http://www.technologyleaders.com/digital-analytics-2/key-performance-indicators-theyre-not-data/

Written by Andrew Edwards on . Posted in Digital Analytics

Many marketers struggle with mapping Data to Key Performance Indicators (KPIs).

In other words, they want to review the reports that come in and see whether they achieved their goals. And in most cases, the data just sits there unyielding. It doesn’t tell the marketer what the marketer needs to know. It doesn’t say “here is your KPI and here is how well you did”.

Data doesn’t know what you want to know.

But that doesn’t mean you can’t get answers to your business questions.

You’ll need to put the data in context. Creating context is one of those cognitive abilities only humans have (so far). It’s a trait not unlike intuition, and relies on the combining of lots of experiential information, plus empirical data, plus what still passes for “gut” in parlance, even though we can be pretty sure even this murky sense of “what’s right” will one day be quantified by neuroscience.

Key Performance Indicator” is a term that goes halfway towards a data definition of “why you have a web site”. Data itself has no idea, nor will it ever, why you built your site–no matter how large nor how small it is.

Think of KPI as metadata about your site. It needs to be layered on top of the reports.

Here is an example of how this mapping of KPI metadata might work. As you may begin to notice, the KPI for even a large, complex site may be rather simple, and that much of the content on the site is supportive in nature (in digital analytics, that would be called “engagement content” because it is supposed to help drive customers to perform the most important KPI, or the “conversion” event).

Let’s say your site wants to disseminate content about different topics your company has deemed important–or that your company wants to develop certain ideas based on some content popularity metrics.

Much of the work lies in reducing the ratio of signal to noise in your data. In other words, for this exercise, you will want to focus on certain reports while ignoring everything else (analytics tools are very good at giving you much more data than you need and burying what you want).

For the sake of this exercise, let’s say the concepts you are market-testing reside in a pdf or other downloadable format.

What you will want to look at is the following:

  • reports that indicate which campaigns drove the most traffic to your “pdf section” (for instance)–possibly a landing page
  • reports that indicate which campaigns drove traffic that moved beyond the landing page and got to the download page
  • reports that tell you how many times a particular asset got downloaded, and which ones were the most popular

So out of thousands of possible reports, we have isolated just a few.

And they will not necessarily be sitting in a single dashboard. You may have to go find them.

Then, in order to describe them to others as KPIs, you’ll have to extract the data from them and create a story to tell.

That story (supported by charts or visuals) might read something like this:

  • Our most popular pdf was “Comparing Icebergs to Big Data”.
  • The campaign most successful at driving traffic to this pdf was called “Don’t Get Sunk in Oceans of Data”.
  • We should devote more resources to our consulting offering about the dangers of mishandling Big Data.
  • We should come up with another campaign that calls to mind disaster on the high seas.

Of course I have oversimplified. But decoupling what you want to know from actual reports; and finding reports that answer your business questions is an exercise that, at least until robots replace nearly all of us, can only be made by a human observer exercising some fairly straightforward powers of observation and deduction.

Key Performance Indicators: They’re not Data
Tagged on:

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>