Measure and Analyze: Find the ideal KPI

"- I did not know you frighten the sheep ...
- Only when they come without a shepherd. "Alatriste

This weekend, while the day was watching football, I found myself wondering "how it plays a coach to decide which players make the team?".

Normally the coach is to first throw if the results do not accompany the team. Is that true in the case of web analyst? Please choose the best KPIs ... send us to the dark ninguneo?

So ... how to decide the winning KPIs? We do not have a global ranking, some KPIs do not mix with others and find the perfect match is very difficult. Yes, but not impossible, so you have to roll up our sleeves and get down to work. In search and seizure of the KPIs and successful warriors.

Where to begin?

First step, the data. We'll take all the data that seem important, if there are none that we highlight ... nothing happens! All are removed, there will be time to decide which we serve and which not.

We take the data from the web tool we have, also collect data from other sources, the DB of our customers, looking through the logs that leaves the system, from anywhere, we have to collect.

If you get it all is because they know your web to the smallest detail, nothing happens, now you know basically. Purpose of this first step: collecting and capturing data.

Second step, sit and think. We need to know what we want, or what we will change and we will transform the business does not go. What we aim to achieve? Not always going to be a sale or sign up for a service, may be more intangible, such as "Who is interested in this product?" or "how is running our customer service?".

Third step, we have the data, we have targets. What run? No, not at all. Now we must decide which data or data combination will take me to decide the perfect KPI to measure this objective and can implement business decisions based on their value.

Here I take a break. It's difficult finding the right KPIs. At first glance, may seem important and reflects perfectly the target. Is it the perfect KPI? Or are we kidding? Here we apply what I have dubbed the method Perero. This take 2 basic questions:

- If the data goes up or down ... is it significant?

If the answer is yes, perfect. We have a KPI whose variability is important, just what we were looking. We continue with the questionnaire.

- If it is significant, do I have the resources to act?

If the answer is ... well, that KPI will be an impressive but not going to contribute anything, I can act, not to me.

That is, not be more than we need. It is well to many interesting data obviously, but if we do not add anything to decide how we can improve, not interest.

I start from the teachings of Avinash Kaushik. To define our KPIs must take into account the difference between the KPIs and Kias (most of it more):

The KPIs (Key Performance Indicator) is equivalent to the basic way of measuring indicators, basically refer to combinations of page views (individual, aggregate, average), visitors (visits, unique, total), time (total, average), referrers ( counters, keywords, trends) ...

The KIAS (Key Insight Analysis), however, involve a change in philosophy when measuring indicators. The highlights include: click density analysis, visitor primary purpose, task completion rates, views segmented, multichannel analysis ...

Do not despair if you do not get the KPIs / KIAS achieved on the first try. You can go testing with several KPIs / KIAS to find that, through experience, to contribute more to the target. Even happened to me, you can find combinations that define a goal far better than you've been given, which will give you an unprecedented satisfaction.

Using a wheel like this, we will put on paper what is really important for each KPI and how it affects business. I find it very interesting as a working methodology for selection and maintenance of KPIs:

What question has made us reach this KPI? Who is important information you can offer? Why is it important? How did we get the data? What questions will arise with the values you can take? What actions should we take if changes for the better? "And worse?

Fourth step. Create the context. Never give the KPI / KIA raw, as if to let go the wrong way in a report. No. He should be evolution, or compared to total sales, or total number of discharges of another service, or what needs to increase to reach a goal ...

It depends on the KPI / KIA, but must always be enclosed within a context. This is where the data segmented by source, number of visits per customer / client not for profit ... If possible, accompany the quantitative data of some complementary qualitative data (through a survey, for example).

We can not have, neither we nor anyone else to receive our report, do any of these questions:

What if in January we sold 25 units of a product? Is it too much? Is it little? Does it meet expectations? How many people are interested but not bought? How many people came to the web by this product? How much is left to meet the target? How many users will look for the internal search engine? What about Google? How many users have become customers with this product?

Fifth and last step. We have all KPIs / Kias, variations, settings, everything perfectly spinning. So what? Well, just missing the icing on the cake, giving you value as an analyst. And no, it is deciding how to present data, that provides for various posts. Lack valuation analyst.

Missing the decisions that are linked to the data. This is what we should indicate the KPI Wheel, on the actions to take with the changes. We must make the interpretation of how to use this data intelligently. How to improve the picture, even a small percentage. Although everything is going as silk, you must propose actions, steps, whatever. This is where we can apply the Karnaugh table. What information is important? Where can I focus? What segment going to improve?. This will be our starting point to determine whether the KPI is valid or not worth it.

We started as coaches, no credibility (except Guardiola). Well, nothing, we play, to give everything, to ignore the bad reviews, step by step, game by game, goal to goal. And if the way we get the league or the Champions Cup ... it's great.

Note: If you want to know more about my work or my interests you can read my blog, "Where is Avinash when you need him?, Or my twitter, @ surprised.

Series 'Measure, analyze and run' by Gemma Muñoz etc:
- Beginning to measure a brand on the internet
- Measuring Social Networking: Twitter
- Web Analytics to optimize content