Proactive Analytics with Alerts

If you’re like me, then you want to keep on top of what’s going on. You want to identify problems and opportunities early, and you want to do so analytically.

An analytical solution gives you a wealth of reports and dashboards. You could start your day by reviewing this information. Alas, after a few days you discover that you don’t really have the time to review all those beautiful reports on a daily basis. You need something more proactive. You need alerts!

iphone_alert

Alerts are not a new idea, but they are as important as ever. News alerts can tell you when your company is in the news. Stock alerts can tell you if your favorite stock is on the move. The same idea is true for your business analytics.

Turning Reports into Alerts

Consider these simple steps for converting any report, pivot, or chart into an alerting tool.

  • Define a threshold and only display information that crosses that threshold.
  • Schedule the report to text or email you the output only if it has any data.

If you pick the right threshold, you’ll only get a text or email when your attention is needed. For example if there’s a significant change from the previous day, if a commitment was breached, or a key indicator crossed a threshold.

Let’s look at a few examples:

  • You’re about to breach your service-level agreement
  • There’s been a drop in performance with significantly longer wait times
  • Your team’s backlog went above your target limit

Implementation

Using Explore Analytics, you can define an output filter. An output filter looks at the output of the report with its aggregated and calculated values and selects only the items that meet the filter selection. For example, if the report summarizes the backlog for each of your team members, the output filter can select members with a backlog of over 50 support tickets. On most days, this report will be empty. When the report is not empty, you’d like to be alerted.

backlog_by_technician

The second step in our implementation using Explore Analytics, is the alerting. We do that by scheduling this report and requesting that it be sent only if it’s not empty.

alert

A couple things to note in the dialog:

  • The email address @vtext.com allows you to send an alert as a text message (SMS) to your phone. The address depends on your mobile carries (the example is using Verizon).
  • The checkbox at the bottom of the dialog means that you will only get the text message if there’s at least one team member with a backlog over 50.

Picking the Alert Threshold

What you want is a text or an email saying “hey, something is going on, you need to take a look.” The message should include the information so that you can immediately act on it.

Therefore, when picking the threshold, consider:

  • Is it actionable? If you’re not going to act on it, then don’t send it.
  • With alerts, you’re looking for exceptions. If you’re going to get the alert every day, then it’s not an exception and you’re likely to say “we should really do something about it” and promptly ignore it.
  • Don’t overwhelm yourself or your team members with too many alerts. Pick the really important ones or set the threshold higher. Let your team members define their own alerts.

Conclusion

Analytical alerts allow you to focus on exceptions and opportunities and proactively manage them. Information is actionable and you can immediately take action or delegate to the right person to take action.

Tracking Trend for Self Service Trend Analysis

Trend analysis is one of the most useful tools for understanding the current state of things and their outlook. Is the business improving or deteriorating? Once visualized, trends are easy to spot. A single glance can tell us whether things are converging, diverging, or moving in a certain direction.

The Problem: Ready Historical Data is Out of Reach

For users, creating a trend chart is easy provided that historical data is readily available. Alas, this is often not the case. Let’s consider an example. An application keeps track of projects and all their tasks. Progress is recorded at the task level and information is rolled up to the project level. So far, this is all very typical. We can calculate the “% complete” of each project simply by adding up the hours completed and dividing by the total hours planned. Yet that only gives us the current status.

For trend analysis, we need to know the “% complete” for yesterday, and the day before, and last week going back in time. We may want to see this information broken down by task and grouped by team or subproject. To do so, the system must keep a history of the updates to each task and be able to tell us the number of hours completed on the task at any point in time.

Operational systems, as opposed to specially-designed data warehouses, typically fall short of this requirement in two ways:

  1. The system might simply update the hours on the task without keeping a history. In that case, our user is out of luck.
  2. The system may separately keep timecards. It may be possible to reconstruct the history from the timecards, but that would be outside the reach of our self-service user, if we assume they don’t have SQL and programming skills.

The Solution: Tracking Trend

“Tracking Trend” is a name for a simple feature that makes life easy for self-service users and allows them to trend data. It breaks the problem into three easy steps.

Step One

The first step is for the user to create a report showing current information. The report would show the “% complete” by project, subproject and team as of the time of running the report.

Step Two

The next step is to “Track Trend” on the report that was created in step one. To “Track Trend” means to create a job that runs the report on a schedule specified by the user, for example every day at 10pm, and capture the output into a table.

Step Three

The table created in step two is perfect for creating trend reports. The table has a date/time field indicating each time the report ran. A trend report can show trends at the project level and allow drill-down to the subproject and team level. Creating such a report is well within reach of a self-service user.

Other Example Applications for Tracking Trends

Tracking trend is useful for any calculated metric, such as inventory on hand, net worth, membership renewal rate, support backlog, top-10 list, and much more. Think for example how easy creating a burn-down chart becomes.

Track Trend in Explore Analytics

Explore Analytics makes tracking trend as easy as scheduling a recurring meeting in outlook. You specify the report (created in step one), the schedule, and a name for the table to hold the output. It’s that simple.

Burn-Down Chart