Goal Funnel Analysis in Splunk

fitzgeraldsteel — Tue, 04 Jan 2011 23:51:40 GMT

How would I query for transactions that first went to page A, and then page B?

For one use case, I'm looking at goal funnels. Goal funnel visualization is a common web traffic analysis, where you look at the percentage of visitors that follow a predefined set of steps. For example, I've got a shopping cart: of the 100% of visitors that enter the site, 30% put something in their shopping cart, 7% start the checkout, and 2% complete checkout.

I see how I can use splunk to examine the weblogs, and group discrete events into transactions. But is there a way to calculate the goal funnel percentages from those transactions?

Re: Goal Funnel Analysis in Splunk

sideview — Fri, 14 Jan 2011 17:19:22 GMT

Sure. Let me kind of manufacture some details. If I'm way off base this might not be useful but it's worth a shot.

Starting point is some transaction like this:

<your search> | stats values(url) as urls by sessionId

(Often people start with transaction but stats is easier once you get used to being specific about what you need. )

From that point you can do the following. There's quite possibly a more elegant way to to this but this is the way that springs to mind::

<your search> | stats values(url) as urls by sessionId | eval added_to_cart=if(searchmatch("url=*/cart"),1,null()) | eval started_checkout=if(searchmatch("url=*/checkout"),1,null()) | eval completed_checkout=if(searchmatch("url=*/checkout_complete"),1,null()) | stats count count(added_to_cart) as step1 count(started_checkout) as step2 count(completed_checkout) as step3

and from there making percentages is just a little bit more eval but I think you see the idea. Hope this helps a bit.

topic Goal Funnel Analysis in Splunk in Splunk Search

Goal Funnel Analysis in Splunk

Re: Goal Funnel Analysis in Splunk