Writing the results
All articles

Toward data driven and strategic procurement – Parsing purchase data for your spend analysis

Purchase orders (PO) are one of the many valuable tools in procurement. They add a level of clarity and formality to the buying and purchasing process, forming a bidding contract between the parties involved.

While there’s a whole host of elements that warrant attention when maintaining an efficient and error free purchasing workflow, here we’re going to focus on the data in your purchase orders (everything you can extract and gather from the orders you issue).

In this article, we'll discuss the importance of purchase data extraction and why it’s essential to analyse your purchase orders. Then we'll look at how automation and better technology can help you make the most of your data, to drive better procurement decision making.

But first, let's start with the basics.

What is a Purchase Order?

The Corporate Finance Institute defines a purchase order as: “a commercial source document that is issued by a business’ purchasing department when placing an order with its vendors or suppliers”. It binds the buyer to a promise to pay the seller for designated products at a future date.

Purchase orders are beneficial to both parties involved. On one hand, they help vendors guarantee the future safety of their cash flow.  On the other hand, they gave buyers control over the quantity and specifications of products bought. And they also enable purchasing departments to keep track of their spend, to use later for bookkeeping or auditing purposes.

Purchase orders vs invoices

Purchase orders and invoices are closely connected. They are  two separate steps in the same process. The purchase order is issued by the buyer and sent to the vendor, detailing the goods/service specifications. The vendor, in turn, sends the invoice after approving the purchase. The invoice details the products/services delivered with all due dates and payments.

What kind of data does a Purchase Order contain?

While there are various types of purchase orders, they generally contain similar details and should reflect the following purchase data:

  • PO number
  • Purchase order date
  • Vendor name and billing address
  • Buyer name and shipping address
  • Contact information
  • Item name
  • Item description
  • Item quantity
  • Item unit cost
  • Line total
  • Taxes
  • Total price
  • Payment terms
  • Delivery date
  • Shipping method and terms

What’s the value in analysing purchase data?

According to a survey by Deloitte, analytics is now considered to be one of the most powerful disruptive forces when it comes to procurement. One important use case which has been increasingly applied in procurement analytics, is in spend management—specifically, spend analytics.

Spend analytics can be a form of descriptive analysis whereby historical spending behaviour and trends are described using data visualisation tools for instance. It can also be in the form of predictive analysis whereby purchase data are collected and analysed through predictive models to aid future decision making. This is conducted to forecast future procurement costs; identify saving opportunities; improve strategic sourcing and make appropriate alterations to the company's overall supply chain strategy.

How to analyse purchase data and perform spend analysis?

Typically, the process of conducting spend analysis encompasses the following steps:

  1. Determine the purchasing departments and business’ objectives
  2. Identify purchase data sources e.g purchase orders pdf documents
  3. Collect and extract the purchase data using document parser
  4. Clean data for errors and categorise for consistency e.g spend category and order frequency
  5. Analyse data through the appropriate models and techniques e.g Time-series forecasting (ARIMA) or Decision trees etc.
  6. Report on insights, KPIs and actionable improvements.

Why is PO data extraction a problem?

There are technical elements to conducting an effective spend analysis, of course, but they’re all rooted in sufficient and good data collection. As stated in the Deloitte Chief Procurement Officer Survey: “no matter how many analytical capabilities an organisation adds to its armoury, bad data often translates into bad analytics”. The chances for bad data are especially high if you manually extract data from purchase order files either scanned or electronically generated, through PO systems.

When dealing with native PDFs (electronically generated), copy and paste will often do a decent job extracting small amounts of information. But when dealing with a large volume of files, such a manual process is not only resource-intensive but also prone to errors.

This could be even more problematic when dealing with paper-based scanned PDF purchase orders - which are still commonly used in business workflows. You won’t be able to copy paste and you will be forced to collect and enter the purchase data by hand into your Excel sheet or database. This is because data in scanned documents are stored as images with embedded texts and tables. They carry no markup nor character level data or hierarchy, resulting in trapped and unorganised unstructured data.

The solution? Eliminate human error and save time through purchase data extraction automation tool like Parsel.

Why choose Parsel to automate your purchase data extraction?

When investigating which technologies are used by high performing procurement departments, it was found that successful teams are 10x more likely to have fully deployed predictive analytics capabilities and are 18x more likely to have fully deployed Artificial intelligence and cognitive capabilities.

At Parsel, we use cognitive data extraction - intelligent data capture algorithms that use AI and machine learning technology to understand the information it is extracting, and categorise it into key-value pairs, tables, and entities.

Parsel's data inference algorithms capture the relevant data from different purchase order formats and layouts. It then converts the data from PDF to Excel and JSON structured data format to later use for your data analytics.

It’s all done online with no installation or manual guidance needed and at 96.6% financial accuracy - higher than any other data extraction tool on the market right now.

Automate Your Purchase data Extraction With Parsel

Want to see how well Parsel cognitive data capture can fit into your supply chain/procurement analysis? Try it for free by signing up today or get in touch to learn more about our Parsel API.

With our Enterprise plan, you can benefit from a fully featured offering, with unlimited monthly page allowance, API access, custom ML models, and direct support from our team of dynamic data scientists and engineers.