Setting up web analytics for online stores, service sites I identified for myself some problems, the importance of which entrepreneurs often underestimate in the context of the company’s development. In order to fully disclose them, I will try to explain in as simple a language as possible how web analytics should work, and what the owner should pay attention to when building a strategy and working with contractors. Separate technical details of setting up tools here will not be affected.
Why do I think I can talk about it? I have my own project – the men’s magazine Brammels. In it, I act as an analyst, as a marketer, and as an entrepreneur. And after half a year of work on it I understand how, in fact, it was still possible to improve the work from the very beginning. Looking at the situation from different roles, you can see more important details that can help my colleagues as well.
First you need to answer simple but relevant questions for some entrepreneurs.
What is web analytics?
It is a tool that helps to control and increase the effectiveness of actions to achieve business goals through Internet marketing.
In short, thanks to web analytics, you can find out how much you earned for each dollar invested, and how to earn more. In the current conditions of the lack of unique products and services – this tool becomes critical for a growing company.
Why do I need web analytics?
Yes, even this question comes from people. But I understand that for one is insight, for another it is obvious. Knowledge of our sphere seems obvious to each of us. Therefore, I will explain the answer in detail.
Analytics is the basis of thinking, logic, personality, ego, mind. It helps people to survive and live better.
Every entrepreneur is engaged in analytics more than anyone else, as it constantly operates in conditions of uncertainty and daily updating of incoming data.
Web analytics is a tool that will help you move from a fun game of luck and intuition to cold calculation and exact numbers. And most importantly – with the mastery of understanding how this tool works, intuition and strategic thinking are also likely to improve. Look at modern athletes, participants of the Olympic Games – will you find there at least one who does not pass tests, does not pass a medical examination and does not measure its results? There are definitely no winners among them.
Now go to the basics.
Working with data includes 3 stages:
- Collect data.
- Analyze the data.
- Change behavior to more effective.
Note – these three stages you probably do every day, whatever you do, although they can all fit into milliseconds. Not only that, you learned to walk because your body performed these three stages. It is these stages that lie in the foundation of the well-known NLP-modeling and any models for improving skills. The main difference of business from your body:
- more variables;
- higher rate of receipt of new data;
- a higher level of abstraction (working with our ideas and judgments, rather than sensual perception of the body);
- more significance of the error.
In short, this is a descent on a steep slope at high speed and it is very interesting, reckless. But in order to get more results and buzz, and not a fracture of limbs, it is better to learn how to correctly manage the data. Now I will go through the points of working with data in the context of web analytics.
At this stage, it is important to get complete and reliable information in web analytics tools. Or in the words of a well-known social technologist – your picture of the world should be adequate. To make the right decisions, the information should at least be:
Complete – enough information to make the right decision or understanding. In order to receive complete information it is important to correctly identify and configure the site’s goals, and the necessary events for tracking. This is critically important. Without a checkpoint in the form of the goals of the company and the site can not ask the right questions.
Reliable – information obtained without distortion. In order for the collected information to be reliable for analysis, you must always configure filters (on employees’ devices), prescribe one type of utm-tags for one source and test the changes made to the analytics code of the site.
Errors made at these stages make the analysis in advance or erroneous, or even useless.
At the data analysis stage, it is important to ask the right questions . The main tool in data analysis is segmentation and data detailing . They help to track which set of characteristics of advertisements, the audience gives the most achievements of the company’s goal, which products, services, pages are more effective and a lot of other useful information. This stage can be described in two main steps:
- Ask the right question.
- Find valuable information for a conclusion or A / B testing (as an answer to a question).
Change behavior to a more effective
Based on data analysis, we either change our behavior to a more efficient one without testing or form a hypothesis, which we test with the help of testing tools. The essence is the same – we need to increase the efficiency of invested funds. If more detailed, we can:
- Disable ineffective advertising campaigns and increase the budget for effective ones.
- Make changes to the site that increase the effectiveness of existing sources and channels of traffic.
- It is more accurate to determine the characteristics of the audience by location, gender, age, involvement, willingness to buy and create more effective advertisements and targeted advertising. And many other features, depending on the level of customized analytics.
We now turn to an important block of errors. Studying them carefully will help you better understand yourself, your business, and the work of contractors. All the basic mistakes that a person makes in reasoning are described in detail in the books of Daniel Kahneman. But here I am bringing precisely the distortions associated with making decisions at the start of a business.
The main mistakes of the owner of the Internet project:
- Underestimation of primary planning.
- Focus on irrelevant details.
- Misunderstanding of the essence of the questions asked.
- Analytics for analytics.
- Retrospective rationalization.
- Risk assessment in decision making.
Now in order.
Underestimating primary planning
Competent tuning of analytical systems should be planned – this is an axiom. This is because the owner himself often cannot accurately form the goals of the site and, even more, divide them into macro and microconversions. This issue should be discussed in the preparation of the project promotion strategy and be described in detail in writing. Competently asked questions to the owner not only simplify the work of the analyst, but also clarify a lot for the entrepreneur himself in understanding where he goes, and by what road signs he orients himself.
Emphasis on minor details
This error most often happens in the early stages of project development. When the data sample is small, and the owner already makes a conclusion about the effectiveness / inefficiency of certain channels, the sources study all the characteristics of the audience. This raises the issue of data completeness. Are they enough to make a decision? It is better to wait, collect more data and make a more statistically reliable decision.
Misunderstanding of the essence of the questions asked
There are two options:
- the owner is aware that he does not understand the question – in this case it is important to explain in simple terms both the question and the answer to it.
- the owner understands the question, but does not understand the meaning of the answer in the whole analytics chain – here it is important to feel and predict why the owner asks the question and what decision he plans to take on its basis; This will help avoid misunderstanding and speak the same language.
To avoid these distortions, it is best to create single documents for the entrepreneur and the contractor in which the goals of the site, micro, macroconversion, KPI are stated in the general conversation. Such a terminological base.
Analytics for analytics
This is a fault for almost all small business owners. They enter analytics once every 2 hours and observe, hope, as their offspring grows. I myself do this sometimes, and this is difficult to avoid. But it is important to learn to use analytics only in the context of specific issues. In addition to taking time, this behavior drains attention and decision-making resources. An illustrative example of limited resource decision-making is given in one piece from the book of Daniel Kahneman.
Basically, right after a short break, the judges came in a more peaceful mood and showed more condescension. They demonstrated a more vivid imagination and the ability to understand that the world and people can change and become better. But as they burned their energy reserves, they began to make decisions that preserve the status quo. I am sure that you ask these judges if they are sure that they made the same right decisions every time, they would be offended. However, the numbers do not lie. And sandwiches too. When we do not have energy reserves, we tend to act recklessly. This phenomenon has been dubbed “ego depletion.” The idea is that making any decision requires you to energy costs. This is a strange kind of exhaustion – you do not feel physically tired, but your ability to make informed decisions is reduced. What really changes is our self-control – our ability to be disciplined, thoughtful, and to calculate the consequences.
Let me give you another amazing experiment. A group of researchers wanted to figure out how decision making influences self-control. To do this, they gathered university students – traditional for psychological research of guinea pigs – and asked some of them to make a number of decisions. Namely: students were shown various products and asked to choose one of them. They were told that they had to choose carefully, because at the end of the experiment they would be given the chosen products and it would depend on their preferences what they would receive as a gift. Another group of students did not need to make decisions. The control group was asked a series of questions. What kind of scented candles do they like – vanilla or almond? What brand of shampoo do they prefer? Which of the cookies do they offer? Then they conducted a classic self-test: how long can you hold your hand in icy water? Whatever efforts you make when making decisions, the same resource is involved in self-regulation. The students from the control group could not hold their hands in icy water for as long as the group that did not make decisions. You can summarize. There are a limited number of sound decisions that you can make in a day, but if you take them more and more, you will destroy your ability to regulate your own behavior. You start to make mistakes – it is possible that very serious. As the Maxwell curve shows, bad decisions affect performance. So go home at five. Turn off the cell on the weekend. Watch the movie. And finally, eat your sandwich – apparently, this is the most important thing. By not working so hard, you will do more and better.
The bottom line is that the brain tries to rationally justify emotional actions in the past. But the same name can be given to behavior when you are looking for the coincidence of some causal relationships in web analytics with a large number of variables. For example, you dropped sales last month and you also turned off contextual advertising during this period. The brain will immediately put its cause-effect order. But it is important to always clear your past data and ask the right questions. What other variables affect sales? What other changes were made to the advertising campaigns, the site, and what remained static? To find the reason it is important to collect a complete picture and always remember about the possibility of chance. There are answers whose search will only drain you. Sometimes it is more effective to build a future picture of the world based on the ideal model your business than digging into the past.
Risk assessment when making decisions
The analyst on the basis of the analysis gives conclusions and suggestions for testing hypotheses, but the question of whether it is worth testing something at all is also influenced by the amount of resources and risks involved. Here lies the main problem in the communication analyst and the entrepreneur. Much depends on the experience of past work, managerial roles. Does the entrepreneur trust the analyst in the miscalculation of risk and decision making, or is the analyst only an adviser? Each situation is individual. Analytics here it is important to explain in detail to the entrepreneur the ratio of risk / income when following different paths. And the entrepreneur will share his rich picture of the world of the entire business so that the analyst can better adjust his own.