Web analysts base their statements on a measurement of operations or events. In other words it is measured
– that something is happening
– how something is done,
– in which order something happens
– how often something happens
However, “the why?” can only be partially addressed by web analytics.
Technical reasons for events can often be found by analytical methods, but the motivation of users or customers is not directly accessible for web analysts.
With the help of web analytics, you can measure page views, clicks and requests. Analysts can examine texts and classify them. We can measure the behavior of people, for example that someone clicks or that someone writes, but we cannot measure the intentions of those who click or write.
A web analyst must be aware of the fact this is his/her view of the things that happen – their own “interpretation of the world”, which might be supported by the facts, but will never be fully proven.
We must always clearly distinguish between the facts and the interpretation of these facts.
Often analysts do not realize when they leave solid ground and start interpreting and guessing intelligently.
We cannot read the minds of users!
Web analysis consists of measurement and analysis of this measurement and thus builds a solid base for good decisions for websites.
Each measurement is fraught with errors. To analyze the measurement and to take rational decisions, it is important to know the possible errors.
Gross “human” errors
- Errors of the experimenter (logic, typos, transposed digits …)
- inappropriate methods of measurement
- inappropriate evaluation and analysis
- disregard of error sources
These errors are avoidable!
- Environmental influences
- Poor ‘calibration’
These errors can not be minimized by repeating the measurement.
Statistical errors / random errors
Errors that can be minimized by repeated execution of the measurement. The error can be analyzed by error calculation.
The law of large numbers ensures that these errors are the smaller if the measurement is performed more frequently.
Over the centuries the natural sciences have acquired some methods of generating solid knowledge – in particular, the use of logic, causal relationships and the review of the generated ‘knowledge’ by measurements. Normally scientists take the following steps:
- Question nature in a intelligent way by taking measurements
- Analyze the measurement and the data
- Try to refute or confirm the hypothesis by further measurements
A hypothesis that neither can be confirmed by the measurements nor can be falsified is not considered scientifically robust. It is best characterized by a statement of faith.
Matters of faith and religion are not the subject of web analytics. In contrary web analytics is the generation of knowledge based on facts.
Thus web analytics should use verifiable and scientific methods and try to provide reproducible and logically conclusive results.
Human beings are exceptionally well prepared to deal with their environment. Our senses let us capture all the important environmental influences. Our brain is trained to make sense of our perceptions.
But sometimes our senses deceive us. Surely everybody knows some examples of optical illusions. A nice collection of optical illusions is available at: http://www.michaelbach.de/ot/index.html
The following image for example shows two squares with the letters A and B that have exactly the same color (# 787 878). However our brain tries but make sense of the colors depending on light and shadows and thus will interpret the colors and so they seem to be different.(Picture under GNU license taken from weltderwunder.de.msn.com )
You see that we can not rely solely on our perception and observation to describe the reality.
We need a measurement – with the measurement observations are quantified to avoid deception (if you prepare and interpret the measurement well).
Just as in the physical world, this statement applies to websites. The measurement of the behavior of visitors and the interpretation of this measurement is commonly referred to as web analytics.
Web analytics is used to avoid deception and make decisions based on a secured knowledge of the user’s behavior and thus contribute to the success of a website.