NHS Supply Chain

1
NHS Supply Chain

• Segmentation Analysis
• Jahanara Choudhury
• Geoff Ellis

2
Background information
3
An introduction
The Department of Health (DoH) engaged with
industry to identify and appoint an organisation
to fulfil a set of key objectives

• Centralising procurement
• Achieving value for money for the NHS
• Reducing management costs
• Managing financial risk
• Making the most effective use of resources

• DHL appointed 1st October 2006 to manage on
  behalf of the DoH an organisation NHS Supply
  Chain
• NHS Supply Chain is the integration of NHS
  Logistics, Part of NHS PaSA and DHL

/
/
NHS Supply Chain has exclusive scope from the UK
Department of Health covering over 5bn of
market. This represents over 28 of NHS
procurement spend.
4
Who do we work with?

• NHS Supply Chain works with around 600 trusts to
  provide an end-to-end supply chain service or one
  of our key services
• Acute care trusts
• primary care trusts
• mental health organisations
• ambulance service
• private sector service providers
• collaborative procurement hubs
• patients in their homes.

5
Our customers
6
Product categories

• Food and kitchen
• office products, print and stationery
• uniforms, clothing and linen
• furniture and office equipment
• medical and facility furniture
• lab/pathology products and equipment
• medical supplies
• medical and surgical equipment
• operating theatre supplies
• patient appliances
• maintenance
• capital equipment.

7
Logistics overview

• 33m lines despatched annually
• national fleet of 340 vehicles
• national trunking fleet to facilitate cross dock
  services and supplier backhauls
• delivery service to all NHS England trusts
• integrated warehouse management and ordering
  system
• dedicated call centre and customer service.

8
Logistics overview

• Core logistics services offering
• 51,000 product lines
• consolidated service using a combination of
  stock, cross dock, co-ordinated direct delivery
  and patient home delivery service
• break bulk to specific order requirements, e.g.
  ward, department, clinic, home
• 48 hour service or 24 hour for critical care, eg
  theatres
• 98 availability, 99 on critical lines
• 98 delivery on time measured to within /- 30
  minutes of promise
• five hour emergency response when needed.

9
Summary
10
Key facts

• We supply to over 250,000 NHS requisition points
• we deliver to over 10,000 delivery locations
• we service 120,000 patients in their own home
• our catalogue range covers over 610,000 different
  products
• we manage over 4 million NHS orders every year,
  servicing over 30 million order lines
• we send out only 52 invoices a year to each trust
• dedicated and closed loop transport fleet.

11
Commercial arrangements

• NHS trusts are not obliged to use NHS Supply
  Chain
• Prices will include a margin as they have
  previously
• Cost base transferred from Department of Health
  for NHS Logistics Authority and NHS PASA
  increased by new recruitment
• No underwrite of profit for DHL but upper cap in
  place.

12
Why do we need to segment?

• To drive increased cost efficiency and profitable
  sales growth by migrating customers towards best
  in class behaviours
• To ensure we selling the right items to the right
  people
• Increase return on investment through targeted
  sales and marketing
• To ensure our service development meets customers
  needs

13
Customer segmentation

• Initial segmentation conducted to understand the
  user group of every single requisition point on
  our system - gt100,000 points reviewed and
  labelled
• Identification of 91 customer departments
• Analysis of 4 million transactions across 10,000
  requisition points as a sample to understand
  buying behavioural types across those segments,
  undertaken by a consultancy with segmentation
  expertise
• Review of outputs and qualification of findings

14
Contract Segmentation

• Phase 2 of segmentation based on products
  analysis
• 26 million transaction records analysed over a
  one year time frame
• Analysis of all classifications of trusts broken
  down into each region and classification
• Departmental contract segmentation completed
• 316 contracts analysed with over 51,000
  individual products
• 6 clusters created incorporating profitability,
  market value and transactional activity
• Customer information analysed at trust and
  department level

15
2D Segmentation

• The combination of segmentation by Product by
  Customer
• Review of outputs and qualification of findings
• Identification of 13 behavioural types/segments
• Next ready to start to use results to drive
  business strategies
• Understand and develop strategy for each segment
• Influence multiple buying behaviours of each
  segment and migrate towards best in class

16
The Use of PASW (or even SPSS!)

• Geoff Ellis

17
Segmentation aim

• A 2D segmentation
• Each product belongs to a segment
• Each Customer belongs to a segment
• Each combination of product Customer belongs to
  a unique segment

18
Some technical details

• 2 Step Clustering
• Similar to the combination of Hierarchical
  Non-Hierarchical (K means) clustering techniques
• Analogous to CHAID
• Variables
• What variables to include
• Continuous / Capped / Binned data
• Are some variables more important than others
• How should they be weighted
• Number of clusters
• How many do you want
• How do different solutions compare
• Missing Data
• Discriminant Analysis on variables present

19
clustering by customer

• the following variables were produced used
• Whether customer placed order in last week
• Whether customer placed order in last fortnight
• Whether customer placed order in last month
• Whether customer placed order in last quarter
• Volume sold of each of 19 product groups
• Whether anything was sold to each of 19 product
  groups
• Capped No of orders
• Capped sales value

• Capped sales volume
• Capped margin
• Capped unfulfilled amount
• Capped Av order value
• Capped Av order volume
• Capped Av order margin
• Unfulfilled proportion
• Capped No products
• No product types
• Av price band
• Capped total quantity supplied (incl carton size)

20
How many clusters?
strongly suggests a 5 cluster solution
21
total supply quantity
22
occurrence of sales of prod type B (staff
uniforms) by cluster
23
cluster by customers
24
cluster by value of sales
25
clustering by product

• the following variables were produced used
• Av price band
• No. departments buying the product
• Unfulfilled prop
• Carton size
• Volume sold to each of 91 departments
• Whether anything was sold to each of 91
  departments
• Capped No of orders
• Capped sales value

• Capped sales volume
• Capped margin
• Capped unfulfilled amount
• Capped No unique req point codes
• Capped Av order value
• Capped Av order volume
• Capped Av order margin
• Capped total quantity supplied (incl carton size)

26
Products by cluster
27
Sales (Value) by cluster
28
Cluster 3
29
Cluster 5
30
combined segmentation
31
definitions
1 A1 high buying customers buying popular general products
2 A2 top customers buying popular general products
3 A3 A4 A5 customers who dont buy much buying popular general products
4 B1 B2 D1 D2 high buying customers buying specialist products without high prices or margin
5 B3 B4 B5 D3 D4 D5 customers who dont buy much buying specialist products without high prices or margins
6 C1 C2 high buying customers buying low volume, high margin products
32
definitions
7 C3 C4 C5 customers who dont buy much buying low volume, high margin products
8 E1 high buying customers buying top selling general products
9 E2 top customers buying top selling general products
10 E3 low purchasing customers buying top selling general products
11 E4 low purchasing but high margin customers buying top selling general products
12 E5 infrequent customers buying top selling general products
33
segment 11

• Av Margin per Order 24.56 Av 2.84
• Av Value per Order 506 Av 33
• Av Supply Qty per Order 106 Av 10.1
• Av Price Band 1.49 Av 1.19
• Av No Prods Types/Cust 6.1 Av 10.1
• high value large orders for specialist products

34
cluster 11 0.7 orders
metrics
key features

• low purchasing but high margin customers buying
  top selling general products
• 1/3 the Av orders per customer
• 1/2 the Av orders per product
• relatively infrequent orders (Av cust 3.5
  days)
• 1.5x the Av carton size
• Items bought have 2x the Av no custs
• Customers buy 1/3 the Av no of products
• top bulk purchasing discount

• customers 258
• products 709
• value 10.3
• volume 6.6
• margin 5.9
• av value 506.28
• av volume 106.3
• av margin 24.56

top departments
top product types

• Audiology (84)
• Domestics (5)
• Pathology (4)

• Lab Chemicals reagents (86)
• Laundry Cleaning Materials (6)
• Lab Instruments Eqpmt (4)

35
Av. margin per order by segment
36
2D Clustering

• Every product / customer has been allocated to
  one of 12 groups.
• The client wanted to target specific customer
  groups with different messages according to their
  behaviour
• Buy a larger range of products, buy specific
  types of products, buy larger quantities less
  frequently
• these 12 segments cannot be applied uniquely to
  specific customers
• a further clustering was performed which grouped
  customers according to their pattern of purchases
  (as defined by volume, value and margin) across
  these 12 segments
• this resulted in 13 groupings of individual
  customers

37
(No Transcript)
38
clustering by department
1 AE
6 Chiropody / Podiatry
40 Orthopaedics
42 Paediatrics
63 Theatre
68 Audiology
86 Supplies Procurement
39
clustering by department
Weighted by Orders
40
Targeted Marketing

• By hospital (postcode) department
• A full day workshop with all marketing staff
• Where are they now?
• Where would we like them to be thats feasible?
• Resulted in 11 specifically targeted marketing
  and communication plans

41
Any Questions?