2. Overview
Detailing 6 key measures across two datasets of a
fast moving consumer goods company.
Dataset 1: Material Master
Ensures accuracy.
Containing Master Data about Packaging Items, Raw Materials and
Finished Products.
Dataset 2: Project Lifecycle Management
Data regarding the product end of life, unattached assets and data
storage management.
3. 3
1. Extraction
Data is first extracted from the ERP system.
Tables may be joined at this stage.
11. Published to Power BI
Pivot Charts (and accessible data model with all
relationships) published to Power BI
https://app.powerbi.com/groups/me/dashboards/56009dfe-4622-
4283-8e1d-4356148190c8?ctid=1a1ab5ab-e103-41b7-ba05-
cc1841979969
15. Alteryx
Folder of files loaded in one step and transformed
en masse
All sheets loaded from all files
Rows and columns cleansed
Columns renamed, metadata set, redundancy removed
Output as Tableau Data Extract and Excel file
Next month, just drag another file into the folder!
Without Master Data, no transactions can occur!
… Therefore data quality is paramount. A recent study showed that the average company loses 12% of revenue due to inaccurate data.
2 Datasets:
1. Data Quality : As a simple example, if the barcode doesn’t match GS1 rules, then large fines will be imposed. If the dimensions don’t match the GS1 rules, then a given supermarket’s online presence may not even list the product.
2. Product Lifecycle Management : PLM is about managing all the information about the product, from a new idea to when it hits the market until it is retired. Here we are focusing on the data quality relating to the retiring of products.
At Unilever, to start, I liaised with Accenture who I discovered would build custom transactions that would report from our SAP systems, however the extent of this was only really to join tables and download.
As you can imagine, the average employee in a supply chain function (where the creation and co-ordination of Master Data occurred) could not easily ascertain where he / she should focus their efforts in order to improve data quality.
I wanted to make this easier for my colleagues and equivalents, so I used the Excel Data Model, which, unlike a traditional Excel workbook, can handle huge amounts of data. This has been cited as the best thing to happen to Excel in 20 years.
Certainly it is very powerful when it comes to building relationships between data sets and making enterprise scale data analysis far easier.
As the tables of data are loaded into the Data Model, you can see on the right that you have the ability to cleanse the data / change the metadata and create calculated columns on the way in.
Excel provides a very easy way to create relationships inside the data model with its diagram view. Here you simply click and drag the fields to create joins.
Pivot tables can be created using Excels data Model in the same way they can be created in a standard Excel workbook. However, not only can much larger data sets be handled, but you have two extremely powerful benefits :
1. The ability to create calculated measures and new fields using the built in DAX language .
2. Being able to use the data across all tables in the data model once relationships are created.
Finally I created a selection of easy to view Pivot Charts, such that people could see directly which errors they had, and therefore how to resolve them.
In order that all relevant parts of the business could see these charts, these were then published to Power BI.
Power BI creates intuitive and interactive dashboards.
However there was a major problem. When the business next wanted to measure the data quality to see where the improvements were, you had to go through all these steps again…. Adding new data to the Excel Data Model, cleansing, transforming, creating relationships and measures etc,…
After leaving Unilever, in order to pursue a career in the Data Analysis / Business Intelligence space, I decided to look for a better solution. Automation was at the forefront of my mind.
After some research, I discovered Alteryx and Tableau.
Alteryx is a leading platform in data analytics that can prepare, join, and analyse data in an easy to understand visual format.
Tableau is a persuasive tool for being able to see their data, gain insights and ask great questions.
Here we see a simple workflow in Alteryx. As the workflow is created, tools are added, such as “Data Input”, which can accept a huge range of sources, “Filter”, “Formula”, “Join” and “Union”, for example - before outputting in a large range of formats.
To re-iterate, the joy of this approach is that this workflow will parse all the data from all the files in a given folder. So when next period for measurement comes, the downloaded data from SAP only needs to be dropped in a folder, saving a huge amount of resource against the previous approach.
Tableau is another self service Business Intelligence tool, effective for being able to allow users to see their data, gain insights and ask great questions.
Like Power BI, Tableau creates innovative and interactive visualisations. With the emphasis again on self service Business Intelligence, end users are able to very simply drag the fields onto the axes, or data area in order to get the visualisation they require.