Applying Filters To Your Log Files

Filters can be very useful for those who work in a marketing/sales department. Even those who use the Freeware version of Expert Data Miner - Log Analyzer may need filters more than they think. If you are an individual who has just started to develop a website, it can be frustrating to see that your own constant modifications distort the statistics regarding the real number of visitors. If you upgrade your site five times per day this will have a significant impact if the initial number of your visitors is very small. There is a very broad choice of filters in EDM,  you can even combine several filters with an AND/OR condition.

Filters can be used from two buttons:

The net will give you a large number of filters, but the right button (some kind of calendar) is just a quick filter on dates.

Here the button with a net was selected. You can choose a filter and re-scan your log file. When a filter is active a message appears in the status bar in order to remind you that the global filter mode is on.

Now if you do not pay attention to this message, you may search for a while before to realize that your old filter needs to be deleted. To delete a filter one has just to select it, press the button Delete and then Save.


Let say that you have a company and you launched a promotion for Canadians for a specific product. You want to target the visitors from Canada who reached your site but who did not visit one of your page (it can be a page where they can sign up for something). In some cases you can contact them by email in order to make a follow up and explain them what they missed. If you have sell a large volume of cheap items, it can be preferable to study their behavior and adjust your web pages rather than to contact them one by one if your initial strategy failed.

Lets say here that your product costs several hundreds of dollars; tracing back your customer one by one (when you can) is not a waste of time here. For that example, you are doing the promotion of a book to some potential resellers, if they purchase 1,000 of them they get a substantial discount. The first step is to define a non cumulative project for this specific goal:

The second step is to define an include filter on your visitors who are Canadians:

However you also want those who did not visit your target page; the combo box OR/AND allows you to define the condition between your two include filters. In this case it should be set to AND. If you need to define two include conditions linked with an OR, you cannot define an AND with your third include condition, the AND/OR flag is used between ANY include filter. So if this scenario happens one day, nothing prevent you from using an OR condition between two include filters plus an exclusive filter on people who DO visit the target page (in this example we target those who did not visit it with an include filter).

The next step is to save your filter and rescan your log with the proper button. When this is done, you go to the last report, visitors-details

It's not a bad idea to select the first 100 items and right click to fetch their DNS because you may need this information. Now let say that you want to know what happened with a specific customer; as usual you select the row, right click and choose to view the click stream :

In some cases the DNS itself contains the name of the company (your visitor). In other cases your visitor can be a registered visitor who logged on, so his authentication ID will be present; if he provided his email some weeks ago when he purchased some article you probably have a table that can link the cookie and email addresse of this visitor to a user session. In the Enterprise Edition of EDM you can import such links and build reports on the fly.

Lets now see another example; you may be puzzled about the real number of visitors that your website receives, in many situation this number is drastically inflated  with some kinds of 'false visitors'. Many of the 'visitors' on www.expertdataminer.com are false visitors. Presently most of them come from a website called software dungeon but in a month they will come from elsewhere. What happens is that several download websites will list this site among many others and each time that someone opens a specific page it will load an icon or a small picture from dozens of remote websites who announce their products in the same category. Most of those people will never ask a page from this website and many will still ignore its existence at the end of the day.  Those people can be filtered out if I create two inclusive filters, each on the Visitors who requested specific file types, except that from the combo box the first target will be Pages and the second Files;  Resources are left out.

In this case an OR condition is selected from the combo box above the button Filter.

If you have a small website, you can also filter yourself out from the statistics. One possible filter is Visitors with a specific authentication ID; you probably use this ID when you try to use your file manager or some other tool. If you are prompted for a userID/password before to connect to those tools (cpanel, file manager or anything like this) your user ID is probably present in your log file. Another possible filter that you can use is Visitors who requested a specific file; Nobody else than you should use your initial management page. In these cases you should use an Exclusive filter before to re-scan your log.

 If you use a filter on a specific page or directory, you must use the standard that appears in your log file, so normally when you are prompted for a page name you should type /mypage.html  or /mypage.asp?id=9643 and not http://mydomain /mypage.html.

 

Next Section: Website Optimization

Back to the home page