How to watch Webvisor correctly?

Only by segments. Viewing the records in a row gives almost no conclusions: most of the visits are ordinary and do not explain anything. Selected groups are useful - those that did not reach the goal, those who left a specific page, abnormally long or short visits.

How many records do you need to look at to make a conclusion?

Usually ten to fifteen within one segment are enough. If after fifteen records the pattern is not visible, most likely the segment was chosen poorly and needs to be narrowed down.

How to understand in Webvisor that the traffic is bot?

By coincidence of signs: the same scenario for different visits, uniform scrolling without stopping on the content, lack of interactions with elements, zero goals achieved, and often at night. One sign means nothing, four together is almost a diagnosis.

What to look for in the records of those who did not complete the application?

The moment of stopping. The man scrolled to the form and left - the question is about the form itself. If you didn’t finish it, the question goes to what was above. Clicked on an inactive element - a question about the layout: people often click on something that is not a button.

Does Webvisor help you understand why conversion rates are falling?

It shows how, but not always why. The recordings are good at identifying mechanical problems: a broken mold, a non-obvious button, an error on a mobile phone. Reasons from the area of meaning - an incomprehensible offer, distrust of the price - are less visible from the records; they require surveys and conversations.

Do I need to watch Webvisor regularly?

Not in a row and not on schedule. A reasonable mode is to open it for a specific question: why the conversion rate on this page dropped, what is happening with the new form, why the channel has abnormal failures.

All articles

PerformanceSeptember 5, 20269 min

Webvisor: how to read posts without spending a week on it

The web viewer shows what a person did on the page, but you can’t look at it in a row - you’ll drown. I’ll figure out how to select records in segments, which five scenarios give the most conclusions, and how records distinguish bots from people.

Webvisor is a tool that everyone uses and almost no one uses. There is only one reason: they try to watch it in a row, drown in boring recordings and conclude that it is of little use.

The value of Webvisor is not in the records, but in the segment for which you selected them. Without a segment, it's just someone else's sessions.

1. Five segments that provide almost all conclusions

Segment	What are we looking for?
Reached the goal, but didn't complete it	What exactly stops you at the last moment?
Exit key page	Problem on a specific screen
Visits shorter than 15 seconds	Mismatch between expectation and content
Visits longer than an hour	Usually not people or abandoned tab
Repeated clicks on a button	The form is not submitted silently

The first line is the most valuable and most underutilized. These are people who already wanted to: they finished scrolling, opened the form, and started filling out. Everything that stopped them is costing you money right now, and you can literally see it with your eyes.

2. How many records to watch

Within one segment, ten to fifteen are usually enough. A pattern emerges quickly: if five people in a row stop at the same form field, you don’t have to look any further.

If after fifteen entries there is no overall picture, the problem is in the segment, not in the quantity. This means that you have brought together too many different people, and it needs to be narrowed down.

3. Shapes: the most common mechanical failure

A scenario that occurs regularly: a person fills out a form, presses a button, nothing happens, he presses again, then again, then leaves.

From an analytical point of view, this looks like a normal failure. From a recording point of view, it’s like a lost application. There are different reasons: invisible error message, field validation without explanation, broken sending on mobile.

Second most common story: clicks on something that is not a button. The title looks like a link, the card looks clickable, the icon looks like a control element. People poke, nothing happens, and it builds up into irritation.

4. Time anomalies

Very short visits. Fifteen seconds means that a person opened the page and immediately realized that he was in the wrong place. You need to look at the first screen: there is usually a discrepancy between the promise in the advertisement and what the person saw.

Very long visits. An hour or more - almost always either an abandoned tab or not a person. The difference is visible in the movement: a person’s cursor stops, returns, twitches; Automated bypass traffic is smooth and monotonous.

5. How to distinguish bots

One sign does not prove anything, a set of signs almost proves it.

the same traffic scenario for different visits;
uniform scrolling without stopping on the content;
no interaction with elements;
zero goals achieved on a large number of visits;
night time and similar devices.

Why is this necessary in practice: such traffic distorts bounces, depth and conversion, and decisions on average numbers are made on corrupted data. Having detected it, the channel or segment is excluded from the conclusions, rather than trying to optimize it.

6. What Webvisor will not show

It answers the question “what did the person do”, but does not answer the question “why” well.

Mechanical problems are clearly visible: a broken shape, an unobvious button, a layout error on a narrow screen. Semantic - almost none. He left because the price seemed high, or because he didn’t believe it, or because he didn’t understand what you were offering at all—it’s indistinguishable from the recording.

For semantic reasons, other tools are needed: surveys, conversations with those who refused, wording tests. What to check on the page itself - in parsing the landing.

7. In what mode to use it?

Not on schedule and not in a row. The web viewer opens to a specific question: why the conversion on this page dropped, what is happening with the new form, why this channel has abnormal failures.

Then half an hour of work gives a concrete conclusion, and not the feeling of “I looked, everything seems to be fine.” Setting up the goals themselves, according to which segments are then built, in parsing Metrics.

8. Summary

The segment is more important than the recording. Five working segments: not reaching the goal, exits from the key page, visits too short, too long, repeated clicks on the button.

Ten to fifteen records within a segment produce the output. If they don’t give it, narrow the segment, and don’t watch another fifty.