Note: This is Part 1 of a 2 Part Series
Your IT organization has a problem. And you don’t even know it.
You’re collecting data and you probably think that you have everything you need. But you likely have trouble connecting that data with any action that had a meaningful and measurable impact on the results your organization delivers. That’s because what you have is data. But what you really need is information that enables action.
The primary focus of any IT Service Management or IT Transformation effort is to improve service delivery and operational efficiency in order to deliver the appropriate level of service in the most cost effective manner as possible. In order to do that, you must be able to measure your performance in a way that enables you to monitor your effectiveness and take the corrective actions necessary to move you towards your goal.
The problem is that most IT organizations collect reams of technical data, but have trouble converting that data into meaningful, results-driven action. The primary flaw is in the data itself. In most cases, how it’s collected and reported makes it almost impossible to take action.
The reason? The data lacks context.
The Need for Context
If you think about how you make decisions in your day-to-day life, you almost never use a single data point to make a decision. Something as simple as choosing a restaurant involves a large number of data points:
- Guest preferences and style
- Type of food
- Average entrée price
- Reviews from previous guests
- Attire – casual, trendy, dressy, etc.
And the list goes on and on. You may make that decision in a split second, but it’s only because you have all of those data points and you can decide what action you’d like to take. But let’s say that you were asked to choose a restaurant, but were only given one of those data elements – say price. Which would you choose? Could you? It would be nothing more than a random selection, because that data point, by itself, is meaningless. You need the context of the other data points to be able to evaluate your choices and decide the correct course of action.
But when it comes to IT operational decisions, IT organizations routinely make decisions with only a single or very limited set of data points. This is particularly true when it comes to process-based metrics. You may look at Mean Time to Restore Service (MTRS) and evaluate if it’s trending up or down. And if it’s going in the “wrong direction,” you may tell someone to fix it. But this almost inevitably leads to bad decisions or misdirection because MTRS, by itself, does not provide enough context to understand the true cause or the corrective action that is required.
Creating Context through Correlation
The reason is that the necessary context is not self-evident with most IT metrics. The mantra of “you can’t manage what you don’t measure” has been engrained in IT organizations for so long that you measure and measure and measure. You measure everything, but the measures tend to be collected and reported independently. But IT systems and the processes that support them are exceedingly complex and almost never operate in isolation. So in order for MTRS to be a useful metric, you must understand it in context. That means that you must correlate it with other key metrics that stand in relation to MTRS and which will enable you to understand why it is trending the direction that it is and to determine if it may be a temporary deviation or a real problem. Only then, can you begin to determine what corrective action is necessary.
The challenge with correlation is that it takes work. It means that simply identifying metrics is not enough. You need to build a model which connects the metrics together in a way that explains their relationship to one another. This requires a deliberate evaluation of each metric to determine what other metrics could impact it and which it might impact.
The pay-off, however, is huge. By understanding the correlation between metrics you will finally be able to transform that data into information. The correlations will enable you to:
- See the cause and effect of process elements on the desired outcomes
- Diagnose the contributing factors to undesirable trends
- Isolate areas that require greater focus for improvement
One of the greatest benefits to a correlation model, however, is often missed – its ability to self-correct and enable continual improvement.
The Self-Correcting Correlation Model
One of the most powerful elements of a correlation model is that it is self-correcting. Because you are using the correlation model to answer the “whys” when you observe an unfavorable trend, a deficiency in the model becomes readily apparent.
Let’s use MTRS as our example. You might define that MTRS should be driven by two correlated metric indicators:
- The number of Incidents in which the initial response target was breached
- The average number of Incidents per day
The logic would read that MTRS will move in correlation to these two indicators. If your teams are not responding according to the agreed upon response timelines, there’s a strong likelihood that it’s going to take them longer to restore service. And if there are simply too many incidents occurring, it is likely to stress your organization’s capacity to respond and thus increase the average restoration time.
But what if MTRS is rising, yet response targets are being met and the average number of incidents is low? In this case, the model will have demonstrated to you that there is a correlation between MTRS and something that you have not yet identified. Which means that you are not yet measuring or managing all of the things necessary to effectively manage MTRS.
In this way, the correlation model provides an incredibly powerful mechanism to ensure its constant evolution. It will continue to evolve to the point that it adequately provides your management team the information needed to effectively identify appropriate corrective actions and continually improve service.
In Part 2, we will describe how to actually build your Metrics Correlation Model.
About the Author:
John has spent more than three decades in the IT services industry, working with industry leaders like AT&T, AT&T Solutions and British Telecom. He has led numerous multimillion dollar, multiyear outsourcing and service-delivery engagements for dozens of Fortune 500 firms.
John has extensive experience helping companies evolve into Digital Enterprises by approaching the change from a behavioral and cultural perspective. Transformation is not something you do to an organization. It’s an attitude and an approach and starts with helping people understand why the “status quo” must change. And more importantly, how this change will impact them.