Welcome back, guys. I'm Katherine McKeever, and today we're actually going to do a how to design a data collection system.
So in our last module, we talked about some of them. Let's layer on some things about data collection systems, and we talked about leading and lagging measures. And what? How do we build a system where we're going to capture what we need without over engineering and
and this module were actually going to go through an example of creating a data collection system.
The first thing up, let's do a quick checklist for you guys. So things that I want you to be able to answer before you walk into Let's do some data collection. The 1st 1 is what type of data are you collecting?
Hopefully, you guys have been paying attention throughout the course, and your answer is continuous. If it's discreet, that's cool to continuous or discrete. What is the operational definition of the data we're collecting? So what exactly does this this definite? What exactly is this measurement?
What does it mean? So if we're gonna use a cycle time,
which seems ominously like a prediction for the future, we would talk about from when we received the phone call to say when we knocked on the door. That's gonna be your operational definition. It can also be a scope. So keep that in mind. What I just explained to you phone call the door is a scope statement.
But it's how we know the everybody is on the same page about what we're measuring.
Are we measuring input processes, output measures? This is going to be important. Um, not only because we're talking about the different types of measures, but what do we do with that data once we get it, If we're measuring input at it, we're going to go back to the people who gave it to us because their outputs become our inputs.
But if we're measuring output data, we're gonna go to our customers and be like, Hey, how's that working? Um,
remember, for Lynne and Six Sigma, we tended to live in the process measure perspective, but it doesn't ever hurt to understand where you are in the what is a process. Um, as far as what you're measuring, it gives you a sense of that bigger picture. The last thing that you really should be able to answer, but must
answer is what's your ideal target? So when we're talking about this, remember you now know that you're working with continuous or discrete data.
We're talking about input, process or output measures. Your ideal target is larger. Better? Is that something where we want higher numbers is smaller. Better?
Is this something where we want smaller numbers? There is also nominal is best. Is zero better? Is this we don't want larger. We don't want smaller. We want nothing. So nominal is best is gonna be something like defects. Nominal is best when we're talking about larger is better and smaller is better
You could be talking about things that will inherently have a measurement so like smaller and better is a time measurement
The smaller your time measurement, the happier you are larger, larger is better revenue number of products that we can put out the percentage of employee satisfaction which is a tertiary benefit Those sorts of things. These are things that I want you to keep in mind as you're going through your data collection
because, remember, you're going to use the same data collection technique
throughout your project. So you've picked your one measure. How you measure it for your baseline is how you're going to measure it for your pilots. Is how you're gonna measure it in six months when you're like, Yeah, let's make sure we're still seeing the benefits that we said we were.
So let's do our example.
We're going to do a room service example, but we're gonna look at it in two different ways, So we're gonna look at it from a continuous and a discreet. So when you do a data collection plan, you want to have your measure on and you wanted to be in a way, that that makes sense, remember, is easy to understand.
You're gonna have your measure type. Is this an output, or is this a
process because it impacts who your customers are Your data type B really explicit with this? Are you talking about discreet or you're talking about continuous Discreet? Are you doing orginal or are you doing categorical on?
So when you talk about operational definition for this example, it's going to be from call to knock on door because we're doing a room service example specifications. So you remember in our, um,
our process capacity modules. Were we talked about CPK where we have what is our customers? Upper and lower control limits. So for this particular specifications, we say 30 minutes.
Um, but our target is a sap. So it's going to be 30 minutes or smaller. Is better or nominal is best. So zero is what we want because I want my eggs right away.
And then let's talk about our data collection type. So the 1st 1 for continuous data. We're gonna look at our employees reporting that. So we as an employee. Okay, I get the phone call at 12. Exactly.
12 midnight. Exactly. I carry the food up 17 minutes later.
Okay? These are good things were within specifications. We are know that this is continuous. So we now have that measurement employees reporting.
Now, let's look at that same example on a discrete measurements. So we same thing. Discrete data call to knock our specifications. 30 minutes. We want a sap. But in order to collect our order to report on just this particular discrete data, the best we can do
is a yes, sir. No. Was it delivered in 30 minutes or was it not?
So if you start thinking about which pieces of data are more meaningful. Start asking yourself is yes. No. Did we need the target or didn't we meet the target? Good enough? Or do you want to know where on that spectrum? Because we're continuous data. Did we fall?
So let's think about what is what did we learn from this? So in our continuous data, remember phone called adore 17 minutes We learned whether or not we met the specifications. So we know that our target is, um,
30 minutes, 17 minutes smaller than 30 minutes. Oh, yea, we did it. We learned where we are. Range of meaning. The specifications is so where we are in our continuum. So delivering and 17 minutes is different than delivering in seven minutes
and still different than delivering in 27 minutes.
So it gives us a sense of how good we are at that. Smaller is better measurement on Ben. If RM blaze can read a watch, it seems really flippant. But remember, you're designing your data collection system for the people who are actually doing the data collection.
So if you were in a scenario where your employees would have no way to tell what time they received their phone call versus what time they knocked on the door. Having them report time intervals
is not going to work when we look at discrete data.
What we learned about it from this process is whether or not we met the specification. Yes. No, this is a binary measure. So if you remember when we talked about types of data, we said that continuous data was much more robust for us because it gives us more information. This is a really great example. Why,
if we have 100% that are no, then we know that we have to do work to get our target down
to less than 30. But if we know that ah, 100% of our deliveries are 27 28 minutes, we know that we want to get it less. But we're still technically within our target. So we do me our customer satisfaction. So,
um, continuous data is best discrete data. We can make it work. Continuous data is going to give us a lot more robust information. And remember, we're going to design this
for our employees doing the collection, but anything that you can measure is a continuous. You can, in fact, measure as discreet if needs be.
So today we went over creating a data collection plan. You know that You need to know your measure. Your data type. Um, your operational definition. You need to know your measurement type input output process. You can You need to know what your specifications are. If you have upper and lower specifications limits,
you need to know what your target is.
So just because we can do it and 30 minutes do we want to do it in 15 minutes and then you have to have a way to collect that reflects that information is so in this day and age, be as creative as possible. But remember that somebody does have to do the data collection. And consistency is key for this one
and our next module, we're going to go over validating the data collection system. So we created it. How do we know it is in fact, reliable and repeatable? So I will see you guys there