More Related Content Similar to Jason Juma-Ross - Accenture Similar to Jason Juma-Ross - Accenture (20) More from Gavin Stewart (12) Jason Juma-Ross - Accenture1. Agile, intelligent marketing
Jason Juma-Ross
@ideasoc
AdTech Sydney, 2013
Copyright © 2013 Accenture All Rights Reserved. Accenture, its logo, and High Performance Delivered are trademarks of Accenture.
2. 2013: another digital media watershed
$000s 2012 Australian Media Spend & YOY Δ
($24m)
4,000,000
$516m
3,000,000
($476m)
2,000,000
1,000,000 ($18m) $21m $11m
$8m ($41m)
0
Television Press Magazines Radio Cinema Out of Home Direct Mail Digital
ROI
Basis
> >
Historical Marginal Optimal
Source: AQX Monthly, Jan 2011 – Jan 2013, IAB Online Advertising Expenditure Report, Dec 2012
Copyright © 2013 Accenture All Rights Reserved. 2
3. 2013: another digital media watershed
$000s 2012 Australian Media Spend & YOY Δ
($24m)
4,000,000
$516m
3,000,000
($476m)
2,000,000
1,000,000 ($18m) $21m $11m
$8m ($41m)
0
Television Press Magazines Radio Cinema Out of Home Direct Mail Digital
ROI
Basis 25% 8%
(example
)
(2%) (28%) (60%) (10%)
Source: AQX Monthly, Jan 2011 – Jan 2013, IAB Online Advertising Expenditure Report, Dec 2012, Accenture analysis: APAC region.
Copyright © 2013 Accenture All Rights Reserved. 3
4. Two types of organisations; two types of
marketing processes
20,000
ANZ Bank 18,000
16,000 Toyota
14,000
Interactive (Online & Direct) Spend ($ 000s)
AMEX
Relatively
NAB unknown: 12,000
iterative
Origin
Telstra
10,000
Westpac Optus MEAN: 9%
8,000
VW CBA
Hyundai
0 20,000 40,000 6,000 60,000 80,000 100,000 120,000
Holden MEDIAN:
Coles 4%
4,000 Relatively known:
stable process
Woolworths
2,000
McDonalds
Target
0
Reckitt B.
Broadcast (non-interactive) Spend ($ 000s) Harvey Norman
Source: AQX Monthly, Feb 2012 – Jan 2013 Top 10 Traditional Top 10 Digital & Interactive
Copyright © 2013 Accenture All Rights Reserved. 4
5. Marketing performance varies enormously
across digital campaigns
• Return (campaign performance) varies
enormously
– CTR median of 0.0005, range of
0.05 (100 times the median)
– Skewed distribution. Only a
Campaign CTR
few, very high CTRs in the >1%
range
– Base scenarios becoming more
costly (having lower ROI)
• Cost (CPX) is even more varied
How can we consistently improve
marketing performance in the
‘relatively unknown’ territory of
Media CPX digital?
Source: Accenture analysis, example digital campaigns aggregate 12m data
Copyright © 2013 Accenture All Rights Reserved. 5
6. Use ‘fast transients’ to deliver relevance at
scale & speed
Search Display Media
KWG 1 KWG 2 KWG N Seg 1 Seg 2 Seg N
de-averaging
Dynamic Landing Pages
Product/Detail Pages
• Integrated data
• Intelligent conversion paths Checkout / Conversion
• Industrialised automation &
decisioning
Copyright © 2013 Accenture All Rights Reserved. 6
7. Delivering relevance has a high cost of
complexity
Customers demand a more granular and continuous content and functionality
development cycle than is possible in the current paradigm
New Paradigm
Complexity Dimensions
Platform based,
componentised dev.,
5 Cust. Segments 12,150 Treatments flexible architecture with
analytics linking
2,430
content, usage, and
3 Channels value
+
Relevance
6 Regions 810
Current Paradigm
135 Monolithic web
27 Brands development & digital
Current supply chain. Analytics
5 Treatments used primarily for
5 Product categories reporting purposes
1 Treatment
Unit Delivery Cost +
Copyright © 2013 Accenture All Rights Reserved. 7
8. Known user profiling to drive content
targeting at the last millisecond
Profile Data Context Cloud Digital Data Warehouse
Repository
Onsite Behaviours Demographic Data
Social Profiles Custom Data Stores
NB. Example ‘Context Cloud’ from Adobe CQ5
Copyright © 2013 Accenture All Rights Reserved. 8
9. Intent can be estimated through combining
unknown user context data
Environmental Variables
• IP address Referrer Variables
• Country of origin • Referring domain
Site Behaviour Variables • Time zone • Campaign ID
• Customer/prospect • Operating system • Affiliate
• New/return visitor • Browser type • PPC
• Previous visit patterns • Screen resolution • Natural search
• Previous Product • Direct/bookmark
interests – top level
• Previous Product
interests – low level Temporal Variables
• Searches • Time of day
• Previous online • Day of week
purchases
• Recency
• Previous Campaign
exposure • Frequency
• Previous Campaign
responses Highly Predictive
Anonymous
Profile For Testing Offline
Variables
Copyright © 2013 Accenture All Rights Reserved. 9
12. Lagging Leading Emerging
STATIC ‘ONE-SIZE FITS PERIODIC, EMPIRICALLY- AGILE, INTELLIGENT
ALL’ WEB SITES DRIVEN ITERATION DELIVERY
Search Social Display Personalisation
(HTML)
Web Skin Core Systems Componentised
(Aligned)
(HTML Layer) Architecture
Search Social Display
Foundational
Components Intelligence Driven
(Analytics) Intelligent
Adaptation
Analytics Applications,
(Reporting) transactional, and
service platforms
Customer
Data Cloud
Copyright © 2013 Accenture All Rights Reserved. 12
13. Relevance = business de-averaged
1960 1980 2000 2020
Broadcast Paradigm Intent Paradigm
Campaigns
Demand
Campaign
Demand
Profile
Profile
Population Population
Demog. Simple Single Uniform Intent Multiple Fragmented Campaign
Segment Offer Channel Campaign Segments Offers Channels Relevance
Local
Mobile
Social
Bundle
Search
Web
eDM/DM
IPTV,
etc
Hindsight based business Relevance, scale, & speed
Copyright © 2012 Accenture All Rights Reserved. 13
Editor's Notes Performance-based shifts in marketing spendExample: deltas for APAC advertisers based on MMM projectsExample: shift in Australia spend from Nielsen data or IAB dataQuestion: how do you do digital well?First, you establish baseline effectiveness… Performance-based shifts in marketing spendExample: deltas for APAC advertisers based on MMM projectsExample: shift in Australia spend from Nielsen data or IAB data Media landscape polarisationThe media landscape is polarising into those that invest and execute well in digital and those that do notA key issue is that digital is a relatively unknown process so effectiveness varies Execution is criticalTremendous variation within channels. How many standard deviations in ANZ Banner CTR dataCTRs vary across media types Integrated: maximise data and reach synergies across media typesIntelligent: optimise full conversion paths for each visitor with full accountabilityIndustrialized: leverage high levels of automation and machine decisioningThe result: de-averaged, scalable digital marketing How to deliver relevance @ scale & speed- Automation platforms- Data-driven decisioning & machine learning- Scale, low cost message & creative developmentDigital & direct processes & mindset