Customers often ask us about Predictive, seeing it as ‘Panacea’ or a ‘silver bullet’ to all online marketing problems.
In fact, the first investment and development area may need to be in analytics, not predictive.
There is a perception that data warehousing and analytics is used by Business Intelligence (BI) tools simply to explore and visualise data, rather than inform any prediction initiative.
In this blog, I will elaborate on this from a technical point of view.
The role of data warehousing
Quoting from wiseGeek,”Data warehousing combines data from multiple, usually varied, sources into one comprehensive and easily manipulated database”. The emergence of big data puts further burden on a data-warehousing platform to be both scalable and fast.
The role of BI
On the other hand, BI mainly refers to the transformation of raw data into meaningful information using various techniques and technologies, for the purpose of making business decisions and developing marketing strategies, etc. These tools mainly offer a variety of visualisation and reporting features.
Analytics vs Predictive analytics
While analytics explores profiles historically (for example: to discover trends, patterns, anomalies and identify clusters of profiles with common characteristics), predictive analytics aims to understand how customers are likely to behave in the future and how they might react to certain events. These patterns may indicate new business development opportunities or reveal hidden problems.
However, neither data warehousing itself nor most BI tools gives you the capability to discover these trends, anomalies or to predict future user behaviour. Even if you know what you are doing, the need to write complex algorithms in BI tools will be a major obstacle.
Does predictive analytics need an additional analytics layer?
Most BI tools are not designed to perform predictive analytics or any other machine learning algorithms. BI tools are really good at their primary functions (reporting and visualisation), but their data exploration capability is limited to the underlying SQL engine. They need to an extra ‘layer’ to perform analytics.
Given the wide range of analytics – especially predictive analytics – methods, and considering their complexity, BI tools don’t seem to be that eager to change their focus and to develop into a proverbial Swiss Army Knife.
Profile Analytics is here
At APSIS, thanks to funding from Vinnova, we are adding a layer to Profile Cloud specifically to allow users to perform analytics across all the Profiles that are stored for their Profile Cloud instance.
Profile Analytics is designed to plug the gap between BI tools and predictive analysis, while providing further invaluable insight into a brand’s customer data.
You have to look backwards to move forwards – using analytics is the only way to identify unprecedented patterns and trends needed to drive real prediction.
This post was written by Shahab Mokarizadeh