questions about this topic.
More than firms expect, on one condition: it has to be obviously useful within the first sentence. A system that says what it is and immediately starts being helpful holds people. A system that opens with a scripted greeting and an unnecessary question loses them at about the same rate as voicemail.
It should take a message and say so, plainly. The failure mode to avoid is a loop where it asks the same question three different ways. Configure a hard limit on clarification attempts and hand off to a callback rather than letting it grind.
An answering service takes a message. The point of doing this properly is qualification and booking: working out whether this is work you want, collecting what you need, and putting it in the calendar or the CRM. If the output is still a message somebody has to read and action, you have paid for a more expensive answering machine.
Both are genuinely harder than vendors admit, and both are worth testing before you sign anything. Ask for a trial on your actual phone line rather than a demo. A system that performs well in a quiet studio and poorly on a construction site is a real risk if your callers are on construction sites.
The platform subscription is the cheap and predictable part. What varies is the build around it: what it is allowed to say, how it qualifies, where the booking lands and what happens when it does not understand. That is priced against your actual call flow rather than a per minute rate. I scope it in a paid MVP Roadmap engagement, and the fee is credited against the build.
How the MVP Roadmap works
Written by
Nikhil Sharma
Founder, DigiBenders
Twelve years shipping software, five of them leading a studio in New Brunswick. I build the software and run the marketing around it, which is an unusual combination and the reason most of my work arrives by referral. One person accountable, and everything ends up in your name.
You read the thinking
Now tell me what you are actually building.
If this was useful, the call usually is too. You describe the problem, I tell you what it takes and whether I am the right person for it.
Thirty minutes, no pitch
Honest read, including when the answer is no
Replies within one business day
Keep reading
