How much Central Stock?
Thursday 24th May 2018
Logo A tool

The 'Central vs Distributed Stock' Conundrum

The issue:-

If all stock is central there's none in the shops, so we sell none.

If all stock is in the shops we have no 'resilience' when one shop sells more than expected and another, less. We'll sell less and markdown more.
We'll have 'burned our boats' with the original allocation.

There's a remarkable parallel with the Laffer Curve. 'At 0% tax rate the government gets nothing. At 100% tax rate nobody will work so the the government again gets nothing. So the answer must be somewhere in between.'

The textbooks didn't help, so we set out to establish the answer.

We showed that, for most supply chains it's best to keep some stock at the centre. Stock kept at the centre is available to replenish the first outstation(s) to deplete its stock by over-forecast sales.
Since nearly 50% of sales will exceed forecast (balancing the other 50% which sell less), we 'know' this will happen. Systems and logic which treat the forecast as an exact figure completely miss the point.
"If you want to make God laugh, show Him your plan"
Distribution systems need to respond to (and therefore be designed for) uncertainty.

By contrast, distributing all the stock bets customer service on the accuracy of the outbase forecasts. With (say) 25 outstations served from one Distribution Centre (DC), that takes a forecastable number (the aggregate DC sales forecast by product) and requires that we split it into 25 different forecasts, each of which are 5 times as inaccurate as the original. Taking a tough problem - forecasting DC demand - and making it 25 x 5 or 125 times tougher is pretty dumb.

Keeping some 'stock' centrally is what those masters of uncertain outcomes - military field commanders - do naturally. They keep troops in reserve to send to the unexpected hotspots.
Committing all troops to the frontline guarantees that a few will be overwhelmed while many others will have spare capacity which hindsight shows they didn't need; they can't then redeploy in time.
In retail terms, out-of-stock in one store combines with slow moving (eventual markdown?) in another. Crazy, yet horribly common!

We've always known this; the new tools allow us to demonstrate and quantify it. The proper balance of distributed vs central stock has surprised everyone.

The demo is highly graphic and interactive - it encourages learning by experiment rather than by lecturing.

The generic lessons are:-

  1. Where there's only enough stock to give each shop one item, do so regardless.
  2. Where supply is continuous, stock enough in each shop to meet the target service level and no more.
  3. In all other cases, keep some stock at the centre.

We think we have this problem too. Please contact us

A lot of logistics maths is plausible. Rather less of it is right.
Home | About Us | Showcase | Research | Cases | The Vaults | Tips & Quips | Contact Us …
© SCT 2002 - 2018. Last updated Thursday 20th May 2010