Most businesses discover their out-of-hours problem by accident. Someone mentions they rang on Saturday and gave up, or a customer says they went elsewhere because nobody picked up.
By then the enquiry is gone and there is no record of it. That is the difficult part of this problem: the cost is real, but it is invisible in your reporting.
This article covers the five options that actually exist, what each one costs, how much effort it takes to run, and the specific way each one fails. There is no single right answer, and the best choice depends on how urgent your callers are.
What do missed out-of-hours calls actually cost you?
The honest answer is that almost nobody knows, because a missed call leaves no trace in the CRM. The number that matters is not how many calls you miss. It is how many of those callers would have become customers, and what an average customer is worth.
You can work it out in about ten minutes. Pull the call log from your phone system or mobile for the last three months, count calls received outside your opening hours, and multiply by your enquiry-to-job conversion rate and your average job value.
Worked example. A drainage firm takes 40 calls a month outside opening hours. Historically it converts one in three enquiries, and its average job is worth £280. That is roughly 13 jobs and £3,700 a month sitting in the missed column, before repeat work and referrals.
Even if only a third of those callers would have gone ahead, the annual figure is well into five digits. That is the number to hold in your head while you compare the options below.
Two things make out-of-hours calls disproportionately valuable. They tend to be urgent, so intent is high. And they arrive when your competitors are also closed, so the first business to answer usually wins the job outright.
Option 1: Divert to a mobile
The simplest option. You set a call divert on your business line so that calls outside opening hours ring a mobile, usually the owner’s.
Cost: effectively nothing beyond your existing call plan.
Effort: a few minutes to configure, then permanent.
Where it fails: it fails in three predictable ways. One person can only take one call at a time, so a second caller during a call gets nothing. Nobody can answer at 2am reliably for months on end without it affecting them. And there is no record of what happened on the call unless the person who took it remembers to write it down.
Diverting is a sensible first step for a sole trader with low call volume. It stops being a system the moment you have more than one person or more than a handful of evening calls.
Option 2: Voicemail and a callback promise
Voicemail feels like cover, and for non-urgent enquiries it partly is. The caller leaves a message, you return it the next working day, and nothing is lost.
Cost: nothing.
Effort: near zero to set up, but it creates a callback queue someone has to work through each morning.
Where it fails: it fails at exactly the moment it matters. A caller with an urgent problem is not looking for a callback tomorrow, they are looking for someone who can help tonight, and the next search result is thirty seconds away.
If you rely on voicemail, at least make the greeting useful. Say when you will call back, give an email address for anything that can wait, and never leave a generic network greeting on a business line.
Option 3: An on-call rota
Rotating out-of-hours cover across your team is the traditional answer, and for genuinely emergency trades it is often unavoidable. Someone holds the phone, gets paid for holding it, and responds.
Cost: real, and usually understated. On top of any on-call allowance you pay employer National Insurance at 15% above the secondary threshold, employer pension contributions of at least 3% of qualifying earnings, and 5.6 weeks of statutory holiday that has to be covered by someone else.
Effort: ongoing management. Rotas need building, swapping and policing.
Where it fails: it fails slowly. Rotas are also constrained by working time rules, including the 48 hour average weekly limit and daily rest requirements, so a rota that looks fine on a spreadsheet may not be lawful once you count the hours properly.
The other failure mode is quieter. On-call duty is unpopular, it wears people down, and the person on call at 11pm is rarely at their best on the phone.
Option 4: A human out-of-hours answering service
An external team answers your calls under your business name, follows a script you supply, takes a message and either emails it over or transfers urgent calls to whoever is on call.
Cost: usually per call, per minute or a monthly bundle with an overage rate, and out-of-hours cover often carries a surcharge. The important detail is that your cost scales with your call volume, so a busy month costs more.
Effort: moderate to set up. You have to write the script, define what counts as urgent, and keep it updated.
Where it fails: two ways. The agent is not from your business and cannot answer anything that is not in the script, so callers with a technical question get a message taken rather than an answer. And a human team has finite capacity, so simultaneous calls queue.
Done well this is a solid option, particularly where callers are likely to be distressed or the subject matter is sensitive. Ask any provider directly how many other businesses each agent covers, and what happens when three of your calls land at once.
Option 5: An AI receptionist
An AI receptionist answers the call in a natural voice, works from a knowledge base you supply, and can book appointments, take details, answer routine questions and escalate anything it should not handle.
Cost: typically a flat monthly fee rather than a per-call charge, which means the cost per call falls as volume rises.
Effort: a setup phase where you supply your FAQs, your booking rules and your escalation criteria, then light ongoing tuning.
Where it fails: it fails on judgement. It is the wrong answer for a distressed caller, a complex complaint or anything requiring discretion, and it should be configured to pass those straight to a human. Accent handling in noisy environments is still imperfect, and any provider claiming otherwise is overselling.
The genuine differentiator is concurrency. Ten callers at 9pm on a bank holiday all get answered at the same time, which is not true of any human arrangement at any price.
Which option suits which type of business?
| Your situation | Best starting point | Why |
|---|---|---|
| Sole trader, low evening volume | Divert to mobile | Free, instant, and the volume does not justify anything else |
| Non-urgent enquiries, office hours business | Voicemail with a stated callback time | Callers will wait if the reason for calling can wait |
| Emergency trade, high job value | On-call rota plus AI or human triage | Someone has to attend, but they should not also be the switchboard |
| Steady out-of-hours volume, routine questions | AI receptionist | Fixed cost, unlimited concurrency, answers rather than messages |
| Sensitive or distressed callers | Human answering service | Judgement and tone matter more than speed |
| Peaky volume, occasional floods | AI receptionist with human escalation | The only option where ten calls at once is not a problem |
Most businesses that take this seriously end up with a hybrid: automated cover handling the routine majority, and a named human reachable for the genuine emergencies.
How do you set out-of-hours cover up this week?
Whichever option you pick, the setup work is broadly the same, and most of it is decision-making rather than technology.
- Pull three months of call data. Count calls received outside opening hours, and separate them into urgent and routine. This tells you which option you actually need.
- Define what urgent means in your business. Write it down as a list of triggers, not as a feeling. Anyone or anything answering your phone needs this list.
- Write your top twenty questions and answers. Opening hours, service area, pricing bands, lead times, parking, what to do in a leak. This document is the input for a script, an AI knowledge base or a new starter.
- Decide the escalation route. Who gets called, on what number, for which triggers, and what happens if they do not pick up. Test the fallback, because it is the part that always breaks.
- Set the greeting and the promise. Say what will happen and when. A caller told “someone will call you before 9am” behaves very differently from a caller who hears a beep.
- Log every out-of-hours contact in one place. Whatever you choose, the enquiries need to land in your CRM rather than in someone’s text messages, or you will be having this conversation again in six months.
- Review after 30 days. Count how many out-of-hours enquiries converted, and compare that against the cost of the cover. The answer is usually obvious by the end of month one.
What this looks like in practice
The businesses that fix this properly rarely do it by adding headcount. They start by measuring what they are missing, cover the routine calls automatically, and keep a human in the loop for the small number of calls that genuinely need one.
If you want a hand quantifying the first part, we will go through your call data with you and tell you what the gap is worth. There is no obligation to buy anything, and the numbers are useful either way.