A better estimate of visit duration

I am new to Matomo, so my comment may just reveal my ignorance! If so, please forgive me.

I have just installed Matomo on 2 websites to replace GA and I am currently testing it. I’m finding the recorded visit duration to be very short. When I check I find that the last page of every visit is logged as zero time, because Matomo doesn’t know how long the visitor stayed. But this makes the average visit duration very small, especially if many visitors only visit one page.

As an example, suppose there are 100 visits over some time and 150 page views, each one last 2 minutes. Only 50 pages will have the 2 minute tie recorded and the other 100 will be exit pages and recorded as zero. Thus the average visit duration will be calculated as 50 * 2 /100 = 1 minute, when the actaul figure was 2 minutes.

Wouldn’t it be more accurate to not set the exit pages as “zero” but to set them as “Not recorded”, and then a more realistic average visit durection would be calculated, albeit only use some of the data?

Hi @EricH
Did you have a look at:

Also Matomo has some configuration called Heart Beat Timer. This can be activated by JavaScript:
Also available on MTM, in the Matomo configuration variable:

Hi Philippe, thanks for replying to my comment.

I have seen that page, and I have read (very briefly) about those methods of measuring the time spent on the last page. My difficulty is I only have two small private websites and I am on the free Matomo plan, so I cannot really afford to pay for the full Matomo product. (I appreciate being able to use it for free, so I don’t expect to have everything I would like.)

But the method used is a poor estimate of the time per page. This is especially so if (as is the case for most websites) that many visitors only visit one or two pages. Then, exit pages will perhaps be half the total pages, and setting their time to zero gives a very poor and understated estimate. Far better, as I wrote, is to set the time for exit pages as “no value” instead of zero, and then the estimate is based only on the pages where a time is actually measured. It wouldn’t be hard ot make this change.

Fortunately, I can do a simple calculation to recover the true average time. But other users of the free plan may not realise this.