WT F I S A DATA STRATEGY ? 
1 1 T H N O V 2 0 1 4 
Alex Loveless 
Head of Data Consulting 
@alexmloveless
Why the f**k a data strategy?
Data = Strategy
What are your business’s short, 
medium and long term goals?
It’s all about the customer 
Customers don’t experience channels 
PPC 
DISPLAY 
ORGANI 
C 
AFFILIAT 
ES 
EMAIL 
MOBILE 
WEB 
customer lifecycle 
JUST GIVE ME 
WHAT I WANT!
Data 
Insight 
Brand Planning 
Omni-channel Planning 
Optimised Brand Experience
creates 
informs 
PPC 
SEO 
Email 
Social 
Web 
Mobile 
behaviour 
social 
3rd party 
purchases 
CRM 
app 
analysis 
value 
prediction 
segmentation 
behaviour 
UX 
data 
media 
customer 
creative 
social
Time Horizon 
Customer Volume 
Short Medium Long 
1 customer ALL my customers 
Web/mobile analytics 
Data Lake/Single customer view 
Attribution Modeling 
Cross Channel Analysis 
DMP/Personalisation 
Ad Targeting/programatic 
Propensity Modeling 
Behavioural Segmentation 
Social Listening 
Social segmentation 
Customer 
clustering 
Propensity modeling
It’s cool man! Just use 
stuff you’ve, like, seen 
in the past to predict 
what might happen in 
the future, m’kay? 
Rev. Thomas Bayes 
Bayes’ Theorem
• Customer data 
• Purchase history 
• Social data 
Find data 
Look for clusters 
(segments) in 
the data 
Look for 
features in the 
segments 
Observe history 
of segments 
Create Bayesian 
training set 
Use to predict 
future states 
• Car lovers 
• Mums 
• Optimists 
• Gender 
• Demographics 
• Frequent purchases 
• Behaviours 
• Purchases 
• Loyalty 
• Likely next product 
• Likelihood to purchase in X days 
• Probable next destination 
• Pregnancy status 
• Likelihood to complain
CONTEXT 
Who are they? 
Where are they? 
What are they 
doing? 
Who are they 
talking to? 
What devices 
are they using? 
How can we tailor their 
experience? 
What do they 
How are they 
feeling? 
want? 
Timing is everything! 
At 
home 
At 
work 
In a 
café 
On the 
move 
Eating 
Watching 
tv 
Social 
web 
Friends 
Family 
Colleag 
ues 
Laptop 
Tablet 
Phone 
Irritable 
Positive 
Stressed 
Excited 
To buy 
something 
To 
browse 
To 
compare 
prices 
Personaliz 
ed email 
Targeted 
content 
Deals 
Male/Fe 
male 
@some 
body 18 - 25 
Optimised 
Experience
Sees 
Display 
ad 
Logs 
onto in-store 
network 
Searche 
s for 
“TV” 
Uses 
app 
from 
café 
Whatsap 
p link to 
friend 
Returns 
unwante 
d item 
Uses 
browser 
Bounces 
Social 
share 
Arrives 
from 
search 
Selects 
Uses app shoe size 
Multiple 
visit no 
purchase 
Opens 
email 
Complains 
Uses 
airport 
kiosk 
In store 
sale 
Social 
campaign
With more data 
resolution increases 
• Behaviour 
• 3rd party data 
• Browse habits 
• Singe device 
• Customer data 
• Email address/social profile 
• Geographic 
• Product affinities 
• App usage 
• Cross device
media CRM 
purchase Social 
DATA 
2nd 
Party 
analytic 
s 
Offline 
Call 
Centre 
How do we 
do this in 
real time?
Data Management 
Platform (DMP) 
Clickstream/Conv. 
3rd Party Data 
Data Exchange 
CRM/Customer 
PPC 
Call Centre 
Targeted Email/Comms 
Display Targeting 
Deals/Incentives 
Mailings/Inserts 
Online to Offline 
Offline/POS 
Web/Mobile 
Personalisation 
Social 
Events 
Call centre 
prequalification 
Loyalty programme 
segmentation 
Predictive 
analysis 
Data Inputs 
Outputs/Actions
OMNI-PPC 
CHANNEL PERSONALISED 
DISPLAY 
ORGANI 
C 
AFFILIATES 
EMAIL 
MOBILE 
WEB 
customer lifecycle 
EXPERIENCE
Call Centre 
Targeting 
Campaign and 
trigger emails are 
personalised 
based on 
customer 
behaviour 
DMP 
Analytics 
Social Login 
Web/Mobile 
email 
Tag Management 
MVT 
Behavioural data is collected via a tag 
management system and fed into a 
profiling DB for segmentation. 
Personalised content is delivered to 
the customer using an MVT tool 
Display Ads 
Social Media 
Profiling DB stiches sessions 
together across device and 
channel and creates 
segments 
Customer profiles are 
enriched with social data 
which is linked via social 
logins 
Call centre agents use 
profiling data to enrich their 
conversations with 
customers 
User experience is tailored to 
the campaign seen prior to 
arrival 
Customer experience is aligned 
across channel and platform based 
on information gathered about 
them from any channel in real time 
or near real time. 
DATA LAYER 
{ 
user: johnsmith, 
id :123456, 
agegroup : 18-25, 
prevsearch: “NewYork” 
“campaigns: [ 
123, 
456, 
334 
] 
} 
The application 
embeds customer 
profile info in a 
data layer on the 
page
DMP CRM Analysis 
Data Lake 
Customer 
Social 
Offline/ 
POS 
2nd Party 
Logs 
Legacy 
Data Broker 
Call 
Centre 
3 Other rd Party 
Data comes in from anywhere 
Data of all 
shapes and 
sizes live in 
the data 
lake 
Data data 
broker 
makes 
sense of 
data on the 
fly 
Do lots of cool stuff with the data 
1 
2 
3 
4
ANALYTICS 
DATA WAREHOUSE 
SINGLE CUSTOMER VIEW 
MASTER DATA MANAGEMENT 
BIG DATA 
BRAND EXPERIENCE OPTIMISATION
1. Business strategy is key – what are 
your big questions? 
2. It’s all about customer insight, 
more is better but any is great 
3. Start big - go small

WTF is a Data Strategy? - WTF Programmatic UK, 11/11/14

  • 1.
    WT F IS A DATA STRATEGY ? 1 1 T H N O V 2 0 1 4 Alex Loveless Head of Data Consulting @alexmloveless
  • 2.
    Why the f**ka data strategy?
  • 3.
  • 4.
    What are yourbusiness’s short, medium and long term goals?
  • 5.
    It’s all aboutthe customer Customers don’t experience channels PPC DISPLAY ORGANI C AFFILIAT ES EMAIL MOBILE WEB customer lifecycle JUST GIVE ME WHAT I WANT!
  • 6.
    Data Insight BrandPlanning Omni-channel Planning Optimised Brand Experience
  • 7.
    creates informs PPC SEO Email Social Web Mobile behaviour social 3rd party purchases CRM app analysis value prediction segmentation behaviour UX data media customer creative social
  • 8.
    Time Horizon CustomerVolume Short Medium Long 1 customer ALL my customers Web/mobile analytics Data Lake/Single customer view Attribution Modeling Cross Channel Analysis DMP/Personalisation Ad Targeting/programatic Propensity Modeling Behavioural Segmentation Social Listening Social segmentation Customer clustering Propensity modeling
  • 9.
    It’s cool man!Just use stuff you’ve, like, seen in the past to predict what might happen in the future, m’kay? Rev. Thomas Bayes Bayes’ Theorem
  • 10.
    • Customer data • Purchase history • Social data Find data Look for clusters (segments) in the data Look for features in the segments Observe history of segments Create Bayesian training set Use to predict future states • Car lovers • Mums • Optimists • Gender • Demographics • Frequent purchases • Behaviours • Purchases • Loyalty • Likely next product • Likelihood to purchase in X days • Probable next destination • Pregnancy status • Likelihood to complain
  • 11.
    CONTEXT Who arethey? Where are they? What are they doing? Who are they talking to? What devices are they using? How can we tailor their experience? What do they How are they feeling? want? Timing is everything! At home At work In a café On the move Eating Watching tv Social web Friends Family Colleag ues Laptop Tablet Phone Irritable Positive Stressed Excited To buy something To browse To compare prices Personaliz ed email Targeted content Deals Male/Fe male @some body 18 - 25 Optimised Experience
  • 12.
    Sees Display ad Logs onto in-store network Searche s for “TV” Uses app from café Whatsap p link to friend Returns unwante d item Uses browser Bounces Social share Arrives from search Selects Uses app shoe size Multiple visit no purchase Opens email Complains Uses airport kiosk In store sale Social campaign
  • 13.
    With more data resolution increases • Behaviour • 3rd party data • Browse habits • Singe device • Customer data • Email address/social profile • Geographic • Product affinities • App usage • Cross device
  • 14.
    media CRM purchaseSocial DATA 2nd Party analytic s Offline Call Centre How do we do this in real time?
  • 15.
    Data Management Platform(DMP) Clickstream/Conv. 3rd Party Data Data Exchange CRM/Customer PPC Call Centre Targeted Email/Comms Display Targeting Deals/Incentives Mailings/Inserts Online to Offline Offline/POS Web/Mobile Personalisation Social Events Call centre prequalification Loyalty programme segmentation Predictive analysis Data Inputs Outputs/Actions
  • 16.
    OMNI-PPC CHANNEL PERSONALISED DISPLAY ORGANI C AFFILIATES EMAIL MOBILE WEB customer lifecycle EXPERIENCE
  • 17.
    Call Centre Targeting Campaign and trigger emails are personalised based on customer behaviour DMP Analytics Social Login Web/Mobile email Tag Management MVT Behavioural data is collected via a tag management system and fed into a profiling DB for segmentation. Personalised content is delivered to the customer using an MVT tool Display Ads Social Media Profiling DB stiches sessions together across device and channel and creates segments Customer profiles are enriched with social data which is linked via social logins Call centre agents use profiling data to enrich their conversations with customers User experience is tailored to the campaign seen prior to arrival Customer experience is aligned across channel and platform based on information gathered about them from any channel in real time or near real time. DATA LAYER { user: johnsmith, id :123456, agegroup : 18-25, prevsearch: “NewYork” “campaigns: [ 123, 456, 334 ] } The application embeds customer profile info in a data layer on the page
  • 18.
    DMP CRM Analysis Data Lake Customer Social Offline/ POS 2nd Party Logs Legacy Data Broker Call Centre 3 Other rd Party Data comes in from anywhere Data of all shapes and sizes live in the data lake Data data broker makes sense of data on the fly Do lots of cool stuff with the data 1 2 3 4
  • 19.
    ANALYTICS DATA WAREHOUSE SINGLE CUSTOMER VIEW MASTER DATA MANAGEMENT BIG DATA BRAND EXPERIENCE OPTIMISATION
  • 20.
    1. Business strategyis key – what are your big questions? 2. It’s all about customer insight, more is better but any is great 3. Start big - go small

Editor's Notes

  • #3 I’m assuming that, because you’re sitting here now you know the answer to this
  • #4 Don’t have to rely on gut feel any more. Go find the answer. Reference “Black Swan” here
  • #5 I want you to bear these in mind throughout this presentation
  • #7 Channel agnostic, device agnostic, joined up, seamless, happy, pleasurable, and above all profitable. So how do we get to that point?
  • #10 18th century Presbyterian minister who, perhaps ironically, showed how to use new evidence to update beliefs.
  • #11 It’s not as hard as it sounds The trick is to operationalise it
  • #13 I’ll take whatever I can get Accuracy is a secondary concern Don’t nneed to know everything just somethingf