Web Analytics Methodology

Web Analysis: Tips and basic methodology

Web analytics comprises a large number of metrics, concepts and ratings that make it seem like a complicated discipline, even impossible for someone who does not possess extensive knowledge about it.

In this article we are going to simplify as much as possible what is the most important thing to know to be able to carry out a web analysis without giving rise to errors and having everything clear, simple and easy. In this way, decisions can be made based on the data collected and not on opinions.

Table of Contents

In web analytics order is fundamental. Many times metrics are used to “make the data speak”, but we do not realize that there is a communication problem derived not from the data that is observed, but from some previous step or implications of the metric that has not been taken into account.

A clear example is found in the bounce rate, an indicator of a very high subjectivity due to the number of factors that influence it and the differences in the results depending on how the segmentation is carried out. However, in web analytics books years ago “it was considered good” a rebound equal to or less than 50%. Today that assessment makes no sense.

“You have to avoid general considerations about metrics, the data should always be judged according to the context of what is being measured and the objectives set”

In order not to fall into this error, it is necessary to develop a well-structured measurement plan that defines exactly what is going to be measured, how it is going to be done, what value is going to be assigned to the indicators… and that’s what we’re going to look at next.

The “model” measurement plan in the world of web analytics is the one so well known to analysts “Avinash Model”.

1. Avinash Model: 5 Steps to Perfect Web Analytics

This model of measurement plan is a must for any analyst or future web analyst. We will dedicate an entire post soon because it really deserves it, but let’s see in general lines what the Avinash model consists of:

1.1.   Set business objectives

What you want to achieve, under the   SMART criterion   (specific, measurable, assignable, realistic, time-related), point it out and have it very clear because it will be the epicenter of the marketing strategy and the measurement plan.

As we talked about in the previous web analytics post, the objectives will be represented with the “conversion” metric, among which we can distinguish between macro conversions (they entail direct income) and microconversions (they contribute indirectly to it).

1.2.Identify strategies and tactics to follow

First it’s the what, then it’s the how. In this step, we decide which channels are optimal for the business and to achieve the objectives, and within the channels, which actions are those that can give better performance and results. In this section, data research, collection and analysis is fundamental. It is perhaps the most laborious step of all but it is the one that will mark the guidelines and the main lines of action on which the plan will be based.

1.3.   Choosing KPIs

What indicators are going to say whether or not the objectives are being achieved? This part is key and is where mistakes are often made for not understanding well the magnitude of the importance of this step and for choosing KPIs to indicators that are not really KPIs.

KPIs are indicators that   strictly show what you need to know about whether you are meeting goals or not   (not those that contribute to your meeting).

This task requires reflection, synthesis and mental clarity, so it is important to take your time and get to create a dashboard with the right and necessary indicators that clarify with as little information as possible whether or not the objectives are being met at a glance.

1.4.   Important: Implementation and configuration of the web analysis tool

We add this step since it is essential to be able to correctly carry out the analysis of metrics and data, and it must be done at this point of the plan, once the objectives, strategies and tactics have been defined, and having chosen the corresponding KPIs. It’s the time when questions like these arise:

What do you want to measure? What tools offer the possibilities needed for this? Is the standard configuration sufficient or do you need to create custom metrics more tailored to your needs? Are qualitative, quantitative or both aspects going to be analyzed?

 “All the effort in identifying and defining the KPIs that we are going to measure would be useless if we have not carried out the correct implementation of the web analysis tool and we have configured the system to tell us exactly what we want to know about the chosen metrics, getting to customize them if necessary”

1.5.   Choose segments

Without segmentation there is no analytics. Global data, aggregated, is the worst thing there is in analytics. They are data that tell you half-truths. The more you segment and “squeeze” that data, they will give you more concrete and accurate information to be able to identify if the strategies and tactics chosen are paying off or if there is one that had not been previously contemplated or some unknown segment that could be analyzed to give a better performance to the business.

1.6.   Assign objective values to metrics.

To know if things are being done well (remembering that it is always done according to objectives), values must be assigned to the metrics in such a way that it says:

“-If my mean value is X, a standard deviation above and below I consider it among my normal values or that indicate the proper functioning of my strategy; Now, if I go from the second standard deviation then it is that either something is being done wrong, or something is being done too well so we have to intensify efforts to analyze what is happening”

Once you have this measurement plan integrated into the business and very clear, it is time to start measuring, let the data be recorded and identify patterns and messages hidden within them segmenting, analyzing, visualizing and interpreting.

And how should the data be interpreted?

2. The intelligence behind the data in web analytics

One of the biggest difficulties faced by an analyst when preparing web analytics reports is the fact of knowing how to discriminate the relevant information, compress it, assess it and conclude the clearest, most concisely and easily the results obtained. But when it comes to decision-making or assessments/recommendations it is fundamental the statistical analysis of such data.

The    Web analysis toolsoffer   data, and you can create custom metrics adapted to the objectives, but what about how the metrics and variables behave between them? How did they covar? How do they correlate? Can you predict some kind of behavior with the data? Is there a relationship between the residues in the dataset?

All the answers to these questions will provide information with which objective opinions derived from data can be generated, which are those that allow decisions to be made with the least possible margin of error and eliminating any type of bias, risk or uncertainty.

When you’ve measured, analyzed, studied, and discovered everything the data can offer, it’s time to make assessments and recommendations.

3. From analysis to web usability

Once you have all the data that says who sees, by what channels, how it behaves and how it converts into a website, that is when you enter the field of web usability.

How to make life easier for the user so that he ends up doing something valuable to me?

Web usability is simply that,   making everything simple,functional, and as comfortable as possible for the user. We tend to go crazy with design and with theories of colors, sensations, behaviors and if we stop to think about it, really the design of the best websites in the world leaves much to be desired from the point of view of “visual graphic design”. How is the design of Youtube, Facebook, Google….?

The analysis of a website should always include ratings and recommendations to improve this aspect. Create more visible buttons, in places where mouse activity is concentrated, insert readable, simple, “usable” texts, with the right and necessary information, insert video elements, images, do not disorient the user, encourage simple and simple navigation … each website can always improve aspects related to web usability.

In conclusion, if you want to perform a perfect   web analysis,   these three requirements must be met fundamentally:

-Develop a measurement plan

-Make use of statistical data analysis

-Conclude reviews and recommend web usability actions

We hope this article has been helpful and helped you understand what a web analysis needs to have. If so, share it!