📜 ⬆️ ⬇️

Qbaka Vision - analysis of user behavior

Hi, Habr. Kubaka is officially with you (it's time).

For those who have not heard about us - we are glad to meet you, we are monitoring javascript errors on the site.
And this year they launched analytics of the behavior of Vision users. About her and want to tell.


')
Popular analysis tools sharpened under the web 10 years ago. They measure page views, not the meaning of action.
You have probably noticed that the basic analytics services are complex and confusing. Because of this, we ourselves didn’t do analysis for a long time, although we wanted to. And where are the obstacles, there is an opportunity to simplify. Vision is an attempt to make analytics closer to people.

The main barriers to analytics are the overloaded interface, the complexity of integration, and sometimes the price. To check, let's say that the user is better logged by searching for a product, you need to add an event handler to the code of the search form, wait for the data to collect, and then somehow build a conversion funnel from this. And so for each new hypothesis. In this case, the new data will be only for the recent period, you will not see the dynamics on them. We decided to go from the opposite: first we collect data, and only then you come up with and test the necessary hypotheses.

Vision automatically records all user interactions with the site (clicks, sending forms, changing fields).
From this data, you can build samples by selecting targets directly in the interface of your site. Like that:

image

The analysis center is a click, but page visits can be added to the chain of events. Events do not have to be on the same page. Actions can be grouped and flexibly configured. You can combine two buttons of the same meaning into one event. Or ignore the URL to view the statistics of the login button in the header of the site. Our favorite mode is “select all similar”, with one click you can select all adjacent blocks on the page. There is segmentation by basic criteria and an API for sending events or a user ID. Counting by session and user. And even the information about how many people left the site after a particular action. And, of course, the history of conversions: you can see how changes affect the passage of the funnel.

image

Why is this all when there is GA and Metric?
The short answer is that they are not sharpened by events. In Vision, most hypotheses can be believed in a few clicks by looking at the data immediately for the entire time since the beginning of the observations. No programming and wait processing.

Why is it when there is a Mixpanel?
Because Mixpanel is too expensive and the barrier to entry is still high. To start using Vision, it is enough to put the monitoring code on the website and that's it.

Why when is Heap?
This is the right question. We will compete with them, it is always good to have an alternative in the market.

But there is a Webvisor!
The web visitor is completely different, he will not show you the connection between events (only individual user sessions, from which you don’t know the number of conversions) and he has strict limitations on the amount of recorded data.

We do not compete with GA and Metric.
Vision complements them. By the way, we do not have data sampling at all. The only restriction is that we keep the data for 1 year.

image

Recording all events is a lot of data. Under this task, we wrote our own event storage and event handling engine, which was rigidly tailored for a specific task. We have to do a complete enumeration of all project data for each request. How the architecture of Vision is arranged we will tell in one of the following posts. The most important thing: we keep any workloads without problems, right now we are receiving data from sites with several million users a day.

Come try and write about your impressions. And if you have no one to do analytics in your company, then write to us at support@qbaka.com - we will connect you with our partners from AIC at a discount, they will undertake interface optimization.

Source: https://habr.com/ru/post/254047/


All Articles