When's my table coming? Again.
A furniture business answering the same delivery questions all day. The common questions were grouped and routed, and the routine ones answered from the order records.
23 requests sorted in a morning
Read the jobCalls an hour, minutes a call and how long callers will wait. How many on, hour by hour, and who hangs up.
Calls in the busiest hour, how the day goes, and who's on. The starting figures are a heating firm's office, so swap in your own.
Callers give up after a minute, five days a week, £14 an hour, and what a new caller is worth, all at our starting figures.
Most phone systems report the average wait of callers who hung up. Double it is a fair guess if that's all you have.
People who haven't bought from you yet: a quote, a first booking.
Our guess. Nobody knows this one exactly. The rest ring back, and are already in your calls.
Kept between 0 and 100,000.
So each hang-up is worth about £30 of new work: 30% × 50% × 1 in 3 × £600.
With the same 6 people at 10 o'clock, one shared line answers 93 in 100 within 20 seconds; split 3 and 3 across two lines, 83 in 100.
At 10 o'clock, 40 calls an hour is 3.3 people's worth of talking, yet the 4 on answer only 73 in 100 within 20 seconds and about 8 callers an hour hang up. It takes 5 to reach your 80 in 100 (they'd answer 85). Across the day it's 85 in 100, which is why the weekly report looks fine. Moving people from 8 till 9 and 3 till 6 to 9 till 12 meets the target every hour with 39 person hours a day instead of 40, and about 26 fewer callers a week hang up. On these sums one more person at 10 o'clock catches about 19 hang-ups a week, roughly £560 of new work, for £70.
| Hour | Calls | People's worth | On now | Answered in 20 s | Hang-ups | Needed |
|---|---|---|---|---|---|---|
| 8 till 9 | 18 | 1.5 | 4 | 95 in 100 | 0.6 | 3 |
| 9 till 10 | 34 | 2.8 | 4 | 80 in 100 | 5.0 | 5 |
| 10 till 11 | 40 | 3.3 | 4 | 73 in 100 | 7.9 | 5 |
| 11 till 12 | 36 | 3.0 | 4 | 77 in 100 | 5.9 | 5 |
| 12 till 1 | 26 | 2.2 | 4 | 88 in 100 | 2.2 | 4 |
| 1 till 2 | 24 | 2.0 | 4 | 90 in 100 | 1.7 | 4 |
| 2 till 3 | 28 | 2.3 | 4 | 86 in 100 | 2.8 | 4 |
| 3 till 4 | 20 | 1.7 | 4 | 94 in 100 | 0.9 | 3 |
| 4 till 5 | 16 | 1.3 | 4 | 97 in 100 | 0.4 | 3 |
| 5 till 6 | 12 | 1.0 | 4 | 99 in 100 | 0.1 | 3 |
| The day | 254 | 21.2 | 40 | 85 in 100 | 27.3 | 39 |
Even the rota that meets your target loses some: with callers who give up after a minute, nobody answers fast enough for all of them.
Everything runs in your browser. Rough sums, not financial advice. Each hour is worked on its own, with calls arriving at random and callers giving up after your average wait. Count people actually answering, not the rota. The new-work figures rest on your guesses about who rings someone else and who'd have bought.
Downloads propellr-phones-reminder.ics. Opens in Outlook, Google or Apple calendars. No sign-up, no email.
Calls don't arrive in a neat line. Forty an hour is one every ninety seconds on average, but some minutes three ring at once and some minutes none do. Each caller waits for the next free person, and gives up after about a minute on average.
For each hour we work out, for every number of people, the share of callers answered within your time and the share who give up first. These are the sums call centres use (Erlang's, with callers who hang up counted in). The people needed is the least number that reaches your target.
The textbook version, where nobody ever hangs up, says four people would answer only 37 in 100 within 20 seconds at 10 o'clock. The real figure is better, 73, because callers who give up shorten the queue for everyone behind them. That's not a saving. It's the 1 in 5 who left.
People's worth of talking = calls an hour × (minutes on a call + minutes of notes) ÷ 60
Answered in time = callers who find someone free + callers who wait, but are answered within your time and before they give up
People needed = the fewest who answer your share in time
Waiting stays small up to a point, then climbs steeply, as it does for jobs in Why does every job take three weeks? On the phones the point comes sooner, because callers won't wait days: at 10 o'clock four people are only two thirds busy and one caller in five still gives up.
Two lines need more people than one because each keeps its own spare person for its own rush. One shared line keeps one spare for both.
| Calls an hour | 3 minutes | 5 minutes | 10 minutes |
|---|---|---|---|
| 5 | 1 (0.3) | 2 (0.4) | 2 (0.8) |
| 10 | 2 (0.5) | 2 (0.8) | 3 (1.7) |
| 20 | 3 (1.0) | 3 (1.7) | 5 (3.3) |
| 40 | 4 (2.0) | 5 (3.3) | 8 (6.7) |
| 80 | 6 (4.0) | 8 (6.7) | 15 (13.3) |
| 160 | 10 (8.0) | 15 (13.3) | 27 (26.7) |
It leaves out calls that spill from one hour into the next, breaks, call-backs you make, and email and chat. Count people actually answering, not the rota.
Or take fewer calls in the first place
A furniture business answering the same delivery questions all day. The common questions were grouped and routed, and the routine ones answered from the order records.
23 requests sorted in a morning
Read the jobCovers and staff hours, day by day. Which days carry more hours than they need, and what that adds up to in a year.
Have a goEnquiries that wait more than a day tend to go elsewhere. Roughly what slow replies cost in a year.
Have a goRough sums only get you so far. Bring the real ones to a first call.
Book a callAsk about this result