Useful Calculated Metrics for SiteCatalyst

Equation on a blackboard

I created a traffic source report in SiteCatalyst Discover recently and wanted to include some basic engagement style metrics e.g. Average Page Views per Visit, Average Time on Site.  However, with some high bounce rates, I wanted to ignore those visits that did bounce, with the metrics only calculated for visits that viewed at least two pages.

The general equation for this for any metric is:

(<Metric> minus <Metric> for Single Page Visits) / (Visits minus Single Page Visits)

The four required numbers are generally pretty easy to get at but it is complicated here as the reports are based on Last Touch Marketing Channels reports.  As discussed in this blog post on Why SiteCatalyst Visits don’t always equal Entries, not all Marketing Channels Visits are recorded in SiteCatalyst as an Entry and therefore cannot be recorded as Single Page Visits.

The solution is to create an estimated value for this by applying the known Bounce Rate for Entries to the number of Visits (assuming the Bounce Rate is consistent).  This equation is:

[Visits]*([Single Page Visits]/[Entries])

This leaves the only missing number as <Metric> for Single Page Visits.  Here, we can apply some logic to either eliminate it or use a substitute.  Common examples are below:

  • Average Time on Site (per visit) – no time is recorded if only one page is viewed so value is zero
  • Page Views per Visit – as only one page is viewed, the number of page views equals the number of visits
  • Orders per Visit (Conversion Rate) – ignoring an issue with the set-up, you can’t place an order on the first page you view therefore the value is zero

So the set of equations to use in Discover to create the Calculated Metrics are:

Average Time on Site

[Total Time Spent]/([Visits]-([Visits]*([Single Page Visits]/[Entries])))

Page Views per Visit

([Page Views]-([Visits]*([Single Page Visits]/[Entries])))/([Visits]-([Visits]*([Single Page Visits]/[Entries])))

Conversion Rate

[Orders]/([Visits]-([Visits]*([Single Page Visits]/[Entries])))

The final metric there is the conversion rate, again adjusted to eliminate those visits that were never a realistic chance of converting.  The conversion action used was to place an order but this could easily be changed to other conversion actions such as Submit Form.  It is important though that the action/event would only be triggered once within a visit.

So what do you do with these calculated metrics?  You can set up a report for Last Touch Marketing Channels and add metrics for Visits, Bounce Rate*, Page Views per Visit, Average Time on Site, Conversion Rate, Orders and Revenue.  And then analyse, interpret and take actions based on the comparisons between channels for these key metrics.  Or if you drill down (and have classifications applied to the Marketing Channel Detail values where required), take actions based on the comparisons between different campaigns, affiliate networks, search engines or landing pages.

Sample Marketing Channels Report layout

Note this report is only possible with Discover, not SiteCatalyst, as Single Page Visits is a traffic metric and not available for the Marketing Channels report.

If you would like to know more about calculated metrics, this is a good post on how to create calculated metrics.   And, within the Knowledge Base, can be found a list of commonly used calculated metrics for SiteCatalyst.

* If you don’t already have Bounce Rate set up as a calculated metric, bad you.  The calculation to use is [Single Page Visits]/[Entries].

Share and Enjoy:
  • Twitter
  • Digg
  • StumbleUpon
  • del.icio.us
  • LinkedIn
  • Facebook
  • Yahoo! Buzz
  • Google Bookmarks

Tags: , , , ,

4 Comments

  1. J March 17, 2011 at 3:25 pm #

    Average Time on Site formula can be significantly simplified from:

    [Total Time Spent]/([Visits]-([Visits]*([Single Page Visits]/[Entries])))

    to [Total Time Spent]/([Visits]-[Single Page Visits]) visits and entries is essentially the same so you dont need to work out the % bounce just factor it out of the equation.

    Alternatively I would go even simpler to my version:

    [Total Time Spent]/[page depth>1]

    • Peter O'Neill March 17, 2011 at 4:29 pm #

      Thanks for the response J but these calculated metrics only relate to Last Touch Marketing Channels where visits and entries can be quite different. Entries (and single page visits) are only counted if that access of the site initiated the visit. Visits on the other hand are counted everytime the website is accessed.

      Your suggested calculated metrics would work with probably all other reports but not these ones. I didn’t know you could do page depth > 1 as a calculated metric – will need to have a play with that.

      Peter

  2. J March 17, 2011 at 6:34 pm #

    Ah – I skim read and didnt appreciate that – I also had to look into the apparent metric variation in the marketing channel report in OSC – it makes sense once you know how the back end is attributing. I revisited the page depth in calc metrics and it doesnt appear to work – sure it used:Fail …retract previous post. The answer to all this pain arguably I guess is to go with a TMS and not rely on Omniture for x channel attribution.

    • Peter O'Neill March 17, 2011 at 7:12 pm #

      No problems. Omniture has other solutions for multiple visit attribution, Unified Sources (VISTA rule) and Channel Manager (JavaScript) – although they do have additional costs involved. My concern with any TMS is whether they will capture all traffic sources, not just the tagged marketing channels, and whether you will also get all the visitor behaviour on the website beyond just the conversion event.

      Of course, I don’t really agree with campaign attribution being accurate/useful in general due to multiple device use – no system can capture all the touchpoints if you are researching at work and purchasing at home. Instead I think people should be looking to assign value to actions/behaviour on the website and ideally testing impact of marketing somehow (both difficult I know).

Post Comment

Notify me of followup comments via e-mail. You can also subscribe without commenting.