• Data Science
  • Shamli Prakash
  • DEC 05, 2017

AI Adoption In Procurement: 3-Step Journey

Ready to learn Data Science? Browse courses like Data Science Training and Certification developed by industry thought leaders and Experfy in Harvard Innovation Lab.


3-Step Journey of AI Adoption within Procurement

Everyone would love to save and bring down what they are spending, including companies of all sizes and across all industries. However, trying to find savings in a company’s spend pool is often like looking for a needle in a haystack. Sure, there are some low hanging fruits too, but they get consumed quickly, leaving procurement teams with the onerous task of finding ever-newer avenues of generating savings. In a previous post I talked about how this is a complex problem to solve and the role AI can play in addressing the challenges involved.

Taking a closer look at the specific aspects where AI can be a game-changer is important to understand it’s broader value in the procurement context. The use of AI and analytics within the procurement domain can be thought of as a 3-step journey.

Step 1: Building a Strong Data Foundation: An organization’s spend data is fraught with issues — it’s usually spread across multiple incongruent ERP and other systems; it has varied formats, structures and nomenclatures; it often has missing information; even the information that is present is rarely reliable with many manual entry errors that creep in over time. Most procurement practitioners therefore vouch for the fact that the task of creating a robust spend data foundation is extremely challenging.

This is where elements of Artificial Intelligence can come to the rescue. It is important to remember that solutions that merely stitch various data sources together are never going to be enough — bringing them together is the smallest part of the problem. AI driven spend analysis would go many steps beyond it to ensure data is not just tied together comprehensively across sources, but is also cleansed and enriched. Critical enrichment steps would cover things like:

  • Continuous data audits to check for and flag incomplete and inconsistent data
  • Text and Natural Language Processing (NLP) based clustering within fields like items, suppliers etc. to ensure information available for analysis is streamlined
  • Accurate spend classification ensuring each of the millions of lines of transactions are correctly and granularly tagged to the items they refer to

Data foundation is probably the most critical aspect of enabling effective procurement initiatives. As they say what you don’t know, you can’t control. And without an accurate data foundation, procurement simply does not know who is spending on what, where, why and how much!

Step 2: Finding the $$$: Once a strong data foundation is in place the world can be your oyster! Having the right data is half the battle won. The next step is to figure out where the savings and opportunities are hiding within this data. Given the size and complexity of typical procurement data, doing it manually is far from efficient. A combination of AI and domain expertise can make the task significantly more fruitful.

Algorithms rooted in procurement knowledge can continuously parse through a company’s spend data and identify opportunity areas. These areas would lie across various levers, all of which would require different types of analyses and consequently algorithms. Some examples:

  • Rate Dispersion: Across the hundreds of thousands of SKUs being bought, identify the existence and extent of differing unit rates being charged by different suppliers. Ensuring it’s an apples-to-apples comparison is key. Once identified savings capture is quick and easy.
  • Working Capital Optimization: Probably the simplest saving to capture but nearly impossible to get a hold of without advanced analytics in place. Where are you paying too early? Where do you have multiple payment terms for the same supplier? Where are you missing out on early payment discounts? These are some of the questions that an AI driven system can serve up to you on a platter.
  • Contract Compliance: Procurement teams spend a lot of time and effort in ensuring they have effective contracts in place with their suppliers. But once the contract is signed, tracking compliance is a whole other ball game! Your machine army can be deployed here as well. Leveraging analytics can help link together contracts and rate cards with transactional data and flag off any instances of non-compliance proactively — from overcharging to missing committed SLAs and everything in between.
  • Fraud and Anomaly Detection: I can bet that most procurement teams would be overjoyed if they had the ability to identify a fraudulent or anomalous transaction as it happens, rather than retroactively, by which time corrective measures may be difficult to deploy. Again — with AI in your toolkit — this is imminently doable!

Step 3: Predictive Analytics: Finally, once you have a data foundation in place and are able to find savings hiding within it, the next step in the journey can be made — that of leveraging advanced predictive modelling techniques to address complex procurement tasks including the likes of:

  • Demand forecasting
  • Inventory management
  • Logistics planning
  • Purchase price analytics

Parts of the three steps described above might be doable to a limited extent through one-time, manual initiatives (although it would cause much heartburn, I can assure you, by way of hours spent wrangling with the data!). However, the key to getting it right is to be able to do it consistently in a sustainable and scalable manner. That’s exactly where data science can be of immense use.

Procurement has unfortunately often been a laggard in adopting new technologies. However, with the proliferation of AI across so many aspects of an organization, there is no reason for it to sit on the fence on this one. The gains can be monumental, both in terms of bottom-line impact as well as the overall effectiveness of how the procurement function contributes to an organization’s larger goals.

The Harvard Innovation Lab

Made in Boston @

The Harvard Innovation Lab


Matching Providers

Matching providers 2
comments powered by Disqus.