| Jul 11, 2016

3 MIN READ

Written by Ashnik Team
Data Pipeline and Analytics

Landing Page Analytics using Lambda Architecture (λ) in E-commerce Applications

What is Landing Page Analytics?
Landing Page Analytics refers to “predicting user’s behaviour” in a web-facing application, with respect to various interactions and activities on the website – view, search and form submission etc.
Today with digitization it is essential to understand user behaviour to improve the business. This is even more important for business with e-commerce offering.
Landing Page Analytics helps businesses to

  • Identify key demographics of potential customer e.g. country, age group, language etc
  • Understand user behaviour and products/categories users are interested in
  • Understand buying nature e.g. specific products are grouped together while buying

Organisation needs this information to identify the current market trend and the customer’s buying behaviour. It will also give the insight to understand current business and possible areas to improve the business.
A Typical Landing Page Analytics Use Case
On an e-commerce site, it is usual for a user to look for a specific product let’s say a Smartphone. During the Purchase action, the user should be offered a list of potential products he/she may opt to buy. The user should also be presented with promotion details for relevant products/options. This all should happen with minimal lag, else the users would lose interest and will be drifted to another site.
Here Landing Page Analytics can be helpful to analyse and get related products and promotion details from data source with the minimal response time. Lambda Architecture provides an efficient way to run this analytical model within the acceptable time interval.
Lambda Architecture ( λ )
Lambda architecture is a data-processing architecture designed to handle massive quantities of data by taking advantage of both batch-and stream-processing methods in Big Data environment. In our example of e-commerce, this architecture will help to achieve Near Real Time analysis to ‘predict the list’ of products user would be interested in.
Lambda Architecture ( λ ) in e-commerce
tamil blog image
A logical architecture looks something like what I have depicted in the diagram. Various components of this architecture are
User Action:  A user action for example in our case the ‘purchase’ action will be fed as an input to analytical model. This input will trigger Both Batch Query and Stream Query to execute analytical model.
Batch Query ( Map Reduce) : Once the analytics model  gets the input, it starts analysis in Batch mode (a typically Map Reduce algorithm) as parallel process.
Stream Query: On receiving the input the analytical model would get triggered in Stream mode as parallel process. In General running the analytical model in stream mode will consume more resource in order to ensure timely completion of the job.
Consolidated Result: Both Batch Query and Stream Query will update the result in a common place to consolidated the results. Later this will be used to make suggestion to the user.
Suggestion:  The user is presented with the final suggestion list based on the results consolidated from Stream Query and Batch Query. The suggestion will be published to the user before it’s relevance is lost. e.g. in our e-commerce example it can be used to suggest product alternatives or additional accessories for the smart phone before the user makes the purchase.
Almost all the web facing application, e-commerce sites and digital platforms needs do Landing page analytics to predict Customer’s behaviour. Lambda architecture is the tool to facilitate the analysis job to run in near real time. There are various components involved in Lambda architecture as explained above and each of them needs to be covered with varied technology/platform. In one of my next blogs I would try to pen down how to choose right fitting technology for setting up Lambda architecture for doing Landing page analytics.
Sundaramoorthy Tamilarasan I Senior BI Consultant, Ashnik 



Go to Top