FIELD GUIDE

Why people leave in ten seconds.

Most people who open a page are gone before they have read a line. Here is what the research measured, which part of that is normal, and what to change on the first screen.

5 sources, each one read in October 2026 and linked at the bottom.

THE SHORT ANSWER

The quick version.

People judge a page in about 10 seconds. If the first screen does not say what this is and what to do next, they go. Some quick exits are normal, so the useful question is which of your pages loses people faster than the others.

THE NUMBERS

What the research counted.

  • 10 seconds

    is when leaving is most likely. Nielsen Norman Group, reading Microsoft Research data, calls the first 10 seconds of a page visit critical to the decision to stay or go. Nielsen Norman Group

  • 30 seconds

    is where it calms down. People who last that long keep leaving, but far more slowly, and often stay 2 minutes or more. Nielsen Norman Group

  • Under 1 minute

    is how long the average page visit lasts, by the same reading. Nielsen Norman Group

  • 205,873 pages

    were in the study behind those figures, each one visited at least 10,000 times. On nearly all of them the pattern was the same: the longer someone has stayed, the less likely they are to leave in the next second. Nielsen Norman Group

  • 55 in 100

    visitors spent fewer than 15 seconds actively on a page, in a month of Chartbeat data covering 2 billion visits. That was 2014, and Chartbeat's customers are publishers, so read it as news sites more than shops. Time

  • 10 seconds, again

    is Google's line too. Google Analytics counts a visit as engaged if it lasts longer than 10 seconds, has a key event, or has 2 or more page views. Bounce rate is the share of visits that did none of the three. Google Analytics Help

WHAT TO CHANGE

Why it happens, and the fix for each.

  1. 01The first screen does not say what this is

    Ten seconds is enough to read one line and look at one picture. Nielsen Norman Group's advice from the same data is to get your value across inside those 10 seconds, because pages that do can hold people for minutes. Nielsen Norman Group

    TRY THISRewrite the first line so it says what you sell and who it is for. Read it to someone who has never seen your site and ask them what you do.

  2. 02There is nothing obvious to do next

    A visit that ends without a tap is not always an unhappy one. Sometimes the page was read and no next step was in view. In eye tracking, the first screen gets more than half of all the looking a page gets. Nielsen Norman Group

    TRY THISPut one button on the first screen, with a label that says what happens when it is pressed. One, not three.

  3. 03The page is not what the link promised

    Someone who clicks an ad for one thing and lands on a page about another has their answer in a second. There is no study behind this one, only your own numbers: compare quick exits by where people came from.

    TRY THISOpen the ad, post or search result that sends you the most visitors, then open the page it lands on. The first line of the page should repeat the promise of the link.

  4. 04Some of it is how people browse

    The Microsoft researchers describe browsing as screen and glean: people screen a page fast, and only settle in to read the ones that pass. A quick exit can also be a visitor who got what they came for, like your opening hours. Microsoft Research

    TRY THISDo not aim for zero. Rank your pages by how many visits end inside 5 seconds, and start with the worst one that also gets real traffic.

ON YOUR OWN SITE

The jargon, said plainly.

Bounce rate is the usual name for all of this. Moxel says it as a count of people.

WHAT IT MEANSfernhouse.shop, week 37
  • Avg. session duration 00:54
    People stay about 54 seconds, then go
  • Bounce rate 40%
    40 of every 100 people left without touching anything
  • Event: input_blur field=card_name
    They clicked the card name box, then clicked away
  • Returning visitors 13%
    13 of every 100 people came back later the same week
  • Exit rate /checkout 52%
    52 of every 100 people who reach payment leave without paying

no glossary needed

Moxel counts a quick exit when a page is left in under 5 seconds, and the fix list names the page once half or more of its visits end that way. At Vibey and Feral it also records taps, so it can say how many visits ended without a tap or a second page. It does not measure how fast your page loads, so check that yourself on a phone.

THE LIMITS

What this cannot tell you.

These studies are averages over other people's sites, and the two largest are more than ten years old. Phones were a small share of browsing then. They tell you what is normal, not what is wrong with your page. On a small site your own numbers are thin too: 6 quick exits out of 9 visits is a hunch, and Moxel labels it as one.

QUESTIONS

Asked a lot.

What is a good bounce rate?
There is no single good number, because it depends on what the page is for. A page that answers one question, like opening hours, can lose most visitors quickly and be doing its job. Compare your pages with each other and look at the one that loses people fastest.
How is a quick exit different from a bounce?
Google Analytics counts a bounce when a visit lasted 10 seconds or less, had no key event and only one page view. Moxel counts a quick exit per page, when that page was left in under 5 seconds. One is about the whole visit, the other about a single page.

SOURCES

Where the numbers came from.

  1. How long do users stay on web pages?Nielsen Norman Group, Jakob Nielsen, September 2011.
  2. Understanding web browsing behaviors through Weibull analysis of dwell timeMicrosoft Research, Chao Liu, Ryen W. White and Susan Dumais, 2010.
  3. What you think you know about the web is wrongTime, Tony Haile, then chief executive of Chartbeat, March 2014. Chartbeat measures attention for publishers, so this is mostly news sites.
  4. Engagement rate and bounce rateGoogle Analytics Help.
  5. Scrolling and attentionNielsen Norman Group, Therese Fessenden, April 2018.

Read in October 2026. Studies get updated, so follow the link for today's figures.

KEEP READING

More from the field guide.

How Moxel records all of this is in the docs. What it never records is on the privacy page.

Find out where
people give up.

Start free