[MUSIC] Welcome back. In the last lesson, we calculated some KPIs, or key performance indicators. In this lesson, we will discuss thresholds and alerts associated with KPIs. Being alerted that a threshold has been crossed tells you that you need to take a look at what's going on. After this lesson, you will be able to explain how to use Tableau to set thresholds to alert you to areas of concern related to your KPIs. So let's get started. [MUSIC] In the last lesson, we calculated some KPIs, or key performance indicators. A KPI alerts us if something is of concern. It's done by setting thresholds. And these thresholds are things that you set based on criteria that were used to set up the KPI. So for example, you see from our last couple of lessons where we developed these KPIs. Here's one where it basically sets a threshold that we determine in terms of how we can, whether or not a particular area is meeting the benchmark or not. So that's exactly what a KPI does. We set the threshold based on some strategic need. And if there's an issue, we get alerted to it. So there's one type of Tableau specific terminology around thresholds and alerts, which we cannot do in this course. And that are the ones that are based around Tableau server's exciting and really cool capability of emailing specific people that will alert them when specific thresholds are breached within a KPI. It sends them an alert and then the person who gets the email can then go onto Tableau and see what the issue is. So for this lesson, I'm going to generalize the definition to one that assumes that the audience of the visual will be viewing the KPI regularly. So we will not be sending an email. But the audience member will log on and see the information on a regular basis. The key about KPIs is that it should be as real-time as possible, and I want to stress as possible. That means that there should be either a connection to data that are refreshed on a regular basis or some other way of refreshing it in a fairly straightforward way. In other words, the data should or could be from an ODBC connection of some sort. It could be hosted on a Tableau server, if you're really lucky, or it could be an Excel spreadsheet that is manually updated. And the latter is not terrible if it's not a big burden. Regardless of the approach, if it's not regularly updated, it's not really a KPI. And by regular updates, I don't mean every minute, or every day, or even every week, it means that the audience knows that when they see the information on the screen, it's as fresh as it possibly could be. So I want to go back to something I showed you in Course 2, which is a control chart. In that example, we designed a kind of a KPI, although I didn't call it that at the time because I wanted to talk about KPIs later. That type of control chart is going to likely be instantaneous. Because oftentimes, control charts are used in manufacturing environments to ensure that there is quality control over something that's produced. So a widget is produced in an assembly line. If an issue crops up, you want to know as quickly as possible whether or not there's a defect in that widget, so that it could be taken out and gotten rid of in some way. If it isn't instantaneous, it's kind of useless. Now again, the key to it being a KPI is what I discussed when defining it at the beginning of the course. It is important, it is measurable, and what it measures is some sort of performance marker, like profit, reliability, or other attribute. A KPI will also have a pre-attentive attribute of some sort that would quickly alert someone of an issue. Now again, Tableau server allows you to email, we don't have to have a server for this course. So let's go back to the control chart. As you learned how to do, we created a parameter that will allow us to see how well we're doing within a set number of standard deviations. If it's outside of that standard deviation, then it might be a problem. The threshold is the standard deviation line, and the alert is the change in color that occurs when it's outside of the line. And the other example, which I just showed you, comes from the KPI, a very simple KPI, but useful KPI that we created, with the sales super store data set earlier. This is, again, it tells us everything we need to know about sales by region and what's not hitting their benchmark. It tells you nothing else. And that's the idea of a KPI. All it does is alert you that you have passed a threshold that you need to be concerned about. Then you can dig into the data and see if it's an issue or maybe it's not. In the next couple of lessons, I'll be creating new visualizations with KPIs using alerts and thresholds, and then we'll get them into dashboards. So just head to the next lesson right now. And we're going to be taking care of that. And we'll be getting a nice cohesive story at the end of the course. See you then, bye.
