Improving Supply Planning - A Demand Driven Journey

© British Telecommunications plc
BT Supply Chain
Brian Dooley FCILT, CDDP, CDDL
Senior Supply Chain Planning Manager
BT’s Demand Driven Journey

A bit about us …

Agenda
• Our journey so far
• The case for change
• Selling the solution
• Results, learnings & next steps

Our starting point
Our transformation journey so far…
• Inventory management and supply chain planning decentralised
• Separate teams within each Line of Business
• No Pan BT standard approach or reporting

Our starting point
September 2014
transformation
programme launched
• Design new operating model for Pan-BT inventory management
• Create Inventory Management Centre (IMC) – responsibility for inventory
planning and execution moved to BT Supply Chain
• Develop insightful inventory reporting

Our starting point
September 2014
transformation
programme launched
The first 18 months …
• Completed design and development of new operating model
• Demand forecasting process
• Safety stock calculation
• Demand & supply balancing through S&OP
• Design and re-organisation of IMC team
• Implementation of new IM processes across the lines of business

Our starting point
September 2014
transformation
programme launched
Achievements to
December 2015 …
• Stable and standardised processes and controls in place
• Reduced inventory holding by 15%
• Systems in place for demand forecasting & inventory optimisation
• Introduction of formal S&OP process

• Forecast accuracy a challenge, particularly at SKU level
• Inventory glide path from DRP not being achieved
• Service issues & expedite costs from none forecast demand
Our starting point
September 2014
transformation
programme launched
Achievements to
December 2015 …
Where do we go
next?

The Case for Change
In a word – FORECAST ACCURACY
Manifested in two key areas; OTIF (availability) and inventory glide
path
“Can we add an extra line to the glide path to account for how wrong
we think the forecast will be”

29%
17% 13% 17% 19% 22% 26% 22% 27% 26% 23% 19%
29%
22%
22%
33% 27%
33% 24%
21%
25% 28%
24% 32%
42%
61% 65%
50% 54%
44% 50% 56%
48% 45% 53% 49%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
FC Error SKU Mix
SKUS <20% Err SKUS 20%-50% Err SKUS > 50% Err
0
2
4
6
8
10
12
14
16
18
20
NumberofSKUs
Forecast Error
Rolling 3 Month FC err %
Half the items have a forecast
error in excess of 50%
At an aggregate level there is a
strong negative bias (over
forecasting)
There is a broad spread of
both under and over forecast
items
Our push to forecast model
results in a bi-modal inventory
distribution of either too
much or too little stock
The Case for Change – Example Forecast Accuracy Results for one Portfolio
Aggregate Demand vs Forecast
Forecast Demand
0
5
10
15
20
25
30
Distribution of SKU’s by Buffer Zone

So What’s the Solution?
Improve forecast accuracy……..
“If I had asked people what they wanted, they would
have said faster horses”
Henry Ford
But…….
“Trying to predict the future is like trying to drive down
a country road at night with no lights while looking out
the back window”
Peter F Drucker

Answer – Become Demand Driven
January 2016 strategic decision to adopt a demand driven approach:
 Enabling us to de-risk the supply chain and address the challenge of demand forecasting
particularly at SKU level
 At first glance this approach made sound sense – de-risks the supply chain and removes
dependency on detailed demand forecasts
 Big question ‘How do we sell this to the rest of the business’
The change challenge:
 Small core team educated in DDMRP (CDDP course)
 Built a ‘proof of concept’ model to test the theory and simulate comparison with
traditional approach
 Introduced concept and outcome of simulations to main line of businesses making up
majority of inventory holdings - all agreed to extend ‘proof of concept’ on larger
portfolio of products
 Potential benefits quantified, supply chain risks assessed and senior stakeholder support
obtained for implementation
 Developed an in house Excel DDMRP model to quickly roll out the proof of concept &
maintain momentum

A Simple Model to Explain the Concept
Demand
Forecast
BT Finished
Goods
In Transit
Supplier
Finished
Goods
Supplier
Manufacturing
Supplier Raw
Materials
100% Coupled and Dependant on Forecast – Forecast Error Propagated Through the Supply Chain
Actual
Customer
Orders
In Transit
Supplier
Finished
Goods
Supplier
Manufacturing
Supplier Raw
Materials
BT
Finished
Goods
DDMRP
Buffer
Driven by Real Demand and Independent of Forecast
PUSHPULL
Buffer sized using:
- Business forecast + planned
events
- Product lead time
- Demand variability
Sized & positioned to cope with
the demand we expect to
happen (have forecasted to
happen) – reviewed monthly as
part of S&OP
Maximum stock exposure is limited to filling the
buffer to top of green, after that all
replenishment orders are driven by real demand
Key Messaging
 Supply chain ‘de-risked’ from forecast
accuracy
 Strong positive impact on cash flow and
ROCE
 Higher service levels and shortened
customer response times
 Overall inventory reduction typically
between 20%-30% across LoBs
 Release of supply chain capacity by not
producing, buying and storing the
wrong stuff
 Cost reduction – lower planning effort
with improved outcomes – more stable
signals to suppliers – lower expedite
costs

01/07/2015
06/07/2015
09/07/2015
14/07/2015
17/07/2015
22/07/2015
27/07/2015
30/07/2015
04/08/2015
07/08/2015
12/08/2015
17/08/2015
20/08/2015
25/08/2015
28/08/2015
02/09/2015
07/09/2015
10/09/2015
15/09/2015
18/09/2015
23/09/2015
28/09/2015
01/10/2015
06/10/2015
09/10/2015
14/10/2015
19/10/2015
22/10/2015
27/10/2015
30/10/2015
04/11/2015
09/11/2015
12/11/2015
17/11/2015
20/11/2015
25/11/2015
30/11/2015
03/12/2015
08/12/2015
11/12/2015
16/12/2015
21/12/2015
24/12/2015
29/12/2015
01/01/2016
06/01/2016
11/01/2016
14/01/2016
19/01/2016
22/01/2016
27/01/2016
InventoryValue(£)
Actual vs DDMRP Simulation – 6 month Inventory Evolution
Actual – over forecast items Actual – under forecast items DDMRP – over forecast items DDMRP – over forecast items
Actual – Total items DDMRP – Total items
Simulation model built using 6 months
history of actual daily orders
Two groups of SKUs from the same family
modelled that had been historically over
and under forecast but at the aggregate
level had excellent accuracy
RESULTS
• Stock is in balance and total inventory
is halved
• NOT Dependant on Forecast so
decoupled from forecast error
• NO Service issues
• LESS Inventory
• STABLE load on factory
• NO costs of failure
Simulation - test the theory
Manufacturing Lead Time

DDMRP is not magic, assuming buffers are not
oversized then a significant increase in demand
will still result in a stock out
The buffer will react and recover but not within
the lead time
Still better than planning to too low a forecast
If the increase is known it can be accommodated
by flexing the buffers (ahead of lead time)
This flex does not have to be driven directly by
an item level detail forecast, buffer sizing can be
discussed as part of the S&OP process and
adjustments and/or simulations run accordingly
01/07/2015
09/07/2015
17/07/2015
27/07/2015
04/08/2015
12/08/2015
20/08/2015
28/08/2015
07/09/2015
15/09/2015
23/09/2015
01/10/2015
09/10/2015
19/10/2015
27/10/2015
04/11/2015
12/11/2015
20/11/2015
30/11/2015
08/12/2015
16/12/2015
24/12/2015
01/01/2016
11/01/2016
19/01/2016
27/01/2016
Item A – 80% Demand Uplift October
RED YELLOW GREEN O/H Avail Stock
01/07/2015
09/07/2015
17/07/2015
27/07/2015
04/08/2015
12/08/2015
20/08/2015
28/08/2015
07/09/2015
15/09/2015
23/09/2015
01/10/2015
09/10/2015
19/10/2015
27/10/2015
04/11/2015
12/11/2015
20/11/2015
30/11/2015
08/12/2015
16/12/2015
24/12/2015
01/01/2016
11/01/2016
19/01/2016
27/01/2016
Item A – 80% Demand Uplift October
RED YELLOW GREEN O/H Avail Stock
A New Role for Forecasting
Soft Benefits
• Better language to communicate
with none supply chain folk
• Greater willingness to share
business insight
• Less emotion when things go
wrong

Sumof3/1/16
Sumof7/1/16
Sumof11/1/16
Sumof15/1/16
Sumof19/1/16
Sumof23/1/16
Sumof27/1/16
Sumof31/1/16
Sumof4/2/16
Sumof8/2/16
Sumof12/2/16
Sumof16/2/16
Sumof20/2/16
Sumof24/2/16
Sumof28/2/16
Sumof3/3/16
Sumof8/3/16
Sumof11/3/16
Sumof15/3/16
Sumof19/3/16
Sumof23/3/16
Sumof27/3/16
Sumof31/3/16
Sumof4/4/16
Sumof8/4/16
Sumof12/4/16
Sumof16/4/16
Sumof20/4/16
Sumof24/4/16
Sumof28/4/16
Sumof2/5/16
Sumof6/5/16
Sumof10/5/16
Sumof14/5/16
Sumof18/5/16
Sumof21/5/16
Sumof26/5/16
Sumof30/5/16
Sumof3/6/16
Sumof7/6/16
Sumof11/6/16
Sumof15/6/16
Sumof19/6/16
Sumof23/6/16
Red (£) Yellow (£) Green (£) Closing Inv (£) Actual Inv (£) Net Flow £
Actual Receipts from
existing POs to this point
44% reduction in
average on-hand
inventory
Buffers dynamically resized using an 8
week historical rolling ADU
Proof of concept simulation – Another example with dynamic buffer
resizing

Count of SKUs
06/06/2016
13/06/2016
20/06/2016
27/06/2016
04/07/2016
11/07/2016
18/07/2016
25/07/2016
01/08/2016
08/08/2016
15/08/2016
22/08/2016
29/08/2016
05/09/2016
OH Red 3 1 2 3 3 4 1 2 3 2 1 1 1 2
OH Yellow 4 6 4 11 6 6 7 6 4 7 8 5 6 9
OH Green 21 24 24 26 36 37 37 36 38 33 32 39 36 37
OH OTOG 86 83 84 74 69 67 69 70 69 72 74 70 72 67
Net Flow Red 1 0 0 0 0 0 0 0 1 1 0 0 0 0
Net Flow Yellow 3 4 3 29 15 6 7 8 13 12 15 9 11 17
Net Flow Green 36 39 52 40 53 64 63 64 58 57 56 59 60 56
Net Flow OTOG 74 71 59 45 46 44 44 42 42 44 44 47 44 42
P12
Q1
Q2
Q3
Jan-16
Feb-16
Mar-16
Apr-16
May-16
Jun-16
Jul-16
Aug-16
Sep-16
Oct-16
Nov-16
Dec-16
Jan-17
Feb-17
Mar-17
DRP
DDMRP
DDMRP Projection
22% reduction in the number of SKUs with too high an on hand
stock
43% reduction in SKUs over top of green – we are not ordering
product we don’t need.
Clear priorities on which SKUs to expedite / ensure on time
delivery
• For the first time since Q1 15/16 inventory has stopped increasing
• Demand has not increased, production has not reduced and we are in
fact still pulling forward production
• DDMRP is ensuring we use capacity to produce what is actually required
and bring the mix back under control
• S&OP is now focused on defining what supply chain capability we need
in place to support the demand we are expecting rather than creating a
master schedule of what we think (hope) we might need
Actual results from Excel model deployed in June 2016
InventoryValue(£)
Inventory Glide Path

Index BT IC Desc ADU MOQ Order Cycle (days) + On Order -Qualified Sales Orders On Hand =Available Stock Recommended Order
1 Part1 601.08 8100 15 32,400 4,000 6,169 34,569 8100
2 Part2 58.69 6000 102 6,000 0 631 6,631 0
3 Part3 938.25 5030 15 45,270 0 10,542 55,812 15090
4 Part4 3417.14 8100 15 146,853 1,950 48,043 192,946 64800
5 Part5 347.85 8075 23 25,089 0 5,092 30,181 0
6 Part6 213.46 4840 23 10,502 1,500 4,491 13,493 0
7 Part7 8.63 12 15 415 13 194 596 0
8 Part8 2569.15 10100 15 125,193 2,500 67,041 189,734 0
9 Part9 190.77 10000 52 10,400 0 5,200 15,600 0
10 Part10 1670.35 4015 15 70,506 3,975 50,872 117,403 0
11 Part11 156.08 8100 52 8,100 790 4,951 12,261 0
12 Part12 6.66 50 15 250 10 221 461 0
13 Part13 711.98 3040 15 14,310 1,940 24,728 37,098 15200
14 Part14 10310.63 14050 15 315,961 13,635 360,075 662,401 0
DDMRP Excel Model - Clear Plan Priorities
Critical Items
• Expedite existing POs
• Call out any delays
• On Hand / ADU = Days of cover
Potentially at risk items
• Monitor
• Check days of cover vs PO due date
• Call out any delays
BT IC Desc On Hand ADU OH Cover PO NUMBER PO DATE
QUANTITY
ORDERED
QUANTITY
DELIVERED NEED BY DATE
Part1 6,169 601 10 xxxxxxxxxxx 25/05/2016 8100 0 30/06/2016
xxxxxxxxxxx 29/06/2016 16200 0 31/07/2016
xxxxxxxxxxx 13/05/2016 8100 0 27/06/2016
Part2 631 59 11 xxxxxxxxxxx 07/06/2016 6000 0 19/07/2016
xxxxxxxxxxx 17/06/2016 15090 0 29/07/2016
xxxxxxxxxxx 29/06/2016 10060 0 31/07/2016
xxxxxxxxxxx 29/06/2016 5030 0 15/08/2016
Part4 48,043 3,417 14 xxxxxxxxxxx 07/06/2016 72900 7047 19/07/2016
xxxxxxxxxxx 29/06/2016 48600 0 31/07/2016
xxxxxxxxxxx 29/06/2016 32400 0 15/08/2016
xxxxxxxxxxx 17/06/2016 8075 0 29/07/2016
xxxxxxxxxxx 05/07/2016 16150 0 16/08/2016
Expedite
Confirm
delivery date

New Focus for S&OP -> DDS&OP
• Less time spent trying to predict the future
– Focus shifts to putting the capability in place to cope with the expected demand and
its variability
• Focus is on managing the tails of the inventory distribution – too much / too
little stock
• Move away from the goal of one number
– Scenarios and an integrated set of plans that support the business strategy
– Then focus on what is changing – the exceptions
• Short term horizon no longer looked at, focus is now on 3 months+
– DDMRP takes care of planning & execution in the short term
0
10
20
30
40
Ideal Distribution
Ideal
Too
Much
Too
Little

DDMRP
currently
deployed via
in-house
solution
Implementation
of software
solution to
underpin
demand driven
operating
model
Work with key
suppliers to
extend DDMRP
back into their
supply chains
Align
functional
KPIs to
support
DDS&OP
Next Steps

Learnings
• Invest in education
– CDDP & CDDL
• Just do it
– Test it on a small subset of products in Excel
• Learn by doing
– As above, just start then shape, develop, and refine as you go – and then invest in
technology
• DDMRP is disruptive
– Embrace the transformation and don’t be afraid to abandon current perceived wisdom

Rethink what your supply chain can deliver

Improving Supply Planning - A Demand Driven Journey

Recommended

Recommended

More Related Content

Similar to Improving Supply Planning - A Demand Driven Journey

Similar to Improving Supply Planning - A Demand Driven Journey (20)

More from Ben Bradley

More from Ben Bradley (12)

Recently uploaded

Recently uploaded (20)

Improving Supply Planning - A Demand Driven Journey