Is Autonomous Sourcing a Pipe Dream?

Archive | Upgrade to Premium | Advertise

 📰 In Today’s Edition…

Hi there,

Tonight I’m:

• Sharing a free Accounts Payable Automation eBook (sponsored)

• Testing out a new Procurement job board

• Testing out a new Key Takeaways Section (in case you’re in a rush this week)

• Giving you the highlights from the first Pure Procurement fireside chat on Autonomous Sourcing. Can you really entirely remove the buyer from the sourcing process? Let’s find out.

Have a great week ahead.

Best,

Joël

P.S. In case you missed it, here was my most popular LinkedIn post of the week:
2 things I wish people had told me about Procurement

📋 Job Board

Here are a few job openings that caught my eye this week:

MacLean-Fogg is looking for a Purchasing Manager
Full Time | On-site | Chicago, Illinois

GE Aviation is looking for a Lead Sourcing Specialist
Full Time | Remote | Cincinnati, Ohio

Vistar is looking for a Merchandising Manager
Full Time | On-site | Boston, Massachusetts | Salary: $100K

HNI Corporation is looking for a Supply Chain Analyst
Full Time | On-site | Muscatine, Iowa | Salary: $60-85K

🔑 Key Takeaways

No time to read the full 🌙 Sunday Night Note? Here are the key takeaways:

1. A Blanket Approach to Autonomous Sourcing Won’t Work. You need to select ‘low hanging fruit’, SKU-based categories to get started.

2. Bad Data Quality and Bad Change Management Will Kill Your Benefits. As with most ProcureTech implementations, we don’t spend enough time on ‘how’ we deploy, thinking technology will be the saviour.

3. Beyond the Gen AI Buzz. The sweet spot for Gen AI in Sourcing is emerging → Sourcing Decision Support.

4. Autonomous Sourcing Currently Falls Flat at Both Ends of the “Sourcing Spectrum” → Hyper complex/strategic sourcing event and in the “Tail of the tail”.

Read the full note below for all the juicy details.

🌙 Sunday Night Note

On Friday, March 15th 2024, I hosted the first edition of Pure Procurement Fireside Chats: Conversations with ProcureTech Leaders about the True Capabilities of Procurement Technology.

The first chat was with Erin Mcfarlane, a leader in autonomous sourcing technology.

Autonomous sourcing is an organization’s ability to automate the sourcing process for using business rules and technology, removing the need for manual involvement of a buyer.

4 Insights on The State of Autonomous Sourcing

1/ A Blanket Approach Won’t Work

If you’re looking for ways to increase the throughput of your sourcing team, automation is the first place to look. However, you shouldn’t apply a blanket approach to all your purchasing categories. That’s a recipe for pain…

Autonomous sourcing is most effective and easy to implement in purchasing categories where requirements are clear, competition is healthy and evaluation of bid results isn’t too complex. The closer you are to “SKU-based purchasing” (e.g. MRO parts, IT hardware, etc.), the better since:

• Your requirements are not open to interpretation

• It’s possible to define clear business rules around:

  • Supplier selection using your purchasing history, preferred suppliers, etc.

  • Bid award (e.g. pick the lowest price that respects the “required by” date. Throw an exception if no bids respect this rules.)

That’s not to say you can’t automate large portion of the sourcing process for other purchasing categories. It just means you should start with the low hanging fruit, gain expertise and build more complex processes/rules from there.

2/ There are 2 Critical Success Factors for Autonomous Sourcing

Even when targeting “widget-type”/SKU-based purchasing categories, you can still fail to automate… Why?

1. Bad Data Quality. Imagine a requisition with a valid part master/description, preferred vendor master and “required by” date listed. Now imagine a requisition that says “see quote”. Which one is easier to source?

   For a human, the difference may be negligible if the quote has all the info. However, for a system? The first one is infinitely easier to source because you can build rules on the “structured data” (data in their respective fields in the proper format).

   • For requisitions/purchases with “structured data”, good data quality therefore means putting good data management processes in place for:

     • Vendor masters

     • Part masters

     • Purchasing commodity codes

     • Average lead times

     • Min/Max inventory levels

     • Etc.

     Business rules can easily be built on top of good structured data to select appropriate suppliers, and automatically award bids.

   • For requisitions with “unstructured data” (e.g. text entered manually by a requester), this means putting mechanisms in place to ensure high quality requisitions are submitted.

   • Investing in data governance will more than pay for itself with the benefits you gain not only around automation but around all the other business capabilities that good data enable.

2. Change Management. People hate change. However, it is the only constant in our lives… When you switch from manual sourcing to automated sourcing for given categories, there is always push back…

   Having a solid change management plan to proactively counter and address this resistance is a key success factor.

   During our chat, Erin mentioned that the data tells her there’s a “3-month adoption hump” for automated sourcing tools.

   Requesters and buyers initially resist the new process/tool because it is uncomfortable… There’s a learning cuvre. But after 3 months, when you ask them if they would go back to manual sourcing, they say: “NO WAY!”

3/ GenAI Has a Concrete Role to Play to Optimize Sourcing

Once you’ve covered all the widget/SKU-based "low hanging fruit”, purchasing categories with more complex sourcing processes can come into play for partial automation with the help of GenAI. Let’s refer to this as Sourcing Decision Support (you’ll see what I mean by that below).

Erin gave the example of parking lot resurfacing (Facilities purchasing category) to illustrate. This isn’t a strategic direct category purchase with a complex strategy… But these types of purchases can quickly add up to millions of dollars and are worth sourcing.

The complexities with sourcing these types of purchases come from the fact that they aren’t purchased repeatedly and that requirements are different for each instance of purchasing. Let’s illustrate with parking lot resurfacing:

• It’s difficult for the Requester/Buyer to know what information should be provided to vendors:

  • Are you ripping up old material?

  • What are the current surface conditions (damage, cracks potholes, etc.)?

  • What material do you want to use for the new surface (concrete, asphalt, etc.)?

  • What’s the square footage of the resurfacing area?

  • What patterns/lines needs to be painted on the new asphalt?

  • Do you have any specific quality requirements for the surface (e.g. high traffic or heavy load areas)?

  • Do you have the required permits/insurance to complete the work?

• It’s difficult for the Requester/Buyer to know what questions to ask and/or compare bids once they are received:

  • How can we compare the bids if they are returned in varying formats (without having to engineer an Excel comparison grid)?

  • Which bid responses/answers should trigger further questions?

• Once you’ve gone through the RFQ process, you still want to integrate the above into your company’s contract templates. This can be a long process…

  • The Buyer/Contract Admin works to fit the ‘square peg into the round hole’ fitting ‘parking lot resurfacing’ into your general Master Service Agreement and Statement of Work templates…

A GenAI supported sourcing tool can help speed up and improve all 3 of these steps:

1. You input your general requirements (parking lot resurfacing) and the Gen AI tool returns a general set of information your should gather and include in your RFQ document.

   The Gen AI tool can also generate the RFQ document with the information provided based on your company RFQ template.

   The Buyer reviews, corrects and perfects the document before sending it out to market.

2. Once you receive responses/bids, you feed them into your Gen AI tool asking for a summary of the key points and main differences between bids. You also ask for what clarifying questions the Gen AI tool suggests to further increase the clarity of the bids.

   The Buyer/Selection Team uses this information as a starting point to evaluate the bids.

3. Once you decide who to award the RFQ to, you use the GEN AI tool to generate a context-specific MSA ans SoW using your company templates. You use this as your starting point for the contract draft (you could also do this earlier in the process to send out the draft with your RFQ as part of step 1).

You’ve just used GenAI as a Sourcing Decision Support tool! And, probably more than halved the time required to run this parking lot resurfacing event…

You can make this above process work today with a Chat GPT license… You feed the Large Language Model (LLM) your RFQ, MSA, SoW templates and vendor responses and you can get to the above..

However, the current problem is Data Ownership and security, as I outlined last fall.

The exciting part?

The ‘fix’ is on its way… I know a few ProcureTech solution providers are working on this problem and getting close to the fix.

4/ Where Autonomous Sourcing Falls Flat

Both ends of the “sourcing spectrum”:

• Super strategic, hyper complex sourcing events with 12,000 conflicting variables to consider, including strategy → Humans are still better at this.

• “Tail of the tail” sourcing (e.g. a one-time 10$ hammer buy). To quote Erin: “who cares about your 10$ hammer” → Best to just optimize the purchasing rules and channels around these types of purchases depending on context (e.g. catalogs, PCards, P2P tail spend solutions, etc.)

There you have it… Some of the key points from the first Pure Procurement Fireside Chat. I hope you’ll join us for the next one in Q2.

Stay tuned to the newsletter for the details. I’ll share them here when available.

Thanks to Martin, DeWayne, Cedric, Shaun for keeping the discussion so lively last Friday.

And a BIG thank you to Erin for being the “guinea pig” for the concept.

Onwards!

💭 Quote of the Week

📊 How Did I Do This Week?

Best comment from last week:

Yes! Sustainable procurement is easier than you think it is.
Read last week’s note to see how. Thanks Joanne!

Vote and leave a comment to get featured next week.

🎁 Earn Prizes

If you 💗’ed today’s edition, share it with another digitally-minded Procurement professional and earn prizes.

You currently have 0 referrals, only 1 away from receiving the Monthly Top 5 Link Archive.

What is the Monthly Top 5 Link Archive?

Share your personalized link with others:
https://newsletter.pureprocurement.ca/subscribe?ref=PLACEHOLDER

That’s it for today.

When you’re ready, here are 3 ways I can help:

1) Pure Procurement Premium Content

For $80 CAD/year, you get access to:

• My 30+ Deep Dive guides to help you get your Procurement digital transformation initiatives right the first time.

• All the templates in the Pure Procurement Store.

You can try Pure Procurement Premium free for 7 days.

2) Sponsor this newsletter to get in front of 7000+ digitally-minded Procurement professionals.

Reply to this email and I’ll send you all the details.

3) Get on the waiting list to work with me on your digitalization initiatives

I’m typically booked at least a few months out. Reply to this email if you’re planning a procurement digitalization initiative and I’ll tell you what I can do.