Friday, November 07, 2008

Should You be Freaking Out about your Analytics? Tracking the Spikes and Dips.

This Blog Post is written by Aaron Lovelace. Aaron is an Analytics and Optimization Analyst on my team and he will be guest blogging on this blog.
One of the challenges in web analytics is knowing whether a spike or dip in web metrics is something to be worried about. One common way that we deal with knowing whether to ‘freak out’ or not is to conduct a historical baseline analysis. Now, if you are a WebTrends customer, there is an easier and less time consuming way. But first, let me explain how we normally go about it so that you can fully appreciate the new development.

For any given site, our team typically maps out historical trends and create a set of standard metrics to measure against. We call this process a "baseline analysis." These baseline metrics can then be used to determine whether what we are seeing in our analytics is normal or not.

If you do this type of baseline analysis on your site and you notice that your metrics are abnormally good, you can turn whatever caused the spike into a best practice. If your metrics are unusually bad, try to avoid whatever caused the problem in the future. Easy enough, right? Well, sort of.

It is easy to see that conducting this type of research is a necessity if you are serious about success, but that doesn’t change the fact that it is a time consuming and laborious process. If you hire a consultant to do a baseline analysis for you, it could get expensive.

So here is the good news— there is a now faster and more accurate way to track spikes and dips in WebTrends… and to know if you should freak out about them.

WebTrends recently announced a new service (through a partnership with Technology Leaders) called “Dynamic Alert.” Among other things, Dynamic Alert allows you to track spikes and dips that deviate from your website’s historical norms automatically.

It should be noted that SiteCatalyst has a feature called simply "Alerts" which allows you to be notified if your metrics exceed a pre-set metric that you specify. With Alerts, you still need to do a manual baseline analysis to figure out which number(s) to enter in the Alerts configuration.

This is what is so great about the new WebTrends tool—you don’t need to pick a number, it just does it for you by automatically analyzing your historical data. One feature I would like to see, however, is the ability to integrate with other analytics tools beyond WebTrends.

Keep an eye out for more information from WebTrends and Technology Leaders about this useful new tool. Although Dynamic Alerts was developed by a partner, if you sign-up through WebTrends, your bill for this service will just be added to your regular WebTrends bill.

Looking to fill your Web Analytics or Online Marketing position? Post your open jobs on WebAnalytics Job Board

New Position

Technical Consultant for Internet Marketing and Web Analyticsat Unica (Waltham, MA)


  1. You make a good point. Contextualizing data is critical because a reading of a single metric by itself gives little value. Looking at data historically, however, is only one way to add context, but it doesn’t tell the whole story; e.g. it does not deal well with macro conditions that are often outside of the firm’s control. That’s where competitive contextualization comes into play: it provides another data dimension and lets you see how well you’re doing against peers and competitors who may be impacted by the same macro conditions. Coremetrics Benchmark provides such capabilities:

    Boaz Ronkin

  2. Hi Boaz, you hit the nail on the head. There several ways to contextualize and add meaning to the data, including as you mentioned, looking at competitive metrics.

    It is easy, however, for people to get overly concerned with comparing their site to the "average" as shown by one of these reports. There are a lot of factors at play that might throw a site off of the 'norm' including pricing, offers, value propositions, etc. So as you mentioned, it is probably best if they are used just as another data point and not as a sole reason to jump into action.


  3. Hi Aaron.
    I'm a big believer of a more automated analytics process. I think that most companies out there are not looking for more custom reports, but actually better ways of understanding and using the vast amount of data they already have.
    Last week we publicly launched our own solution and vision for this which we call "Pro Active Analytics" (you can read more on that here:
    I'll be really happy to actually share this with you and get a few moments of your time to hear your thoughts and what we need to do better.
    If you are interested, please ping me at shahar at nuconomy dot com.
    Thanks, Shahar Nechmad
    NuConomy, CEO


I would like to hear your comments and questions.