Anybody familiar with fractals will know that the answer to the question is "infinite". Fortunately most teams in service and knowledge work organisations don't need to calculate the length of your country's coastline. Unfortunately, many people seem to equate estimating the length of time it takes to answer the emails in an inbox as being a problem of the same order of magnitude, which isn't the case.
Having worked in the service sector now for almost 45 years, by far the most common problem I am asked to solve is “how long is a piece of string”, metaphorically speaking of course.
Anybody who has every asked a service or knowledge work team to plan their work or estimate how long their list of tasks will take, will inevitably be posed this challenge.
What’s more, it usually comes from a wise, elder owl in the team who has gained many years’ experience proving that creating a realistic scenario where, using the number of emails (easily found) as the main indicator of workload, could provide a reasonable expectation of the effort to answer them.
They are usually very happy to explain, in a kindly way that, if you have as much experience as they do, you’d understand why that's not possible.
They then delve into their extended repertoire of “experience” that goes something like this:
“I once worked on a case that took three years to close”………..insurance underwriter
“An application could have any number of applicants in it”…….loan approval officer.
“How can you tell what’s in the email before you read it? ……almost everybody.
“A phone call could last forever”……….almost everyone except a contact center agent ( they have to hang up after 2 minutes 🙂 )
“it depends on who gives the estimate”…….software developer.
The problem of course is that all of these answers have an attractive “common sense” aura of mystery about them.
Let me state at the outset that, in 45 years of addressing these problem, I’ve never found a team whose suite of tasks cannot be estimated and planned within the level of accuracy required for planning workload and estimating the required capacity.
My favorite approach to the problem takes the following steps:
ME: How long did it take to get into work this morning?
THEM: about 30 minutes
ME: You start at 9am, so you left home at 8.30?
THEM: No, I left at 8.15 in case traffic is heavy
ME: What’s the best time it takes you to get here?
THEM: If traffic is light, maybe 20 mins
ME: Great so we can plan that the journey will always take between 20 minutes and 45 minutes.
THEM: (smiling) Unless there’s a crash on the motorway.
ME: So how long would it take then?
THEM: (smiling even broader) It could be 1 hour or it could be 4 hours
ME: So why don’t you leave at 5 am every day?
THEM: (looking at me like I’m crazy), Then I’d be sitting in the office 4 hours early every day because at 5am it would only take me 15 minutes to get here if there’s no crash!
ME: So how often is there a crash?
THEM: About once a month
ME: So what happens when there is a crash and you leave at 8.15
THEM: I’m late !!!
ME: So you're fired?
THEM: No, you can’t do anything if there’s a crash, everybody here knows that!
ME: OK, so back to our problem, how long does it take you to answer the email query inbox?
THEM: No idea!
How many are in it now?
THEM: Let me see.........1,752
ME: How low does it get?
THEM: Never seen it below 1,000
ME: What was the largest number you've ever seen ?
THEM: I've seen it as high as 2,200
ME Why was that?
THEM: Marketing sent out a letter with a wrong date on it.
ME: Does that Happen often?
THEM: If its not Marketing its Sales
ME: But that doesn't happen any more
THEM: All the time, its why our numbers are all over the place.
ME: Has it ever been empty?
THEM: Not in the last 15 years, which is how long I am here.
ME: Well let's leaves the Sales & Marketing problems alone for a moment
ME: If we took the first 100 emails in your inbox and I offered you $1,000 to accurately estimate (+/1 10%), how many minutes it will take you to clear them, could you do it?
ME: Well, lets say I have the $1,000 here now, what time would you guess?
THEM: About 500 minutes
ME: So, 5 mins per email?
THEM: No, they mostly take about 2-3 minutes each but I need to allow for some that take longer
ME: So why 500 minutes?
THEM: I doubled what I thought it takes
ME: What would happen if you did that travelling to work and left at 7.30 each morning. Would that help when there’s a motorway crash?
THEM: ……….emmmh, no, I’d just be here way early every morning there's not a crash.
ME: An on the motorway crash days, would it help?
THEM: No, I’d still be late.
ME: Great, remind me to explain later why averages are a mostly useless measure of anything.
ME: So If I ask one of your work colleagues, do you think would they give me the same answer?
ME: So if we make it a competition and I ask you both to do 100 emails each and the most accurate estimate gets the $1,000, would that be OK?
THEM: Then we would just take exactly the estimated time we give to do them (smiling)
ME: OK, so lets leave the estimate and just make it a competition, I'll pick the first 200 emails in the inbox (100 each) and whoever finishes first (no shortcuts!) wins the $1,000.
THEM: That wouldn’t be fair as it would depend on the type of emails we both get.
ME: So, there are types of emails?
THEM: Sure, long one and short ones.
ME: So how long are the long ones?
THEM: About 20 minutes
ME: Why are they longer?
THEM: Agents always have 4 or 5 queries in an email, they stack them and send them together.
ME: And how long are the short ones?
THEM: 2-3 minutes
ME: Why do agents sometime just have one query?
THEM: They don't, the short ones are from the call center teams, they send them across when they finish a call.
ME: So there are agent emails and contact center emails in the inbox?
ME: Any other types in there?
ME: So if we give the agents and call center teams a different email address each , then the agents email inbox would only have emails that are around 20 minutes long, on average and the contact centers inbox would have emails in it that are about 2-3 minutes (2.5) on average?
THEM: I thought you said averages are a mostly useless measure of anything?
ME: Glad you were listening, I said mostly (me smiling now)
ME: Averages are mostly useless when they contain large variances like 2, 3 and 20 minutes. It gives a result of (20 +2.5)/2 or 11.25 minutes. This sounds really accurate ( has 2 decimal places so it must be !!)
THEM: a bit like me leaving at 7.30 am, useless
ME: (Smiling broadly), exactly.
THEM: So splitting the emails into two inboxes means each has either 20 minute emails or 2-3 minute emails.
ME: Mostly, they might have a motorway crash.
THEM: What do you mean
ME: Well, they should mostly have 2-3 minute emails and 20 minute emails but there might be an outlier.
THEM: What’s an outlier?
ME: Well, let’s say that, of the 100 emails in the agents (20 minutes) inbox, some might be only 2-3 minute queries?
THEM: That sometimes happens, so how do we manage for that?
ME: Of the 100, what would your experience tell you would be the number of shorter emails?
THEM: less than 5.
ME: And would there be any that might be longer?
THEM: Again, less than 5
ME: And how long might they take.
THEM: about 30 minutes, they might have a few extra queries.
ME: Have you ever heard of weighted averages?
THEM: Yes but not sure how they work.
ME: So let’s apply them here. If , as you say, max 5% are shorter and 5% are longer, then 90% are 20 minutes?
THEM: Yes, I follow that.
ME: So let’s use your estimates for the longer ones (30 minutes) and 2.5 minutes for the shorter ones, OK?
THEM: Ok, sounds good
ME: So that means 5 will take 2.5 minutes, 90 will take 20 minutes and 5 will take 30 mins, have a guess at what you think the average of those would be, would it be closer to the 30, 20 or 2.5 minutes?
THEM: emmm, closer to the 20 mins?
ME: Why do you say that?
THEM: there’s more of them?
ME: That's a good part of the answer, so a rough estimate might still be 20 minutes?
THEM: sure, I could go with that.
ME: if we wanted something more accurate, would you say higher or lower that 20?
THEM : Because of the 30 minute email?
ME: Let’s do the numbers (5 x 2.5= 12.5) + (5 x 30 = 150) + )90 x 2 = 1800) / 100 = 19.625
THEM: so its smaller!
ME: Yes, our judgement is often biased by the bigger number, it's a bit like the motorway crash!
ME: But as you can see, given we are just trying to plan how long it will take for a person or team to clear the inbox, it doesn’t really matter whether it’s 19.625 or 20, both are close enough for estimating purposes.
THEM: So what happens if we get the estimates wrong.
ME: We Learn !!!! If we make an estimate and track the actual results then that’s our biggest lesson!
After a while, we’ll get better and our “reasonable expectations” will improve. More importantly, by estimating, capturing the actual times and comparing the difference (there is always a difference unless you have a crystal ball or are rigging the numbers), we’ll learn why things are not as they seem and why email times vary.We'll see that “normal variation” might be 10-15% just because “that’s life” while “abnormal variation” usually has a “root cause”
THEM: What/s a root cause.
ME: That’s for another day!
THEM: Great, looking forward to it and I'll set up the separate email addresses and let you know what happen.
What do you think will happen, and how long will it take "them" to clear the email backlogs from both inboxes if the split between "contact center" emails "agent emails" is 50-50?
If you want to easily do this for all your teams tasks then take us up on our offer of a free demonstration of ServiceForce, our Lean Management System for Services by clicking on the this link: ServiceForce from Expertivity