Wednesday, December 07, 2011

Do you need Real-Time Web Analytics?

Real-time web analytics provides you a view into what is happening on your site at that very moment. It is really interesting to see where visitors are coming from, what keywords they are searching, what pages they are viewing etc. Though most of the time that’s where it ends i.e. it is interesting but not very valuable. As many web analysts have stated time and again, the value of the analytics comes from the action you take on that data. So, unless you are going to take actions in real time you really don’t need real-time analytics. However, I can understand the temptation to use Real Time Analytics for instant gratification.

Side Bar:
Recently Google Analytics joined the bandwagon of providing Real-Time analytics. Other notable real time web analytics vendors include Chartbeat, Woopra. As of now, Google analytics only provides a very limited view of real time stats at this point, though I am assuming that it is just the beginning and Google will roll out more stats in its real time reports. 

Few cases where you might want to (or be tempted to) use Real-Time Analytics
  • You launched a new campaign e.g. paid search, email newsletter, TV ad , and would like to see how people are reacting to those campaigns.
  • You added new promotions on your site and want to see how visitors are reacting to those promotions, so that you can tweak those promotions in real time.
  • You added new stories, links etc. and want to see if anybody is clicking on them so that you can make some changes based on instant feedback. I can see the usefulness of this feature for news and media sites.
  • You made some technical changes e.g. changed tracking code and want to see if those pages are being recorded in Google Analytics. Real time reports can serve as QA tools.
  • You launched a new feature on your site, launched a video, deployed a new game and would like to know if your visitors are using it or not.
Keep in mind that even if you are ready to make changes in real time, you might not have statistically significant results based on few data points that you get in real time reports. If you have nothing better to do then you can for sure kill your time with some real time view into your site traffic.

Views from Twitter

What do you think? Have you found Real time analytics to be useful? How are you using it? Please share your views.

Web Analytics Jobs


  1. Anil,

    The term "real-time" has always meant too many different things and unfortunately it has been tied to whichever vendor the client is using.
    The thing to keep in mind when deciding if you *need* real-time, and if so which type of real-time is right, is to understand your business and what the company will use the information for.
    Many times I have seen a company demand that real-time yet when asked "What team would use the information to make changes with-in seconds or minutes based upon what they see?" the company typically will say either - 'Well, no one today, but...' or 'My executive wants to know the information ASAP so we must have it.'

    Both of these points are a complete failure to understand what real-time provides.

    Real-time reporting is great if you have nothing better to do than sit there and watch the wave of usage. The unfortunate truth is that most offerings that provide this kind of reporting are essentially 'bean counting' and there is no understanding of WHO, rather they focus on session level information. Without sessionization what is the value of the information if you have nothing to contrast it against – especially when you are trying to make important decisions.
    If you are a retail site, perhaps real-time reporting is important to make sure that the shopping experience is not interrupted and if so which part do we need to fix.

    The real value that I think real-time offers is when a company is able to operationalize the information into their content management systems and for many media/content companies this is where I see real-time being most important.
    If you use Yahoo’s homepage as an example, the main content in the middle of the page should not be stagnant for 30-45 minutes, never mind 24 hours. If an article is seeing higher readership or another one is simply wasting space, you need to adjust accordingly. That said, the process of adjusting those items should not be manual. Understanding the consumption of the content needs to occur in a way that the CMS is able to make the change without human intervention.

    In my opinion the current state of real-time is much like the way web analytics started – IT driven. Until the information is able to influence immediate change and has clear marketing or financial implications it is simply a waste of analysis.


  2. Great post Anil. "unless you are going to take actions in real time you really don’t need real-time analytics" says it all.

    I absolutely agree that there is risk with jumping to conclusions on real time data. This is true even for data that has only been collected for a few days. I recently saw a request for data on the number of visitors from the new Kindle Fire device to a specific site. The Kindle had only released two days before. I thought it was very cool that the data was available, but the first thought to cross my mind was "I hope they don't make any business decisions based on this data".

    I find that when I am doing analysis on recent data that I will often intentionally exclude the current day. I can't compare the current day against previous days as I don't have the full data set yet. If I'm doing the analysis for any sort of lengthy period then the numbers keep going up (especially if I'm pulling the numbers during peak traffic hours).

    Real-time reporting is helpful when validating a implementation, but I prefer to use a web proxy tool like Charles ( as it is not only real-time, but I can do things like swap out a dev code file on a production website to test how the code update will impact the site. I'm also able to see if a page is firing off the image requests exactly when I want it to (very useful if you have multiple people testing at once).

    One place where real time reporting is helpful is with automated alerts. If you have a process to pull in new data every 15 minutes to 30 minutes on a key report then you could setup a automated alert to notify you if the data falls above or below is some sort of predetermined threshold. Take the process a step further and setup the alert to send a text message to the individual(s) in your organization that can take action on the data.

    Adobe Consultant


I would like to hear your comments and questions.