How machine-learning can make procurement decision-making easier – tips from Amazon

National | Retail | Technology

Author: Dave Brittain, Head of Amazon Business UK

‘Machine-learning’ is a buzzword most business leaders are familiar with. Whether you are running a large or small business, you know that MI is something you probably need to be on top of and that, potentially, it could have an impact on efficiencies and your bottom line.

But the truth is when you scratch beneath the surface, many CEOs don’t know how best to take advantage of it on a day to day basis… particularly when it comes to procurement.

In the past, when it came to evaluating procurement data, companies would need to invest in experts such as business intelligence (BI) engineers, data scientists and IT professionals who would create complex analysis models from the data. Today thanks to MI you don’t need to be an expert to take complex data and build narratives to make informed decisions – it’s much more straightforward and accessible.

Let’s say you wanted to evaluate the order history data of thousands of employees to make purchasing decisions. Using some online purchasing solutions, you can create a simple narrative to give top management a senior level overview – without first having to analyse complex graphics or decipher data tables line by line. You can also drill down to get much more detailed information at a granular level.

Essentially, there are two other ways Amazon Business allows you to break down this data from your order history using machine-learning.

  • Aggregated visualizations offer the possibility to gain deeper insights, as the data can be filtered by unique patterns or specific details.
  • Raw tabular data allows a very detailed analysis of the granular information on which the more general narratives and visualisations are based.
Dave Brittain, Senior Manager of Amazon Business.
Dave Brittain

I’ve outlined these options in more detail below:

Aggregated visualisations

The reason aggregated visualisations are important is that the majority of purchasing managers need more than simple narratives in order to be able to understand the purchasing behaviour of their employees exactly. For this group, a more detailed dashboard with extensive KPIs, diagrams, drilldowns and filter functions is suitable.

For example, if a manager has a Guided Buying policy in place for his company, the dashboard can be used to check whether and to what extent these policies are being adhered to and how they affect the purchasing behaviour of his employees.

Expenditures can be broken down by category, preferred and restricted items, users and groups, and viewed at a glance in a bar chart.

A purchasing manager can even trace non-compliant purchases back to the responsible employee and thus further optimise his procurement management.

Tabular raw data

While simple narratives and visualisations are usually sufficient, there may be situations where financial or procurement experts need to analyse the data very deeply and in detail in order to get to the bottom of a particular issue.

For a full investigation, users can get deep into the analysis by first setting relevant filters and then accessing the data in more detail until they finally reach the tabular raw data on which the visualisations and narratives are based, and which are relevant in this context. If you don’t want to take the detour via the filter system but want to analyse a specific graphic in more detail, you can export the raw data directly from it as a CSV file.

Final word on MI

The key thing to remember here is that when it comes to making informed decisions, data can be key. Being able to access that data quickly and efficiently doesn’t mean you have to hire a team of data scientists, there are simple products and solutions on the market designed to make your life easier.

Did you enjoy reading this content?  To get more great content like this subscribe to our magazine

Tagged with: ||

Reader's Comments

Comments related to the current article

Leave a comment

Your email address will not be published. Required fields are marked *