Ask an agency owner where their unbilled time goes and you will usually hear about a project that ran over. Ask their team the same question and you get a different answer: email.
They mean the correspondence around the work rather than the work itself: reading a client's request properly, chasing the missing detail, the reply that turns into three replies, the "quick question" that takes forty minutes because answering it honestly meant opening the file. Almost none of it is logged, almost all of it is chargeable, and at month end it is invisible.
Why the hours go missing
The cause is friction rather than carelessness, and it compounds:
- The time tracker is somewhere else. Logging fifteen minutes of email means leaving the mailbox, opening another application, and reconstructing which client and project it belonged to. The tax on recording the work is a meaningful fraction of the work.
- The amount feels too small to bother with. Fifteen minutes does not feel worth the trip. Fifteen minutes six times a day, twenty days a month, is thirty hours.
- Nobody is sure it is billable. If the retainer's current position is not visible, the safe assumption is that the client would object, so the entry never gets made.
- The request never reached a queue at all. It arrived in one person's mailbox, was handled there, and left no trace anybody else could see.
If you want the figure for your own team, the unbilled client email calculator works it out from your headcount, rates and email load.
Personal mailboxes are the underlying problem
Most agencies start with client requests going straight to whoever owns the account. It works until it does not, and the failure modes are all the same shape:
- A request sits for four days because the account manager was on leave and nobody knew it existed.
- Two people answer the same client differently, because neither could see the other's reply.
- An account changes hands and the entire request history goes with the person who left.
- Nobody can answer "what is this client waiting on" without asking three people.
A shared inbox fixes all four, provided it keeps individual ownership. Shared visibility without an assignee produces the opposite failure: everyone can see the request and nobody is responsible for it.
What a working setup looks like
The mechanics matter less than the properties. A client request system that works for an agency has four:
1. One queue, with an owner per request
Everything a client sends lands in a place the whole team can see, and each item has somebody's name on it. Reassignment is a click, not a forwarded email.
2. Requests attached to the client, automatically
If filing a request under the right client is manual, it will be skipped under pressure and the data will be wrong exactly when you need it. Matching senders to your existing client records, by address and by email domain, means the filing is already done by the time anybody reads the message.
3. Time logged where the request is
This is the one that changes the numbers. If starting a timer or recording a duration happens in the same window as the reply, prefilled with the client and project the request is already linked to, the friction disappears and the entries start appearing. Every one of those entries is an hour that used to be given away. It is the argument for running the queue inside the tool that already tracks the time, which is what Hour Cap's shared customer support inbox is.
4. The budget visible at the point of decision
The person deciding how much effort a request deserves should be able to see how much of the retainer is left while they decide. That is a different thing from a monthly report. A report tells you what happened. A meter next to the reply box changes what happens.
What to do with the numbers once you have them
Agencies who start measuring correspondence time usually go through the same three stages.
First, mild shock: the volume is higher than anyone guessed, and it is concentrated in two or three accounts.
Second, a scoping conversation. Not a confrontation, because the data is not an accusation. "Your retainer includes twenty hours and we are averaging twenty-six, of which about seven is correspondence" is a sentence that leads somewhere useful, and it is only available to an agency that recorded the seven.
Third, a change in how retainers are sold. Once the support component is a known quantity it gets priced into the retainer, and the account stops being quietly subsidised by the team's evenings.
The measure of success
Faster replies are the wrong measure. Agencies rarely have a response-time problem, and importing help-desk metrics into an agency imports the wrong goal.
The measure is whether the hours around the work now appear on invoices. If your realisation rate moves and your retainers stop mysteriously running over, the system is working. If the queue is organised and the write-downs match last year's, it is not working yet.