What’s a Heuristic in Supply Chain Problem-Solving?

A heuristic is like a street-smart shortcut your supply chain takes when the perfect answer would take way too long to calculate.

Instead of crunching every possible number, a heuristic says: “I don’t need perfection — I need something good fast.”

In supply chain terms, it’s a practical rule of thumb that helps you make solid decisions when you’re juggling trucks, warehouses, inventory, and deadlines, without melting your brain (or your computer). Think of it as a clever hack for messy logistics puzzles.

If you’re talking inventory with anyone in supply chain, there’s one heuristic that instantly makes you sound competent: ABC analysis.

A simple idea from the mid-20th century; rank products by revenue, set a basic service level, and move on. It’s a quick way to decide how to carefully manage each item.

Interestingly enough, this simple approach  still runs billion-dollar supply chain networks in 2025.

With all our data, machine learning, AI and compute power…
Why does a 70-year-old shortcut still dominate the conversation?

Setting Up The Experiment

To play with the idea, I built a small experiment:
1,000 fake products with realistic demand, lead times, and costs.

Then I compared two approaches:

ABC analysis heuristic: Top 20% of products by revenue get a 99% service level, the middle 30% get 95%, the rest 90%.

SKU-level optimization: Newsvendor is a math problem solved for ordering just enough so that your chance of selling out is perfectly balanced with your fear of leftovers. In this method, I use the Newsvendor formula to pick an “optimal” service level for each individual SKU which balances holding cost (0.25 annual carry rate) against stockout cost (set to 2× the product margin). 

I expected SKU-level optimization to hold less inventory, raise service levels, and reduce total cost.

It did.

Newsvendor Wins!

The Newsvendor approach didn’t just edge out ABC; it crushed it.

Higher service levels. Roughly 60% lower total cost.

A single example GIF below shows the pattern:

Newsvendor carries a bit more inventory; costlier than ABC.
ABC hits a stockout around Week 30.
That one miss reverses ABC’s cost advantage.

Service levels: 100% for Newsvendor, 98% for ABC.
Newsvendor has lower overall cost and better service level.

Huge Problem Though…

Those inputs I used
(a 0.25 carrying cost and a stockout cost set to 2× margin for all products)
are broad assumptions.

If I change the stockout cost assumption, the SKU-level advantage shrinks.
If I change it enough, it disappears entirely.

We Just Need to Correctly Estimate Stockout Cost Right?

Researchers have tried.
The work is clever, careful, and deeply specific to each setting.

You can get a number.
It’s just rarely the number.

In one business, a stockout creates a $13 short-run loss and a small long-run dip.¹
In another, it depends entirely on whether a near-perfect substitute sits two inches to the right.²
Online, it shifts with delivery speed, transparency, promotions, and even the mood of the checkout fee.³

The methods are heavy. Event studies after stockouts, survival models for repeat purchase, MDCEV and choice models for substitution.
And they all drift over time.

Once you estimate it, you have to maintain it.
New promos. New packaging. New competitors. New SLAs.
Your beautifully calibrated number starts aging immediately.

So yes, you can build a better engine than ABC.
But it needs careful fuel and regular maintenance.

Most teams don’t have that budget; time, data, or attention.

Meanwhile, ABC runs on autopilot.If the goal is simply to get in the ballpark, ABC gives you most of the value for a fraction of the effort.

Why We Fall Back on Heuristics

Heuristics stick around because they’re light.
You pick them up, use them, and keep moving.

Reading those papers on stockout-cost estimation, you can feel the weight of it all.
Data, modeling, re-estimation, drift.
It’s simply too much.
Suddenly the “better” method feels heavier than the problem you were trying to solve.

Heuristics survive because they’re cognitively efficient.⁴
They give you an answer quickly enough that you can keep moving.

The point isn’t “don’t try to be precise.”
It’s that precision only wins when its marginal benefit exceeds its marginal cost.

The interesting question becomes:
What would have to be true for a precise method to honestly be worth it?

The decision to use them would need to be as light as the heuristic itself.

In inventory management, we're not there yet. So conversations drift back to ABC.

ATTRIBUTION
¹ https://www.kellogg.northwestern.edu/faculty/anderson_e/htm/personalpage_files/Papers/Measuring_and_Mitigating_the_Costs_of_Stockouts.pdf

² https://www.dii.uchile.cl/~amusalem/OOS%20paper%202010.pdf

³ https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4524615

⁴ Lindy.