[ Digital Measurement ] Analy&cs workshop on how to turn data into ac&onable insights
Jan 22, 2015
[ Digital Measurement ] Analy&cs workshop on how to turn
data into ac&onable insights
[ Company history ]
§ Datalicious was founded in 2007 § Strong Omniture web analy&cs history § One-‐stop data agency with specialist team § Combina&on of analysts and developers § Making data accessible and ac&onable § Driving industry best prac&ce § Evangelizing use of data
June 2010 © Datalicious Pty Ltd 2
[ Challenging clients ]
June 2010 © Datalicious Pty Ltd 3
[ Data driven marke:ng ]
June 2010 © Datalicious Pty Ltd 4
Data Pla<orms Data collec:on and processing Web analy:cs solu:ons Omniture, Google Analy:cs, etc Tagless online data capture End-‐to-‐end data pla<orms IVR and call center repor:ng Single customer view
Insights Repor:ng Data mining and modelling Customised dashboards Media aKribu:on models Market and compe:tor trends Social media monitoring Online surveys and polls Customer profiling
Ac:on Applica:ons Data usage and applica:on Marke:ng automa:on Aprimo, Trac:on, Inxmail, etc Targe:ng and merchandising Internal search op:misa:on CRM strategy and execu:on Tes:ng programs
[ Today ]
§ Capturing data – Op&ons, limita&ons, innova&ons
§ Genera&ng insights – Process, metrics, examples
§ Taking ac&on – Media, targe&ng, tes&ng
June 2010 © Datalicious Pty Ltd 5
[ Capturing data ]
101011010010010010101111010010010101010100001011111001010101010100101011001100010100101001101101001101001010100111001010010010101001001010010100100101001111101010100101001001001010
June 2010 © Datalicious Pty Ltd 6
[ Digital data is cheap ]
June 2010 © Datalicious Pty Ltd 7
Source: Omniture Summit, MaS Belkin, 2007
[ Digital data op:ons ]
June 2010 © Datalicious Pty Ltd 8
Source: Accuracy Whitepaper for web analy&cs, Brian CliWon, 2008
+Social
[ On-‐site analy:cs tools ]
June 2010 © Datalicious Pty Ltd 9
Source: Forrester Wave Web Analy&cs, 2009
Google: ”forrester wave
web analy:cs pdf” or
hKp://bit.ly/aTLAKT
[ What pla<orm to use ]
June 2010 © Datalicious Pty Ltd 10
Time, Control
Soph
is&ca&o
n
Stage 1: Data Stage 2: Insights Stage 3: Ac:on
Third par&es control most data, ad hoc repor&ng only, i.e. what happened?
Data is being brought in-‐house, shiW towards insights genera&on and data mining, i.e. why did it happen?
Data is fully owned in-‐house, advanced predic&ve modelling and trigger based marke&ng, i.e. what will happen and making it happen!
[ Governance and data integrity ]
June 2010 © Datalicious Pty Ltd 11
Source: Omniture Summit, MaS Belkin, 2007
© Datalicious Pty Ltd
[ Free off-‐site analy:cs tools ] § hSp://www.google.com/trends § hSp://www.google.com/sktool § hSp://www.google.com/insights/search § hSp://www.google.com/webmasters § hSp://www.google.com/adplanner § hSp://www.google.com/videotarge&ng § hSp://www.keywordspy.com § hSp://www.compete.com § hSp://www.alexa.com § hSp://wiki.kenburbary.com June 2010 12
[ Search at all stages ]
June 2010 © Datalicious Pty Ltd 13
Source: Inside the Mind of the Searcher, Enquiro 2004
In Australia Google has a market share of almost 90% of all searches, making it a very large and reliable data sample
[ Search call to ac:on for offline ]
June 2010 © Datalicious Pty Ltd 14
[ Client side tracking process ]
June 2010 © Datalicious Pty Ltd 15
Source: Google Analy&cs, Jus&n Cutroni, 2007
What if: Someone deletes their cookies? Or uses a device that does not support JavaScript? Or uses two computers (work vs. home)? Or two people use the same computer?
[ Tag-‐less data capture ]
June 2010 © Datalicious Pty Ltd 16
Google: “atomic labs” www.atomiclabs.com
The study examined data from two of the UK’s busiest ecommerce websites, ASDA and William Hill. Given that more than half of all page impressions on these sites are from logged-‐in users, they provided a robust sample to compare IP-‐based and cookie-‐based analysis against. The results were staggering, for example an IP-‐based approach overes&mated visitors by up to 7.6 &mes whilst a cookie-‐based approach overes:mated visitors by up to 2.3 :mes. Google: ”red eye cookie report pdf” or hKp://bit.ly/cszp2o
[ Overes:ma:on of unique visitors ]
June 2010 © Datalicious Pty Ltd 17
Source: White Paper, RedEye, 2007
[ Maximise iden:fica:on points ]
June 2010 © Datalicious Pty Ltd 18
0%
20%
40%
60%
80%
100%
120%
140%
0 4 8 12 16 20 24 28 32 36 40 44 48 Weeks
Probability of iden&fica&on through cookie
June 2010 © Datalicious Pty Ltd 19
Datalicious SuperCookie Persistent Flash cookie that cannot be deleted
[ Mobile page headers ]
June 2010 © Datalicious Pty Ltd 20
Source: Mobile Tracking, Omniture, 2008
MSISDN = Mobile Number
[ Single-‐sign on ]
June 2010 © Datalicious Pty Ltd 21
Facebook Connect gives your company the following data and more with just one click! ID, first name, last name, middle name, picture, affilia&ons, last profile update, &me zone, religion, poli&cal interests, interests, sex, birthday, aSracted to which sex, why they want to meet someone, home town, rela&onship status, current loca&on, ac&vi&es, music interests, tv show interests, educa&on history, work history, family and email Need anything else?
[ Research online, shop offline ]
June 2010 © Datalicious Pty Ltd 22
Source: 2008 Digital Future Report, Surveying The Digital Future, Year Seven, USC Annenberg School
Google: ”digital future report 2009 pdf” or hKp://bit.ly/ZkLvr
[ Offline sales driven by online ]
June 2010 © Datalicious Pty Ltd 23
Cookie
Website.com Research
Credit Check Fulfilment
Phone Orders
Retail Orders
Online Orders
Credit Check Fulfilment
Credit Check Fulfilment
Website.com Research
Website.com Research
Online Order Confirma:on
Virtual Order Confirma:on
Virtual Order Confirma:on
Virtual Order Confirma:on
@
@
@
Cookie Cookie
Adver:sing Campaign
Tying offline conversions back to online campaign and research behavior using standard cookie technology by triggering virtual online order confirma&on pages for offline sales using email receipts.
[ Summary: Capturing data ]
§ Plenty of data sources and plajorms § Especially search is great free data source § Maintaining data integrity takes effort § Cookie technology has its limita&ons § New tag-‐less technologies emerging § Maximise iden&fica&on points § Offline can be &ed to online
June 2010 © Datalicious Pty Ltd 24
[ Genera:ng insights ]
101011010010010010101111010010010101010100001011111001010101010100101011001100010100101001101101001101001010100111001010010010101001001010010100100101001111101010100101001001001010
June 2010 © Datalicious Pty Ltd 25
[ Corporate data journey ]
June 2010 © Datalicious Pty Ltd 26
Time, Control
Soph
is&ca&o
n
Stage 1
Data Stage 2
Insights Stage 3 Ac:on
Third par&es control most data, ad hoc repor&ng only, i.e. what happened?
Data is being brought in-‐house, shiW towards insights genera&on and data mining, i.e. why did it happen?
Data is fully owned in-‐house, advanced predic&ve modelling and trigger based marke&ng, i.e. what will happen and making it happen!
[ The ideal analyst ] § Business minded – Semng realis&c improvement goals
§ Technically savvy – Bridging gap between business and IT
§ Strong sales skills – Raising awareness for the value of data
§ Seniority and experience – Needs to be taken serious across organisa&on
§ Posi&on within hierarchy – Able to analyse without loyalty conflict
June 2010 © Datalicious Pty Ltd 27
[ Process is key to success ]
June 2010 © Datalicious Pty Ltd 28
Source: Omniture Summit, MaS Belkin, 2007
Reach (Awareness)
Engagement (Interest & Desire)
Ac:on (Ac&on)
+Buzz (Sa&sfac&on)
Quan&ta&ve and qualita&ve research data
Website, call center and retail data
[ Defining metrics frameworks ]
June 2010 © Datalicious Pty Ltd 29
Social media data
Media and search data
Social media
[ Key metrics by website type ]
June 2010 © Datalicious Pty Ltd 30
Source: Omniture Summit, MaS Belkin, 2007
[ Conversion funnel 1.0 ]
June 2010
Conversion funnel Product page, add to shopping cart, view shopping cart, cart checkout, payment details, shipping informa&on, order confirma&on, etc
Conversion event
Campaign responses
© Datalicious Pty Ltd 31
[ Conversion funnel 2.0 ]
June 2010
Campaign responses (inbound spokes) Offline campaigns, banner ads, email marke&ng, referrals, organic search, paid search, internal promo&ons, etc
Landing page (hub)
Success events (outbound spokes) Bounce rate, add to cart, cart checkout, confirmed order, call back request, registra&on, product comparison, product review, forward to friend, etc
© Datalicious Pty Ltd 32
[ Addi:onal success metrics ]
June 2010 © Datalicious Pty Ltd 33
Click Through
Add To Cart
Click Through
Bounce Rate
Click Through $
Click Through
Call back requests
Store Searches ? $
$
$ Cart Checkout
Pages Per Visit
?
Video Views
June 2010 © Datalicious Pty Ltd
Exercise: Metrics framework
34
Stage Metrics Data Sources
Reach
Engagement
Ac:on
+Buzz
[ Exercise: Metrics framework ]
June 2010 © Datalicious Pty Ltd 35
Stage Metrics Data Sources
Reach Impressions, Searches
Ad Server, Google
Engagement Video Views, Product Views
Web Analy:cs Pla<orm
Ac:on Orders, Store Searches
Web Analy:cs, Call Center
+Buzz Comments, Men:ons
Social Analy:cs Pla<orm
[ Exercise: Metrics framework ]
June 2010 © Datalicious Pty Ltd 36
Customer data
[ Combining data sets ]
June 2010 © Datalicious Pty Ltd 37
3rd party data
+ The whole is greater than the sum of its parts
Web analy:cs data
[ Behaviours vs. transac:ons ]
June 2010 © Datalicious Pty Ltd 38
one-‐off collec&on of demographical data age, gender, address, etc customer lifecycle metrics and key dates profitability, expira:on, etc predic&ve models based on data mining
propensity to buy, churn, etc historical data from previous transac&ons
average order value, points, etc
CRM Profile
UPDATED OCCASIONALLY
+ tracking of purchase funnel stage
browsing, checkout, etc tracking of content preferences
products, brands, features, etc tracking of external campaign responses
search terms, referrers, etc tracking of internal promo&on responses
emails, internal search, etc
Site Behaviour
UPDATED CONTINUOUSLY
[ Store searches vs. actual loca:ons ]
June 2010 © Datalicious Pty Ltd 39
[ Enriching customer profiles ]
June 2010 © Datalicious Pty Ltd 40
Source: Hitwise, 2006
All you need is an address
[ Hitwise Mosaic segment swing ]
australia.com vs. newzealand.com australia.com vs. bulafiji.com
June 2010 © Datalicious Pty Ltd 41
Source: Hitwise, 2006
[ Hitwise Mosaic segment swing ]
australia.com vs. newzealand.com australia.com vs. newzealand.com
June 2010 © Datalicious Pty Ltd 42
Source: Hitwise, 2006
[ Single source of truth ]
June 2010 © Datalicious Pty Ltd 43
Insights Repor:ng
[ De-‐duplica:on across channels ]
June 2010 © Datalicious Pty Ltd 44
Banner Ads
Email Blast
Paid Search
Organic Search
$ Bid Mgmt
Ad Server
Email Pla<orm
Google Analy:cs
$
$
$
Central Analy:cs Pla<orm
$
$
$
June 2010 © Datalicious Pty Ltd
Thinking outside the box
45
[ Search and brand strength ]
June 2010 © Datalicious Pty Ltd 46
[ Search and the product lifecycle ]
June 2010 © Datalicious Pty Ltd 47
Nokia N-‐Series
Apple iPhone
www.google.com/trends
[ Search and media planning ]
June 2010 © Datalicious Pty Ltd 48
www.google.com/adplanner
June 2010 © Datalicious Pty Ltd 49
June 2010 © Datalicious Pty Ltd 50
June 2010 © Datalicious Pty Ltd 51
Fiat 500: Online influencing offline
Google: “slideshare fiat 500 case study” or hKp://bit.ly/lh7bx
[ Search driving offline crea:ve ]
June 2010 © Datalicious Pty Ltd 52
June 2010 © Datalicious Pty Ltd 53
June 2010 © Datalicious Pty Ltd 54
Sen:ment analysis: People vs. machine
Google: “people vs machines debate” or hKp://bit.ly/8VbtB
[ Social metrics and tools ]
June 2010 © Datalicious Pty Ltd 55
Google: ”slideshare
al:meter report” or
hKp://bit.ly/c8uYXT
Source: Social Marke&ng Analy&cs, Al&meter, 2010
June 2010 © Datalicious Pty Ltd
Exercise: Sta:s:cal significance
56
June 2010 © Datalicious Pty Ltd 57
How many survey responses do you need if you have 10,000 customers?
How many email opens do you need to test 2 subject lines if your subscriber base is 50,000?
How many orders do you need to test 6 banner execu:ons if you serve 1,000,000 banners
June 2010 © Datalicious Pty Ltd 58
How many survey responses do you need if you have 10,000 customers?
369 for each ques:on or 369 complete responses
How many email opens do you need to test 2 subject lines if your subscriber base is 50,000?
381 per subject line or 381 x 2 = 762 email opens
How many orders do you need to test 6 banner execu:ons if you serve 1,000,000 banners?
383 sales per banner execu:on or 383 x 6 = 2,298 sales
[ Summary: Genera:ng insights ]
§ Right resources and processes are key § Define a flexible metrics framework § Maintain framework to enable comparison § Combine data sets for hidden insights § Establish a single (data) source of truth § Think outside the box and across channels § Data does not equal significance
June 2010 © Datalicious Pty Ltd 59
[ Taking ac:on ]
101011010010010010101111010010010101010100001011111001010101010100101011001100010100101001101101001101001010100111001010010010101001001010010100100101001111101010100101001001001010
June 2010 © Datalicious Pty Ltd 60
[ How to drive ROI ] § Increasing revenue – Increasing overall amount of sales – Increasing the average revenue per sale
§ Reducing costs – Increasing media effec&veness – Increasing website conversion rates – Increasing online self-‐service usage
§ Improving customer experience – Reducing steps necessary to complete a task – Perceived value or quality of the final solu&on
June 2010 © Datalicious Pty Ltd 61
[ How to drive ROI ]
June 2010 © Datalicious Pty Ltd 62
Media or how to op:mise the channel mix
Targe:ng or how to increasing relevance
Tes:ng or how to maximise conversion
[ Success aKribu:on models ]
Banner Ad $100
Email Blast
Paid Search $100
Banner Ad $100
Affiliate Referral $100
Success $100
Success $100
Banner Ad
Paid Search
Organic Search $100
Success $100
Last channel gets all credit
First channel gets all credit
All channels get equal credit
Print Ad $33
Social Media $33
Paid Search $33
Success $100
All channels get par:al credit
Paid Search
June 2010 63 © Datalicious Pty Ltd
[ First vs. last click aKribu:on ]
June 2010 © Datalicious Pty Ltd 64
Chart shows percentage of channel touch points that lead to a conversion.
Neither first nor last-‐click measurement would provide true picture
Paid/Organic Search
Emails/Shopping Engines
Banner View
TV Ad
Print Ad
[ Path to purchase ]
Banner Click
SEM Generic
Partner Site
Direct Visit
June 2010 65 © Datalicious Pty Ltd
$
SEO Generic $
SEO Branded
Banner Click $
Social Media
Email Update
Direct Visit $
[ Forrester media aKribu:on ]
June 2010 © Datalicious Pty Ltd 66
Google: ”forrester aKribu:on
framework pdf” or
hKp://bit.ly/dnbnzY
Source: Forrester, 2009
[ Customer data journey ]
June 2010 © Datalicious Pty Ltd 67
To reten:on messages To transac:onal data
From suspect to To customer
From behavioural data From awareness messages
Time Time prospect
June 2010 © Datalicious Pty Ltd 68
June 2010 © Datalicious Pty Ltd 69
On-‐site segments
Off-‐site segments
[ Matching segments are key ]
June 2010 © Datalicious Pty Ltd 70
On and off-‐site targe:ng pla<orms should use iden:cal triggers to sort visitors into segments
[ Off-‐site targe:ng pla<orms ]
§ Ad servers – Google/DoubleClick – Eyeblaster – Faciliate – Atlas – Etc
§ Ad Networks – Google – Yahoo – ValueClick – Adconian – Etc
June 2010 © Datalicious Pty Ltd 71
hSp://en.wikipedia.org/wiki/Contextual_adver&sing, hSp://hubpages.com/hub/101-‐Google-‐Adsense-‐Alterna&ves, hSp://en.wikipedia.org/wiki/Central_ad_server, hSp://www.adopera&onsonline.com/2008/05/23/list-‐of-‐ad-‐servers/,
hSp://lists.econsultant.com/top-‐10-‐adver&sing-‐networks.html, hSp://www.clickz.com/3633599, hSp://en.wikipedia.org/wiki/behavioural_targe&ng
[ On-‐site targe:ng pla<orms ] § Test&Target (Omniture, Offerma&ca, TouchClarity) § Memetrics (Accenture) § Op&most (Autonomy) § KeWa (Acxiom) § AudienceScience § Maxymiser § Amadesa § Certona § SiteSpect § BTBuckets (free) § Google/DoubleClick Ad Server (free) June 2010 © Datalicious Pty Ltd 72
[ Prospect targe:ng parameters ]
June 2010 © Datalicious Pty Ltd 73
[ Vodafone affinity targe:ng ]
June 2010 © Datalicious Pty Ltd 74
Different type of visitors respond to different ads. By using category affinity targe&ng, response rates are liWed significantly across products.
Message CTR By Category Affinity
Postpay Prepay Broadb. Business
Blackberry Bold - - - + 5GB Mobile Broadband - - + - Blackberry Storm + - + + 12 Month Caps - + - +
[ Affinity targe:ng ]
§ Func&on of behavioural targe&ng – Grouping of visitors into major segments – Based on content and conversion behaviour – Ease of use vs. reduced targe&ng ability
§ Most common affini&es used – Brand affinity – Image preference – Price sensi&vity – Product affinity – Content affinity
June 2010 © Datalicious Pty Ltd 75
[ Coordinate the experience ]
June 2010 © Datalicious Pty Ltd 76
By coordina:ng the consumer’s end-‐to-‐end experience, companies could enjoy revenue increases of 10-‐20%.
Google: “get more value from digital marke:ng” or hKp://bit.ly/cAtSUN
Source: McKinsey Quarterly, 2010
Avinash Kaushik: “The principle of garbage in, garbage out applies here. […] what makes a behaviour targe<ng pla=orm <ck, and produce results, is not its intelligence, it is your ability to actually feed it the right content which it can then target […]. You feed your BT system crap and it will quickly and efficiently target crap to your customers. Faster then you could ever have yourself.”
[ Quality content is key ]
June 2010 © Datalicious Pty Ltd 77
June 2010 © Datalicious Pty Ltd
Exercise: Targe:ng matrix
78
Phase Segment A Segment B
Awareness
Considera:on
Purchase Intent
Up/Cross-‐Sell
Reten:on
[ Exercise: Targe:ng matrix ]
June 2010 © Datalicious Pty Ltd 79
Phase Segment A Segment B
Awareness Seen this?
Considera:on Great feature!
Purchase Intent Great value!
Up/Cross-‐Sell Add this!
Reten:on Discount?
[ Exercise: Targe:ng matrix ]
June 2010 © Datalicious Pty Ltd 80
Google: “change one word double conversion” or hKp://bit.ly/bpyqFp
[ ClickTale tes:ng case study ]
June 2010 © Datalicious Pty Ltd 81
[ Tes:ng pla<orms ]
§ Test&Target (Omniture, Offerma&ca, TouchClarity) § Memetrics (Accenture) § Op&most (Autonomy) § KeWa (Acxiom) § Maxymiser § Amadesa § SiteSpect § ClickTale (cheap) § Unbounce (cheap) § Google Website Op&miser (free) June 2010 © Datalicious Pty Ltd 82
[ Summary ]
§ There is no magic formula for ROI § Focus on the en&re conversion funnel § Media aSribu&on is hard but necessary § Neither first nor last click method works § Create a coordinated targeted experience § Content is always king no maSer what § Test, learn and refine con&nuously
June 2010 © Datalicious Pty Ltd 83
June 2010 © Datalicious Pty Ltd 84
Contact me [email protected]
Learn more
blog.datalicious.com
Follow us twiSer.com/datalicious