How to Create Omni-Channel Customer Insights

Add bookmark

Data-driven companies worldwide are looking to create reliable omni-channel customer insights from the data they collect. The main question is: How? Although companies are considering and trying different methods, these all have their shortcomings. The truth is, the difficulty is not in bringing unstructured data to a data lake. It is the next step that creates the main bottleneck: creating crystal clear insights from that data lake.


Let us first take a look at the most common ways companies attempt to obtain the omni-channel insights they want. Generally speaking, they take one of the following three approaches to create insights from the data they collect:
1. Extending (web) channel-specific tools with data from other channels. In this approach, they bring ‘external’ data to the channel-specific tools, typically web analytics tools like Adobe Analytics. Some even do this in Google analytics.
2. Use enterprise channel-crossing reporting tools. Using tools like Business Objects, Cognos or Micro Strategy to import multiple data sources.
3. Connecting a visualization tool to a dataset. In this method, tools like Tableau or Qlik are connected to a data lake.
So with this in mind, the first question we can ask ourselves is where the shortcomings of these three approaches are.


The first approach has proven to be not very efficient for the following reasons:
  • Data models: Web analytics tools sore data in data models for quick visualization. This is ideal within the walls of the web analytics data silo, but it makes the data unusable for other causes because you cannot take it out of the data model properly without losing or polluting your data.
  • Processing time: Most web analytics tools need a day or more to process data. And extracting data from the tool will take even more time, making quick responses impossible and leaving customers unsatisfied.
  • Poor flexibility: To really optimize your customer journeys, you want to stay flexible when it comes to the data you collect. Web analytics tools do not offer this flexibility, as the data collection model is fixed.
  • No data ownership: Most analytics tools save data at their side, which means that you lose data ownership and cannot guarantee your customers’ privacy.


When looking at option two, the possibilities are far better than with option one. However, these tools come with big investments. Full implementation costs can run up to a six figure number, which is quite a threshold for many companies. Furthermore, the flexibility and accessibility often leave room for improvement. Therefore, many companies move to option three.


In this approach companies focus on creating a so-called data lake. Often, this is created by exporting data from their analytics tool and several other systems. This mostly is raw, unstructured data. Since data storage is relatively cheap and BI departments often have a visualization license in place, this option is more cost-efficient than option two.
However contrary to popular belief, connecting the data to a visualization tool like Tableau or Qlik does not deliver the results these companies are looking for. The visualization tools are optimized for data visualization and not for processing data efficiently, which leads to long loading times for reports and very slow interfaces. That is a shame, because the visualization capabilities are excellent. The purpose of the data lake is to contain all data and not to optimize it for a specific use. There is a gap between the unstructured data in the data lake and the visualization tools expecting a dataset to be delivered efficiently.


But do not write off these ‘option three visualization tools’ just yet. Because they are in fact part of a proper solution. If you can bridge the gap between channel-specific data dumps and these tools, you can visualize omni-channel data in an agile and flexible way. Luckily, there is technology that can do so: the DimML data science language. If you know what you want to visualize you can very easily ‘prepare’ the data you want to use by putting it in the format your tool needs. For instance if you want to combine a page name to a client segment, you can create a table with that information, which can very easily be read and visualized by Tableau or Qlik. By adding a prepared data component you will have the extensive possibilities of the data lake while maintaining the excellent reporting capabilities of the visualization tools. The intermediate, flexible component will give you the missing link in your omni-channel customer insights pipeline.


In our experience, this is the right approach if you want to create a flexible and agile solution that delivers reliable omni-channel customer insights for better understanding and decisions. You have the opportunity to connect, enrich, prepare and even reprocess data for practically every business intelligence visualization tool.
Get in touch with Ronald via: