Digital Marketing: Promises, (Web Analytic) Promises

“An ounce of performance is worth pounds of promises” – Mae West

 

It seems like analytics is back in vogue with marketers these days, even more so if you use “big data” as a qualifier. But how much value do web analytic tools deliver?

 

The customer’s journey is inherently more complex these days and marketing reflects this through multi-channel campaigns.  It is common for marketers to touch a customer multiple times over multiple channels over time. So when a conversion does ultimately occur, how do you determine the primary contributor to the conversion event?   Was it the digital sign that the customer saw while he/she was stuck in a traffic-jam on the freeway?  Was it the two emails they received over the span of the last month with the promotion and positioning content?  Was it something they heard while talking to the customer service agent while calling in to report a problem with a different product they had purchased?  Or was it a specific sequence of interactions the customer had with the company that caused them to buy the product? 

 

You might think that marketers should be able to answer this very simple question from their web analytics platforms.  Isn’t that what all web analytics platforms promise?  Feed us the click-stream and/or all metadata and everything shall be revealed in the dashboard.

 

Traditional web analytics platforms tend to focus on collecting click-stream data from websites. For most marketers, it serves as a report generator that gives them some historical information about how many visitors came to their website, how long they spent on the site and which pages they viewed.  Interesting information, but none of these metrics are measuring true business outcome. 

 

Avinash Kaushik, a leading analytics evangelist and subject matter expert talks about this at length on his blog and in his book.  He states that while these metrics are important, they do not move the needle in terms of business performance and do not help in really quantifying the economic value of the customer interaction.  I’ve heard of some marketing departments that didn’t notice they hadn’t been receiving their weekly web analytics reports in over three months.  

 

It’s not that web analytics aren’t useful; they happen to serve a very tactical purpose.

Here are three main reasons why:

 

1)      Web analytic reports or analytics aren’t actionable.  Most web analytics platforms are completely different from the tools that marketers use on a day-to-day basis to manage their campaigns or publish to their web sites.

2)      Web analytics aren’t tracking business metrics. Marketers are not using their platforms to track business metrics like conversion rate, revenue or value per acquisition, what are the high value segments etc.

3)      Web analytic platforms look at the data in aggregate.  To quote Avinash Kaushik “All data in aggregate is crap!!”  Without performing any kind of segmentation analysis, looking at the data in aggregate will give you completely erroneous insights.  This is like analyzing Ford’s annual sales in aggregate, finding that the F-150 pick-up truck is its best-selling truck and then marketing the F-150 to customers in New York City or San Francisco.  The problem is that most analytics platforms will not let you do dynamic segmentation with real-time data even today.

 

Marketers need strategic analytics that gives them better insights about business outcomes.  They are looking for answers to questions like the following:

 

  • What are my high value segments of customers?
  • Which products or key messages should be used in promotions?
  • What is the most likely item a customer will purchase next?
  • How do I identify a high-potential prospect?
  • How does a non-purchase correlate to web site navigation (pages visited etc.)?
  • What % of customers who purchased a product have a positive sentiment about the company in their social communities (Likes on Facebook or positive sentiment in tweets related to product/brand or positive/negative sentiment when positing comments or reviews)?
  • Why did my website traffic drop by 20% from a region as soon as I turned off my billboard ad?
  • If I have a $100 to spend on my marketing campaign in a region, how should I distribute that across paid search, email, banner ads, social media and so on?

 

These questions are easier asked than answered.  Just collecting click-stream data or related metadata is not sufficient and, more often than not, it requires correlation of information from other systems like billing systems, or product catalogs, or CRM systems, or even offline transactional databases.  Further, most web analytics platforms are really poor at handling dynamic segmentation beyond the standard canned reports that come with the product and anything beyond that will require additional modules or access to data so you can do it yourself.

 

So the big question is: How do marketers answer these questions in today’s world? 

 

Either they ignore them or they go the other extreme by building large data warehouses, employing large teams of skilled resources and using statistical packages like SAS to build and run pre-defined models.  It takes most organizations about 4-7 days to collect all the data they need from the disparate systems and most of this data is structured in nature.  Now add in another layer or complexity by including customer behavior in third-party applications like Facebook or Twitter or other social communities where the nature of the information in mostly unstructured.  It can take anywhere from a week to 4 weeks to run the models and generate the reports for marketing.  Imagine a high volume retail organization trying to rely on these reports to determine business performance and adjust their campaigns or promotions during Christmas shopping season.  By the time you get the insights your opportunity window is long gone and you’ve lost your customer.

 

The solution is simple.

 

Marketers need is a system that can analyze both structured and unstructured information, scale to handle billions of interactions, and analyze the information in real-time and be able to answer the questions like the ones posed above in real-time or near real-time.  Is it possible to build such a system for marketers that sits between the under-served (tactical analytics) and the over-served (large data warehouses with specialized resources)?  Yes, of course and today, there are a number of small start-ups as well as large companies attempting to solve this very same problem.  Funnily enough, the traditional web analytics players seem to be trapped within the paradigm they invented and are unusually quiet about this topic.  What is going to distinguish the real solutions from the pretenders is going to be the depth of the science applied to the problem. 

 

More on that next time..  Stay tuned..

Leave a Comment

We encourage you to share your comments on this post. Comments are moderated and will be reviewed
and posted as promptly as possible during regular business hours

To ensure your comment is published, be sure to follow the Community Guidelines.

Be sure to enter a unique name. You can't reuse a name that's already in use.
Be sure to enter a unique email address. You can't reuse an email address that's already in use.
Type the characters you see in the picture above.Type the words you hear.
Search
Showing results for 
Search instead for 
Do you mean 
About the Author
Sunil Menon is CTO and heads up Product Management for Marketing Optimization at HP Autonomy. Sunil has over 17 years of technical and busi...


Follow Us
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation