Workforce Management calculator

[RoxaneB]

Hello,

Wondering if anyone out there does a manual calculation for workforce management (i.e., determining productivity levels and calculating the number of bellybuttons - or FTEs - needed)?

We are not a true call centre - our customer service representatives are handling calls (typically from sources external to the organization), processing electronically-submitted offers of work from our funders, processing electronically-submitted updates from our funders providing updates on client care/information, emailing internal staff such as schedulers and front line staff (e.g., nurses).

For now, my manual calculator can track volume of work (in the above paragraph) along with the average time for each activity. In some cases, I am also able to show "unfinished" work - work left on the table, so to speak - which allows me to show how many staff we would need for 100% completion of all work (e.g., answering all calls, processing all updates, etc.) I've also factored a buffer that includes meetings, breaks, work not directly related to the activities above, training, etc.

I'm able to show historical productivity for each activity by site and employee, as well as overall productivity. At first glance, it looks like I need to purchase the add-in to obtain the WFM forecasting formulas. The trend line is not an adequate option.

If anyone out there is using a manual approach to this, does it sound like I'm on the correct path? Is there a way to forecast resource needs by day and time without purchasing the add-ins?

Trying to help our local leadership teams understand if they have the right people doing the right things at the right time, and this seems like a complicated, but data-driven approach to take.

Thanks!

 

[Steve Prevette]

I should point out there are some grave hazards to what you are doing, and its unintended consequences. If call center personnel are monitored solely on how many calls, how many minutes per call, they will "dump" cases that are "too hard and will take a long time" on others, or simply hang up on the customer. Response call times can be very skewed - designing the personnel manning levels to the 'average' will doom you to be overloaded half the time, and completely swamped now and then, unless you take into account the variability in the service times.

 

[RoxaneB]

I should point out there are some grave hazards to what you are doing, and its unintended consequences. If call center personnel are monitored solely on how many calls, how many minutes per call, they will "dump" cases that are "too hard and will take a long time" on others, or simply hang up on the customer. Response call times can be very skewed - designing the personnel manning levels to the 'average' will doom you to be overloaded half the time, and completely swamped now and then, unless you take into account the variability in the service times.

It sounds as if you are making the assumption that call time is a metric. It is not. It is, however, a variable in determining if we are staffed at adequate levels. Being a healthcare organization with a focus on community care, we fully recognize that for some of our clients, speaking with one of our agents may be the one interaction they have that day. Our agents are aware of this and they know that they are not monitored (nor penalized) for lengthy calls. Red flags for us are variables (note: not metrics) such as hold time, number of transfers, etc. which can adversely impact the caller experience. Agents are also evaluated for call quality behaviours (i.e., phone etiquette) including greeting, empathy, authorization checks, tone, probing, active listening, etc.

We need to start somewhere, and starting with average is a good spot, recognizing that we will never be perfect in terms of the phone service (for which, btw, we have a service level of 80% within 20 seconds).

 

[Steve Prevette]

If you truly are looking for a service level of 80% within 20 seconds - do you keep track of that information (or metric)? How far away from it are you? Is it common cause (the system is not capable) or special cause that impacts you.

That would help you assess where you are. If average is sufficient for you, well it is a minimum starting point.

There are operations research tools (Queuing Theory and Monte Carlo Simulations) that network engineers use when trying to determine if a computer network will get overloaded and how often. Chap_01.pdf (washington.edu) gives a reasonable intro. Of course, it can get very complex very fast. Queuing at Bank with 2 Tellers/Servers Monte Carlo Simulation - YouTube is one approach.

I should answer your question of " Is there a way to forecast resource needs by day and time " as Yes, and it is a common task out there in operations research. It is a combination of analyzing your past data, and modeling the variation in the data, not just the average, and determining what is your risk profile against manning levels. Rather similar also to "stock out" analysis - what is the probability that a responder will not be available on the next call, or that response will take longer than 20 seconds.

 

[RoxaneB]

If you truly are looking for a service level of 80% within 20 seconds - do you keep track of that information (or metric)? How far away from it are you? Is it common cause (the system is not capable) or special cause that impacts you.

Yes, our telecom system is able to track and display this data for all of our locations and down to the daily level. What I cannot answer is my own question regarding having this level of data down to the employee level, but I would presume yes as each agent does need to sign in to the system. With that being said, there are also cultural issues to overcome with some agents using the login credentials of other agents.

As for achieving our target, some sites are while others are not. Yes, we're looking at why some sites are doing well and also while some sites are struggling. We are also taking the approach that simply throwing more people into the headcount numbers is NOT the first (nor ideal) solution - preferring to look at process adherence, process improvements, and system automations at this early stage.

That would help you assess where you are. If average is sufficient for you, well it is a minimum starting point.

That's all I want at this point in the process. A starting point. I fully recognize that we will not have a perfect solution of the gate and that there will be future iterations of our solution. We have limited resources to work on this and limited time - so rather than spin our wheels on perfection, we're taking small steps. Starting off with building a Model T as our first iteration instead of Cadillac.

There are operations research tools (Queuing Theory and Monte Carlo Simulations) that network engineers use when trying to determine if a computer network will get overloaded and how often. Chap_01.pdf (washington.edu) gives a reasonable intro. Of course, it can get very complex very fast. Queuing at Bank with 2 Tellers/Servers Monte Carlo Simulation - YouTube is one approach.

I hate to sound like I'm shooting down your feedback, Steve, but these are simplistic and somewhat single-pathway approaches. Again, our agents are not a call centre and they currently multi-tasking. The only activity where they don't multi-task is when they are on a call. Then, they focus solely on that call. But a day-in-the-life-of essentially goes with start to process an offer, try to determine if we can accept (based on capacity and client care requirement), oh hey the phone is ringing, answer the call and try to focus on providing an exceptional client experience, disconnect, go back to the offer, accept (or reject) the offer, move on to a referral, start to build a client care schedule, oh hey the phone is ringing, answer the call and try to focus on providing an exceptional client experience, disconnect, new offer and process before it times out, go back to building that client care schedule, crap there's an update in the system for client who is being seeing later today, start to process the update, oh hey the phone is ringing...and so on.

This is why I'm starting with average times for our variables of time in terms of calls, offers, referrals, frequency updates, and emails in the high-level calculator and dashboard. And recognizing that one of the process improvements MIGHT be assigning people to specific sub-tasks within the role. Which is why I'm looking for the calculator that will allow me to enter in historical data to show productivity and to forecast out who needs to be doing what and when based on our historical performance at a daily and hourly level.

I should answer your question of " Is there a way to forecast resource needs by day and time " as Yes, and it is a common task out there in operations research. It is a combination of analyzing your past data, and modeling the variation in the data, not just the average, and determining what is your risk profile against manning levels. Rather similar also to "stock out" analysis - what is the probability that a responder will not be available on the next call, or that response will take longer than 20 seconds.

You cut off my question, Steve, in your quote. I asked " Is there a way to forecast resource needs by day and time without purchasing the add-ins? " I obviously understand the need to gather the past data, not just the average - I'm using average right now in a high level calculator for productivity (it's not spot on but it is directionally correct). The detailed data is what can be used forecast out and there is a way to do it in Excel but it seems to require the purchase of add-ins. I was looking for something that wouldn't cost anything.

Ummm..."response longer than 20 seconds" does not mean that a call lasts 20 seconds. Unsure if you understood that traditional SLA metric or am I misunderstanding your interpretation?

 

[Steve Prevette]

It does sound like you have a plan and are sticking to it. I suppose I don't understand the question "without purchasing the addins? What addins to what software? I'll just say that any qualified person should be able to write a simulation using off the shelf Excel Spreadsheet. I actually do not know what "addins" to recommend anyway. Good luck!

 

[excel-template]

Hello @RoxaneB, I have done something similar for a client 7 years ago in excel you are welcome to the file. Please contact me and I can send you the file.

Thanks,

John

 

[TerenceWood]

Hello @RoxaneB, I have done something similar for a client 7 years ago in excel you are welcome to the file. Please contact me and I can send you the file.

Thanks,

John

share your work with me plz

 

[neno74]

share your work with me plz

Hello can you share your file with me , please. Thank you in advande