Forum Moderators: DixonJones

Message Too Old, No Replies

Vistior path analysis

What exactly do YOU measure?

         

fom2001uk

11:40 am on Jan 13, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Since an average site has several hundred pages, and an average visit involves several page views, then the total number of permutations for paths is well into the thousands.

That's why the "most popular paths" report is a waste of time. There are just too many, all representing tiny fractions of paths taken (with lots of single page visits near the top usually).

So my question is how do use path analysis to get meaningful info? What filters or criteria do you apply?

A few ideas :

Common paths to sales/orders?
Common paths to abandonment?
Paths that begin or end on key pages (which pages?)
Paths that are referred from a specific channel (search engine, affiliate, etc)?

What paths are you investigating, how are you doing it, and why (what insight does it give you?)

Brett_Tabke

11:49 am on Jan 13, 2003 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



The one that is important to me, are those that come in on the home page:

- where do that go? (what info were they looking for). Pump up that content and expand it.
- what page did they bail out on after coming off the homepage? That page is weak and they didn't find what they were looking for.
- what page did they enter on, that most often ended up in a sale or commision?

fom2001uk

12:28 pm on Jan 13, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Some good advice there Brett. So you filter for the homepage as the entry page, then look at their paths, or filter for key exit pages, etc?

How long a path are you looking at though, are there limits?

Brett_Tabke

12:44 pm on Jan 13, 2003 (gmt 0)

WebmasterWorld Administrator 10+ Year Member Top Contributors Of The Month



Ya, there are too many other entry points on most sites to be of value. Those other entry points are more a reflection of your promotion and se activity than anything else.

aspdaddy

3:11 pm on Jan 13, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



My log entries dont have any session id's, so I guess this kind of analysis cant be done from log files. Can you reccomend s/ware for this?

fom2001uk

4:22 pm on Jan 13, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



I don't think you need session IDs. There are a few ways to determine unique visits (agent, IP combination, cookies, etc).

But I suppose different software packages will offer diffrent levels of visitor path analysis. We use 2 packages and one of them is much better at visitor path analysis than the other. Sticky mail me and I'll give you more details.

Tabas

9:20 pm on Jan 15, 2003 (gmt 0)

10+ Year Member



Good points. I also want to filter on entry pages linked from emails sent to customers. Haven't tried it yet, but I read that you can add "?whatever" to the end of a URL in an email so it logs differently than other entries on that page. I'm curious if (client) visitors from emails have different or longer paths than those from other sources.

123loganalyzer (www.123loganalyzer.com) seems to offer good path analysis and strong filtering capability. It also includes some marketing analysis features, like "Conversion Analysis By Target Web Page Views". Anyone tried it? I'm very new to all this so appreciate any feedback or similar alternatives.

aspdaddy

11:08 am on Jan 16, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Tabas,

Fom is right, with a combination of Time/IP/User Agent/Screen Resoultion etc. you could reasonably determine sessions from the log files.

With something like M/S Log Parser you could query for those fields on any referer with?whatever, then run a second query to get the click stream of each...in theory :)

Ash

10:18 am on Jan 17, 2003 (gmt 0)

10+ Year Member



Brett, when assessing entry points of value aren't 'clicks-to-preferred-outcome' a valid part of the equation?