Website Analytics: Which Events to Track to Understand Customer Behavior

Why track events if the counter is already installed

A basic Yandex.Metrica or Google Analytics counter records page views and navigation between them. You see that 200 people visited a service page and three left an inquiry. But why the other 197 left is unknown.

Events show micro-actions: a person opened the form, started filling it out, got to the phone field and closed the tab. Or clicked 'Request a callback' but the form did not open due to a bug. Without events you only see the fact of departure. With events - the reason.

In practice it looks like this: a client comes in asking to increase site conversion. We open analytics, look at the event funnel and see that 40% of visitors click the 'Calculate cost' button, but half of them fail to submit the form. The problem is not traffic and not copy - it is phone number validation blocking submission. Fixed it in an hour, conversion went up by half. Without events we would have spent a week looking for the cause.

Button and link clicks

This is the basic event to set up first. Track all target actions: clicks on 'Order', 'Leave a request', 'Contact us', catalog buttons, transitions to messengers.

Often it happens: the button exists but no one clicks it. Or they click, but not where expected. For example, the homepage has a big 'Get a quote' button, and at the bottom of the page - an inconspicuous 'Request a calculation'. Click statistics show everyone presses the bottom one, because it sits after the service description when a person is ready to act. The top button just takes up space.

Another case from practice: a client complained the messenger button was not working. We looked at events - it gets clicked 50 times a day. The problem was different: after clicking it opened Telegram but without a pre-filled message, and people did not understand what to write. Added a greeting text to the link, number of conversations went up.

Form interaction

Here you need three events: start of filling, filling each field, and submission. If 100 people started filling the form and 30 submitted it - there is a problem somewhere. Look at which field people stop.

Typical situation: the 'Comment' field is required, though the customer has nothing to write there. Or the email field requires a dot and Latin characters, but a person enters a Cyrillic address. The form does not let it through, there is no error message, the user leaves. Without field tracking you only see the fact: form not submitted.

Another point - multi-step forms. If your form has three screens, track transitions between steps. Often people leave on the second step because it asks too much. In one project we removed the tax ID field from the first step, moved it to the end - number of completed applications went up by a third.

Scroll and time on key pages

An event like 'scrolled to middle of page' or 'spent more than 30 seconds on page' shows whether people are reading your content. If 500 people visit a service page per month and 50 scroll to the pricing description - the rest leave earlier. That means the top section did not hook them.

We usually set marks at 25%, 50%, 75% and 100% scroll depth. This shows where exactly people stop. For example, if 80% reach the 'How we work' block and then leave - either the information is sufficient and they moved to the next step, or something repellent follows.

Time on page without events is an imprecise metric, because the counter does not know if a person is reading or just opened the tab and went for coffee. A combination of time and scroll gives a more honest picture: if someone spent three minutes and scrolled to the end, they studied the material. If three minutes but no scroll - most likely just did not close the tab.

Actions in catalog and product cards

For e-commerce sites and service catalog sites it is critical to know what people do inside sections. Track: click on product card, add to cart, use of filters, sorting, click on 'More details'.

It often turns out that filters are not used at all, though budget was spent developing them. Or that products are only sorted by price, and other parameters interest no one. In one project the client insisted on a manufacturer filter - after launch we saw it was used four times in a month. But the in-stock filter appeared in every second session.

Another useful event - click on product image to enlarge. If people zoom photos, it means image quality matters and needs improvement. If no one clicks - perhaps the photos are already large enough, or the product does not require detailed inspection.

Errors and technical problems

Set up an event for any error: form did not submit, file did not upload, payment did not go through, 404 page. A user will not write you about an error, they will just go to competitors. And you will think the site is fine.

Typical example: the payment button leads to a payment gateway, but for some users the session does not transfer due to browser settings. Payment fails, the person tries again, fails again, leaves. Without error tracking you only see abandoned carts and do not understand the issue is technical, not pricing.

Another point - slow loading. If a page loads longer than five seconds, set up an event. Yandex.Metrica will show how many users encountered a delay and on which pages. Sometimes the problem is local: one section lags due to heavy images or an unoptimized script.

Contact views and attempts to reach out

Click on phone number, click on email, opening map with address, transition to social networks - all this needs tracking. A person who clicked the phone is ready to call right now. If there are many such clicks but few calls - either the number is wrong or you are not answering.

In practice this point is often missed, which is a mistake. In one project the homepage had a widget with phone and callback form. Events showed that people click the number three times more often than fill the form. Made the number larger and added a 'Call now' button - number of inquiries went up, though traffic stayed the same.

Another case: a client did not understand why the site had few leads, while traffic was good. Looked at events - turned out people were massively transitioning to Instagram via the header link and leaving applications there. Simply more convenient for them in social networks. Left the form on the site for those not using social networks, but put main emphasis on transitions.

How to figure out which events to track for your site

Start with business goals. What is most important to you: inquiries, calls, purchases, registrations? Track all steps on the path to that goal. If the goal is an inquiry, set up button click, start of filling, field filling, submission. If purchase - add to cart, proceed to checkout, filling delivery details, payment.

Then look where conversion drops. Yandex.Metrica and Google Analytics build funnels: you see how many people performed the first action, how many reached the second, where they dropped off. If half is lost on the second step, dig deeper: what is wrong there, why people are not moving forward.

Do not set up events for the sake of events. If you do not know why to track a logo click, do not track it. Every event is data that needs analysis. Better five events you regularly check and use for decisions than twenty no one pays attention to.

What to do with the data after setup

Events by themselves are useless if not analyzed. Once a week or at least once a month go into analytics and look at funnels. Where does conversion sag? Which buttons do not work? On which pages do people sit long but do not perform target actions?

Next - hypotheses and tests. If you see people start filling the form but abandon it at the phone field, try making it optional or changing validation. If they click the button but nothing happens, check the code. If they scroll to the middle of the page and leave - rework the second half.

At EFIMOV DEV when developing sites we immediately build key events into the technical specification: which actions matter for the business to track and how to set it up. After launch we watch statistics especially carefully for the first month and correct what does not work. Usually during this time all bottlenecks that are invisible at the prototype stage come to light.

If this is already on your plate, here is how we do it end to end: Corporate website development.

Frequently asked