What the evidence says about the evening #
InStep AI called 3,047 unique UK property business phone numbers, taken from 3,169 branch records, and Property Industry Eye reported the result on 5 August 2026.
94.1%
Of calls between 6pm and 9pm went unanswered
InStep AI, reported 5 August 2026
93.8%
Reached nobody, across the whole test
InStep AI, reported 5 August 2026
4.1%
Were answered by a person
InStep AI, reported 5 August 2026
96%
Missed by mid-sized agencies, the most of any group
InStep AI, reported 5 August 2026
The rest of the test split small: 1.1 per cent reached an automated message, 0.7 per cent reached a voice AI system and 0.3 per cent were not in service. Independents missed 93.6 per cent and the largest networks 92.2 per cent.
The demand is real. The same piece reported that 27 per cent of leads arrived outside office hours, including 31 per cent of viewing requests and 31 per cent of applicant registrations. InStep AI co-founder Reid Cockrell described it as "a structural gap between when the industry is open and when its customers are free to call".
On the management side, Lettie's data for 1 June to 31 August 2026, reported on 15 September 2026, found tenants logged half of all maintenance cases outside Monday to Friday office hours, 17.4 per cent of them at weekends, and 73.1 per cent of heating and hot water cases out of hours.
Two streams, wrongly treated as one #
The reason out of hours feels unsolvable is that agencies treat it as a single problem when it is two.
Stream one is the emergency. No heating in February, a leak coming through a ceiling, a broken lock on a front door. This needs a person with authority to spend money and the judgement to know what is actually urgent. Lettie's figure of 57.5 per cent of High or Emergency cases arriving outside weekday office hours shows the volume is genuine, and it is the part of out of hours that nobody should try to automate.
Stream two is ordinary demand, arriving late. A viewing request. A question about whether a property is still available. A tenant reporting a dripping tap. A landlord asking about a statement. None of this is urgent. All of it is time sensitive in a competitive sense, and all of it is currently handled by being ignored until morning.
What the morning costs #
An unanswered evening message does not disappear, it converts. It becomes a longer conversation because the person is now annoyed. It becomes a phone call rather than a reply, because they have followed up. It becomes a lost applicant, because they booked two viewings with agencies that answered. And it lands in the same inbox as everything else, so it has to be sorted before it can be worked.
There is a complaints tail as well.
Complaints frequently start life as an out of hours message that nobody answered for two days.
Which parts run on rules #
Rule driven
- Acknowledging that a message arrived
- Capturing who it is, which property, and what they want
- Telling them when they will hear back
- Answering the handful of factual questions where the answer is a lookup
- Putting the rest in the right queue for the morning, sorted
Not rule driven
- Deciding whether a report is an emergency
- Authorising spend
- Sending someone to a property at night
- Any conversation with a distressed tenant
None of that should be handed to a rule.
The useful distinction is between answering and triaging. Triage, done consistently and recorded, is rule driven work. Answering, in the sense of deciding what happens next, is not.
What to measure #
For one month, log every contact that arrives between 6pm and 8am with three fields: what it was, whether it turned out to be urgent, and how long it took to resolve once the office opened. Most agencies who do this find the genuinely urgent share is small, the volume is larger than they thought, and the cost is concentrated in the morning rather than the evening. That result points at triage, not at a rota.
The daytime version of the same problem is in portal enquiries and viewings, and the repair side in maintenance reporting. The free tools will convert the log into hours.