Taking Online Analytics to the Next Level, Part 2: Building Your Online Ecosystem


Share on LinkedIn

In part 1 of this series I provided an overview of Google Analytics with Remarketing, a valuable tool set for online marketers that would like to target customers (both known and anonymous) online based on interests. But what if you want to integrate all that rich online data within other sources to build a complete view of the customer and apply those insights in all available channels? This is currently not possible using Google Analytics as data access is limited. What are your options? Below are a few tips to get you started-

Online Data Capture: The traditional definition of online analytics is the capturing of online interactions, into a data store, and delivering insight from that data through a reporting interface. You have a few online analytic solution options that provide native access to the online data including IBM NetInsight (Unica) and Omniture. Those solutions, as well as others, provide access to the raw data so you may incorporate into your full customer intelligence ecosystem. Another option, that continues to gain tracking, is the open source solution PIWIK. PIWIK includes a slick reporting interface and a data consolidation process that rivals the big boys. PIWIK will not scale to huge websites anytime soon (this is not their target market nor should it be) but it is a viable option for small to medium size sites.

Create a link between online and offline sources: You need a process built into your online analytic solution which allows you to track a known / registered users throughout all online behaviors and tie those online interactions to offline data. Several options exist including tracking the registered user ID behind the scenes since you typically use that attribute to adjust the online experience.

Technology: Once the online interactions are captured, and a link to offline sources is available, what technology stacks are available to store and analyze your integrated online and offline data.

  • Database solutions: The key question is data size. As a rule of thumb, if the full data size is 2 Terabytes plus, which is typically driven by online interactions (although a few industries have huge non-online sources including telcos), then a traditional relational database may not work well given the analytic queries required. Options include:

Database appliances: Solutions such as Greenplum, Teradata and Netezza make it easy to store and analyze large data sources. This is typically the most expensive option but for good reason … it is the easiest to get up and running and traditional database skills translate for the most part.

Open Source Distributed Systems: Systems such as MongoDB, Cassandra & Hadoop are quickly gaining popularity and are evolving. These options are cheaper than database appliances but require deeper skills (deeper technology and analytic skills) and thus increased resource costs.

Hybrid: In some environments a hybrid relational database & open source distributed system may make sense. As an example, if your database environment includes many small to medium size sources then only a few huge online sources a hybrid environment could be a good option. Use the distributed system to ‘crunch’ the huge sources, aggregate, and place the results in a relational database with the other sources.

  • Analytic Solutions: Many good analytic solutions exist. Three key factors are your database solution, your analytic requirements, and your internal analytic skills. See this post for additional detail on available analytic solutions.

Making the insights actionable: Once you have a consolidated offline and online database, and have built customer segmentations, how may it be used in all channels?

  • Online: The process is exactly the same as Google Analytics with remarketing. You find segments of customers (known and anonymous) that show interests then you select those customer’s IDs and ingest them into your ad and dynamic content (web) servers so they may be targeted. This process is typically manual at first but may be automated over time.
  • Email and offline channels: Now your non-online channel efforts may be informed by all online interactions and known online interests. And, in many environments, it is the online interactions that best show interest and provide insight into when to reach out to customers on a one-to-one basis. As an example, if a customer is on your site and browsed shoes yesterday it would be best to make sure that they see shoe ads when they visit again and when they do visit again and click on a shoe ad but still do not purchase, an email providing a shoe offer should be sent that same day. With an integrated system this is possible and just as importantly the efforts may be measured and enhanced.

Republished with author's permission from original post.

Roman Lenzen
Roman Lenzen, Partner and Chief Data Scientist at Optumine, has delivered value added analytical processes to several industries for 20+ years. His significant analytical, technical, and business process experience provides a unique perspective on improving process efficiency and customer profitability. Roman was previously VP of Analytics at Quaero and Director of Analytics at Merkle. Roman's education includes a Bachelor of Science degree in Mathematics from Marquette University and Masters of Science in Statistics from DePaul University.


Please use comments to add value to the discussion. Maximum one link to an educational blog post or article. We will NOT PUBLISH brief comments like "good post," comments that mainly promote links, or comments with links to companies, products, or services.

Please enter your comment!
Please enter your name here