, my LinkedIn inbox is filled with knowledge scientists reaching out.
Identical questions. Identical issues. Is provide chain knowledge science the appropriate transfer?
After 10 years in provide chain knowledge science, together with 5 years writing on this weblog, I’ve developed robust views on this query.
Provide chain is an distinctive playground for knowledge scientists.
Wealthy issues, lovely arithmetic and tangible impacts.
However I’m not right here to let you know what’s greatest in your profession.
On this article, I wish to give an sincere view of the alternatives that excite me and the challenges that frustrate me.
Extra importantly, I’ll present learn how to discover this area your self utilizing the tutorials and case research shared throughout this weblog.
You’ll be able to then check your self to see whether or not provide chain analytics resonates with you.
Why do we’d like Provide Chain Analytics?
What’s a Provide Chain?
A Provide Chain is often outlined as a number of events exchanging flows of fabric, info or financial assets with the final word objective of fulfilling a buyer request.
Factories, warehouses and planning groups use programs to speak and alternate info.
These programs retailer a considerable amount of transactional knowledge in databases that displays the exercise of your total provide chain.
- Warehouse Administration Methods (WMS) retailer all of the inbound (flows coming to the warehouse) and outbound (flows leaving the warehouse) transactions.
- Enterprise Useful resource Planning (ERP) programs can retailer all the acquisition orders (to suppliers) and invoices (to prospects)
- Transportation Administration Methods (TMS) will file all of the shipments leaving the warehouse and notify you when they’re delivered.
As it’s not possible to trace every cargo throughout the availability chain, these knowledge stay the one approach to monitor your operations.
Due to this fact, Provide Chain Analytics has emerged as a technique that organisations use to achieve insights from knowledge related to all processes throughout the worth chain.
We have to exploit this knowledge, however for what?
Descriptive Analytics: Deliver Visibility to Operational Groups
At first, firms want visibility.
In a previous expertise, I labored with a Logistics Director who couldn’t inform me what number of pallets they’d of their largest distribution centres.
That is the place we’re for many firms in 2026.
I admit, this isn’t as cool as machine studying or superior optimisation but.
But it surely’s the place most provide chain analytics journeys start, and the place you may ship fast worth!
I found in my first years as a Provide Chain Resolution Designer that operational groups had been drowning in knowledge however unable to see patterns.
They know one thing went fallacious, however they’ll’t clarify why.
Operations supervisor: we don’t handle to extend our capability of orders ready per day and we don’t why!
In a warehouse storing merchandise for a cosmetics retailer, I used to be requested to help the reengineering of operations to extend e-commerce capability.
Round November eleventh, you’ve gotten an enormous e-commerce pageant in China throughout which volumes are multiplied by ten.
Operations supervisor: we recruit extra operators however it doesn’t influence the capability.
To seek out the foundation trigger, I made a decision to go on-site and observe a shift throughout peak hours.
And I rapidly noticed that many operators had been packed in some alleys of the warehouse ready for his or her flip to choose merchandise.
I assumed that was most likely the foundation trigger, however I wanted a approach to show that and talk it to the operations supervisor.
With this heatmap, displaying the share of orders per storage location, we unlocked the conditions by rapidly understanding the foundation trigger.
They knew that some areas contained excessive rotation SKUs, however to not this extent.
Operations Supervisor: we should unfold the excessive rotations throughout the warehouse to keep away from congestion.
This easy visualisation, which I realized from EDAs in Kaggle, was the start of a extra complicated optimisation examine that I documented on this collection of articles.
Even when this isn’t deep studying or complicated optimisation, by no means underestimate the worth of the right visualisation fixing the appropriate operational downside.
The reengineering examine that began from this easy visible helped us to resume the contract with this buyer and earn a number of million euros.
What in order for you extra technical problem?
Diagnostic Analytics: Information supporting Root Trigger Evaluation
We will now step up the technical complexity.
Let me introduce a technique that turned my favorite playground for provide chain knowledge science: Lean Six Sigma.
Lean Six Sigma is a step-by-step method to course of enchancment that makes use of statistical instruments to validate assumptions.
Why Lean Six Sigma? As a result of it forces rigour by way of statistically backed assumptions.
Operations groups make assumptions each day which can be most frequently based mostly on their expertise, however not often on knowledge, as they’re too deep within the day-to-day administration.
We will help them with statistics utilizing the Lean Six Sigma method.
I first used this method for a undertaking supporting the transportation crew of a manufacturing facility in North America.
The Inbound Transportation Supervisor acquired merchandise from two completely different routes.
- Route 1: with tough street circumstances and heavy site visitors
- Route 2: with no site visitors and a lovely trendy street
An exterior service supplier allocates the cargo to a pool of drivers (D1, D2, D3).
What’s the issue?
When an order is allotted to the northern regional hub, the lead time for the request to be accepted is 35% greater than on the southern hub.
Transportation Supervisor: We assume that drivers keep away from as a lot as potential to be allotted to the north route.
Earlier than leaping to conclusions (and beginning a battle with drivers), we determined to analyse the previous cargo confirmations.
We used cross-validation and the Chi-Squared check to indicate that there is no such thing as a vital proof that the motive force’s allocation is linked to the Hub.
This helped the crew to analyze different potential root causes to resolve the issue.
For extra particulars in regards to the method (and examples with supply code), you may take a look at these brief explainer movies:
With these descriptive and diagnostic instruments at hand, you will discover the foundation explanation for most operational points in warehouses, factories, and planning groups.
- Why do we have now a drop in productiveness with this shift of operators?
- Why are there +20% choosing errors on this space?
- Why do we have now a further day of lead time on this particular freight street?
Answering these questions with statistically backed methodologies may help groups implement motion plans.
Can we help these motion plans with data-backed prescriptions?
If we summarise what we’ve seen up to now, within the steady enchancment journey, you may
- Assist to establish the issue with a fastidiously chosen visualisation
- Talk the insights to operational groups
- Use statistical strategies to seek out the foundation trigger
It’s time now to supply solution-oriented insights to help operational enhancements.
Prescriptive Analytics: Optimisation for Choice Making
The thought right here is to resolve an optimisation downside linked to a efficiency indicator.
It may be:
- Staff leaders who wish to optimise their hiring of short-term employees
- A planner who expects to extend the vehicles’ filling charge
- A warehouse supervisor who wants to scale back cupboard space
Mainly, we wish to maximise or minimise a particular goal operate whereas respecting explicit constraints.
On this weblog, you will discover dozens of examples utilizing linear and non-linear programming to optimise a particular course of.
I’ll use the Provide Chain Community Design downside as an example the challenges chances are you’ll face when conducting this type of examine.
A multinational firm with markets and factories in numerous nations wish to redesign its provide chain community to scale back prices and environmental footprint.
Their provide chain director expects you to inform them the place to open factories to minimise the general manufacturing value.
These are the choice variables in your optimisation mannequin which you could construct utilizing the Python library PuLP.
I confronted many points accumulating enter knowledge and fixing the target features.
What are the challenges?
Certainly, more often than not, the issue isn’t said absolutely and comprehensively.
As an example, in a undertaking for a style retailer, it took us weeks to finalise the target operate.
Certainly, after presenting the preliminary outcomes, we seen that the target of lowering the general manufacturing value (throughout all nations mixed) can have an effect on markets by way of the Value of Items Offered (COGS).
You can’t have eventualities during which it’s costlier to provide in India than within the USA.
That is the place I had the chance to help the shopper in adjusting their enterprise and operational assumptions, drawing on my understanding of the mannequin.
These sorts of strategic initiatives may help you shine together with your abilities as you immediately cope with decision-makers on initiatives that influence profitability.
Our worth-added right here shouldn’t be traces of code, however bridging the hole between enterprise facets and optimisation levers.
For extra particulars, I clarify on this video how I exploit AI with an MCP server linked to Claude Desktop to resolve this downside:
And you can too discover a full case examine on this In direction of Information Science article:
To realize this, you want a primary understanding of the operations coupled together with your knowledge science abilities.
I’ve some content material for that.
What do you might want to begin?
The largest challenge I noticed when managing analysts and knowledge scientists on analytics initiatives was their restricted understanding of operations.
This creates a belief deficit.
Operational groups may even see knowledge scientists as individuals who’ve by no means set foot in a warehouse.
You want operational data to reach provide chain analytics.
To not change into an professional.
However sufficient understanding to speak successfully, body issues precisely, and design options that truly work in follow.
This implies studying the fundamentals: how warehouses and factories function, how transportation networks operate, how stock flows by way of a provide chain.
Right here’s the place to start out.
Be taught Provide Chain Processes with 5-min Explainer Movies.
In this playlist of 40+ short explainer videos, I share a condensed model of my 10 years of expertise as an answer design supervisor and as a logistics efficiency supervisor.
It begins with the fundamentals of warehousing and transportation operations.
These movies will give the fundamentals to know:
- Warehouse processes: receiving items (inbound), storage of products (stock) and transport of products (outbound)
- Transportation Administration: Full Truck Load vs. Much less Than a Truck Load, efficiency indicators and value buildings
The main target is operational and monetary, as this stays a key concern in logistics operations and is my main experience as a Provide Chain Resolution Supervisor.
These movies provides you with the fundamental data to know a lot of the logistics-related case research introduced on this weblog.
In every of those articles, you will discover the answer’s supply code linked to a GitHub repository and an explainer video summarising the case examine.
What are you able to do with that?
I often advise taking part in with the enter knowledge, parameters, and eventualities to adapt the answer to your organization’s downside.
You may be imaginative or ask LLM to generate new eventualities and mess around with them.
Don’t forget that the target is to develop your technical abilities in addition to your operational understanding.
Productise your options for higher adoption.
On this weblog, we wish to construct options that influence operations.
We have to guarantee their adoption with a user-friendly deployment.
Due to this fact, I devoted a number of tutorials and articles to the productisation of algorithms and visualisations.
In my last article, I confirmed you step-by-step learn how to deploy a list simulation software utilizing Python’s Streamlit library.
This method can be utilized for any of the 50+ analytics options shared on this weblog.
Do I’ve ebook recommandations?
Sure!
My YouTube Channel is known as Provide Science, a reference to Wallace J. Hopp’s ebook The Provide Chain Science.
If you happen to like arithmetic and wish to apply it to precise operational case research, this ebook is for you.
What’s subsequent?!
I hope that you’re now satisfied that you’ve every part available to change into a provide chain knowledge scientist whose abilities are valued for his or her influence on operations.
As somebody who not too long ago began my very own firm offering analytics merchandise, I can affirm there’s a want for these abilities.
And we will have enjoyable engaged on these sorts of initiatives!
About Me
Let’s join on LinkedIn and Twitter. I’m a Provide Chain Engineer who makes use of knowledge analytics to enhance logistics operations and scale back prices.
For consulting or recommendation on analytics and sustainable provide chain transformation, be at liberty to contact me by way of Logigreen Consulting.
In case you are eager about Information Analytics and Provide Chain, take a look at my web site.
