Multi-channel Analytics by Datalicious

> Multi-Channel Analytics <
Measuring and optimising
a multi-channel world

> Workshop overview
 About Datalicious
 Metrics framework
 Media attribution
 Channel integration
 Re-marketing
September 2014 © Datalicious Pty Ltd 2

101011010010010010101111010010010101010100001011111001010101
010100101011001100010100101001101101001101001010100111001010
010010101001001010010100100101001111101010100101001001001010
> About Datalicious

> Using data to fatten the funnel
“Turning data into actionable insights to widen the conversion funnel”
Media Attribution & Modeling
Maximise reach, awareness & increase ROI
Targeting & Merchandising
Improve engagement, boost loyalty
Testing & Optimisation
Remove barriers, drive sales
Boosting ROMI

> Wide range of data services
Data
Platforms
Data collection and processing
Adobe, Google Analytics, etc
Web and mobile analytics
Tag-less online data capture
Retail and call center analytics
Big data & data warehousing
Single customer view
Insights
Analytics
Data mining and modelling
Tableau, Splunk, SPSS, R, etc
Customised dashboards
Media attribution analysis
Marketing mix modelling
Social media monitoring
Customer segmentation
Action
Campaigns
Data usage and application
SiteCore, ExactTarget, etc
Targeting and merchandising
Marketing automation
CRM strategy and execution
Data driven websites
Testing programs

> Veda group strategic value add
Veda
Data
Datalicious
Technology
Products and services
for smart data driven
marketing in a multi-channel
Inivio
Analytics
world

> Best of breed technologies

> Datalicious product development
“Collecting, analysing and actioning data”
Analyse
Segment
Engage
Measure
dataexchange

> Clients across all industries

101011010010010010101111010010010101010100001011111001010101
010100101011001100010100101001101101001101001010100111001010
010010101001001010010100100101001111101010100101001001001010
> Metrics framework

> AIDA and AIDAS formulas
Old media
New media
Awareness Interest Desire Action Satisfaction
Social media

Reach
(Awareness)
Engagement
(Interest & Desire)
Conversion
(Action)
+Buzz
(Delight)
> Simplified AIDAS funnel

> Marketing is about people
People
reached
People
engaged
People
converted
People
delighted
40% 10% 1%

> Standardised roll-up metrics
People
reached
People
engaged
People
converted
People
delighted
Unique browsers,
search impressions,
TV circulation, etc
40% 10% 1%
Unique visitors,
site engagements,
video views, etc
Online sales,
online leads, store
locator searches, etc
Facebook
comments, Tweets,
ratings, support calls, etc
Response rate,
Search response rate,
TV response rate, etc
Conversion rate,
engagement rate,
checkout rate, etc
Review rate,
rating rate, comment
rate, NPS rate, etc

> Provide context with figures
People
reached
Brand vs. direct response campaign
People
engaged
People
converted
People
delighted
40% 10% 1%
New prospects vs. existing customers

> Provide context with figures
 Brand vs. direct response campaign
 New prospects vs. existing customers
 Competitive activity, i.e. none, a lot, etc
 Market share, i.e. small, medium, large, et
 Segments, i.e. age, location, influence, etc
 Channels, i.e. search, display, social, etc
 Campaigns, i.e. this/last week, month, year, etc
 Products and brands, i.e. iphone, htc, etc
 Offers, i.e. free minutes, free handset, etc
 Devices, i.e. home, office, mobile, tablet, etc

Google: “google analytics custom variables”

> Conversion funnel 1.0
September 2014
Campaign responses
Conversion funnel
Product page, add to shopping cart, view shopping cart,
cart checkout, payment details, shipping information,
order confirmation, etc
Conversion event
© Datalicious Pty Ltd 20

> Conversion funnel 2.0
September 2014
Campaign responses (inbound spokes)
Offline campaigns, banner ads, email marketing,
referrals, organic search, paid search,
internal promotions, etc
Landing page (hub)
Success events (outbound spokes)
Bounce rate, add to cart, cart checkout, confirmed order,
call back request, registration, product comparison,
product review, forward to friend, etc
© Datalicious Pty Ltd 21

> Additional success metrics
Click
Through $
Click
Through
Use additional metrics closer to the campaign origin
Add To
Cart
Click
Through
Page
Bounce
Click
Through
Call back
request
$ Cart
$
Checkout
Page
Views
?
Product
Views
Store
Search ? $

Exercise: Statistical significance

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 executions
if you serve 1,000,000 banners
Google “nss sample size calculator”

How many survey responses do you need
if you have 10,000 customers?
369 for each question or 369 complete responses
How many email opens do you need to test 2 subject lines
if your subscriber base is 50,000? And email sends?
381 per subject line or 381 x 2 = 762 email opens
How many orders do you need to test 6 banner executions
if you serve 1,000,000 banners?
383 sales per banner execution or 383 x 6 = 2,298 sales
Google “nss sample size calculator”

> Conversion metrics by category
Source: Omniture Summit, Matt Belkin, 2007

> Relative or calculated metrics
 Bounce rate
 Conversion rate
 Cost per acquisition
 Pages views per visit
 Product views per visit
 Cart abandonment rate
 Average order value

> Align metrics across channels
 Paid search response rate
= website visits / paid search impressions
 Organic search response rate
= website visits / organic search impressions
 Display response rate
= website visits / display ad impressions
 Email response rate
= website visits / emails sent
 Direct mail response rate
= (website visits + phone calls) / direct mail pieces sent
 TV response rate
= (website visits + phone calls) / (TV ad reach x frequency)

Exercise: Metrics framework

> Exercise: Metrics framework
Level Reach Engagement Conversion +Buzz
Level 1,
people
Level 2,
strategic
Level 3,
tactical
Funnel
breakdowns

> Exercise: Metrics framework
Level Reach Engagement Conversion +Buzz
Level 1,
people
People
reached
People
engaged
People
converted
People
delighted
Level 2,
strategic
Display
impressions ? ? ?
Level 3,
tactical
Interaction
rate, etc ? ? ?
Funnel
breakdowns
Existing customers vs. new prospects, products, etc

> NPS survey and page ratings
Page ratings

Google: “google analytics custom events”

> Importance of calendar events
Traffic spikes or other data anomalies without context are
very hard to interpret and can render data useless

> Potential calendar events
 Press releases
 Sponsored events
 Campaign launches
 Campaign changes
 Creative changes
 Price changes
 Website changes
 Technical difficulties

101011010010010010101111010010010101010100001011111001010101
010100101011001100010100101001101101001101001010100111001010
010010101001001010010100100101001111101010100101001001001010
> Media attribution

> Duplication across channels
Paid
Search
Banner
Ads
Email
Blast
Organic
Search
$ Bid
Mgmt
Ad
Server
Email
Platform
Google
Analytics
$
$
$

> Duplication across channels
Display
impression
Paid
search $
Ad
Server
Bid
mgmt.
Web
analytics
Display
click
Ad server
cookie
Organic
search
Bid mgmt.
cookie
Analytics
cookie
Ad server
cookie
Analytics
cookie
Analytics
cookie

> De-duplication across channels
Central
Analytics
Platform
$
$
$
$
Paid
Search
Banner
Ads
Email
Blast
Organic
Search

> Campaign flows are complex
= Paid media
= Viral elements
= Sales channels
Direct mail,
email, etc
Paid
search
Facebook
Twitter, etc
CRM
program
POS kiosks,
loyalty cards, etc
Organic
search
Home pages,
portals, etc
YouTube,
blog, etc
Landing pages,
offers, etc
PR, WOM,
events, etc
TV, print,
radio, etc
Call center,
retail stores, etc
Display ads,
affiliates, etc

Exercise: Campaign flow

> Success attribution models
Banner
Ad
Banner
Ad
$100
Organic
Search
$100
Email
Blast
Paid
Search
$100
Paid
Search
Paid
Search
Banner
Ad
$100
Affiliate
Referral
$100
Success
$100
Success
$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
partial credit

> First and last click attribution
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

> Ad clicks inadequate measure
Only a small minority of people actually click on ads, the majority
merely processes them (if at all) like any other advertising without an
immediate response so advertisers cannot rely on clicks as the sole
success measure but should instead focus on impressions delivered

> Indirect display impact

> Full purchase path tracking
Influencer Influencer Closer
$
Introducer
Paid
search
Display
ad views
TV/print
responses
Display
ad clicks
Online
leads
Affiliate
clicks
Organic
search
Social
referrals
Offline
sales
Organic
search
Social
buzz
Direct
site visits
Emails,
direct mail
Retail
visits
Lifetime
profit

> Full purchase path tracking
$
Introducer
Paid
search
Display
ad views
TV/print
responses
Display
ad clicks
Online
leads
Affiliate
clicks
Organic
search
Social
referrals
Offline
sales
Organic
search
Social
buzz
Direct
site visits
Emails,
direct mail
Retail
visits
Lifetime
profit

> Purchase path example

> Path across different segments
$
Introducer
Channel 1
Channel 1
Channel 1
Channel 2
Channel 3
Channel 2 Channel 3
Channel 4
Channel 4
Channel 2 Channel 3 Product 4
Product
A vs. B
Clients vs.
prospects
Brand vs.
direct resp.

> Understanding channel mix

What promoted your visit today?
 Recent branch visit
 Saw an ad on television
 Saw an ad in the newspaper
 Recommendation from family/friends
 […]
How likely are you to apply for a loan?
 Within the next few weeks
 Within the next few months
 I am a customer already
 […]

> Website entry survey
De-duped Campaign Report
Channel
% of
Conversions
Straight to Site 27%
SEO Branded 15%
SEM Branded 9%
SEO Generic 7%
SEM Generic 14%
Display Advertising 7%
Affiliate Marketing 9%
Referrals 5%
Email Marketing 7%
Greatest Influencer on Branded Search / STS
} Channel % of Influence
Word of Mouth 32%
Blogging & Social
Media 24%
Newspaper
Advertising 9%
Display Advertising 14%
Email Marketing 7%
Retail Promotions 14%
Conversions attributed to search terms
that contain brand keywords and direct
website visits are most likely not the
originating channel that generated the
awareness and as such conversion
credits should be re-allocated.

> Website entry survey example
In this retail example, the
exposure to retail display ads
was the biggest website traffic
driver for direct visits as well as
visits originating from search
terms that included branded
keywords – before TV, word of
mouth and print ads.

> Adjusting for offline impact
-5 -15 -10
+5 +15 +10

> Purchase path vs. attribution
 Important to make a distinction between media
attribution and purchase path tracking
– Not the same, one is necessary to enable the other
 Tracking the complete purchase path, i.e. every paid
and organic campaign touch point leading up to a
conversion is a necessary requirement to be able to
actually do media attribution or the allocation or
conversion credits back to campaign touch points
– Purchase path tracking is the data collection and
media attribution is the actual analysis or modelling

> Where to track purchase path
Web Analytics
Referral visits
Social media visits
Organic search visits
Paid search visits
Email visits, etc
Ad Server
Banner impressions
Banner clicks
+
Paid search clicks
Lacking ad impressions
Less granular & complex
Lacking organic visits
More granular & complex

> Purchase path data samples
Web Analytics data sample
LAST AD IMPRESSION > SEARCH > $$$| PV $$$
AD IMPRESSION > AD IMPRESSION > SEARCH > $$$
Ad Server data sample
01/01/2012 11:45 AD IMP YAHOO HOME $33
01/01/2012 12:00 AD IMP SMH FINANCE $33
01/01/2012 12:05 SEARCH KEYWORD -
07/01/2012 17:00 DIRECT $33
08/01/2012 15:00 $$$ $100

> Media attribution models
$
Introducer
?%
?%
?%
?%
?%
?% ?%
?%
?%
?% ?% ?%
Product
A vs. B
Prospects
vs. clients
Brand vs.
direct resp.

> Full vs. partial purchase path data
Display
impression
✖ ✔ ✔ ✔
Display
impression
Display
impression
Email
response
Search
response
✖ ✖ ✔ ✔
Display
impression
✖ ✖ ✔ ✔
Display
impression
$
Display
impression $
Display
impression
Display
impression
Direct
visit
Display
impression $
Display
impression
Search
response
Display
response
Search
response $
✖ ✔ ✔ ✔

> Full vs. partial purchase path data
Display
impression
✖ ✔ ✔ ✔
Display
impression
Display
impression
Email
response
Search
response
5% to 65% variance
in conversion attribution
for different channels due to
partial purchase path data
✖ ✖ ✔ ✔
Display
impression
✖ ✖ ✔ ✔
Display
impression
$
Display
impression $
Display
impression
Display
impression
Direct
visit
Display
impression $
Display
impression
Search
response
Display
response
Search
response $
✖ ✔ ✔ ✔

> Purchase path for each cookie
Mobile Home Work
Tablet Media Etc

Display
impression
0%
$100
Display
impression
Display
response
Search
response
0% Last click
attribution
Even
attribution
Weighted
attribution
0% 100%
25% 25% 25% 25%
X% X% Y% Z%

> Google Analytics models
 The First/Last Interaction model plus …
 The Linear model might be used if your
campaigns are designed to maintain
awareness with the customer
throughout the entire sales cycle.
 The Position Based model can be used
to adjust credit for different parts of the
customer journey, such as early
interactions that create awareness and
late interactions that close sales.
 The Time Decay model assigns the most
credit to touch points that occurred
nearest to the time of conversion. It can
be useful for campaigns with short sales
cycles, such as promotions.

Exercise: Attribution models

$
Introducer
?%
?%
?%
?%
?%
?% ?%
?%
?%
?% ?% ?%
Product
A vs. B
Prospects
vs. clients
Brand vs.
direct resp.

> Media attribution example
Even/weighted
attribution
COST PER CONVERSION
Last click
attribution

Even/weighted
attribution
?
TV/Print
? ads
?
Direct
mail
Internal
COST PER CONVERSION
Last click
attribution
?
Email
?
Website
content

Increase
spend
Increase
spend
ROI FULL PURCHASE PATH
TOTAL CONVERSION VALUE
Reduce
spend

101011010010010010101111010010010101010100001011111001010101
010100101011001100010100101001101101001101001010100111001010
010010101001001010010100100101001111101010100101001001001010
> Channel integration

> Tracking offline responses online
 Search calls to action for TV, radio, print
– Unique search term only advertised in print so all
responses from that term must have come from print
 PURLs (personalised URLs) for direct mail
– Brand.com/customer-name redirects to new URL that
includes tracking parameter identifying response as DM
 Website entry survey for direct/branded visits
– Survey website visitors that have come to site directly
or via branded search about their media habits, etc
 Combine data sets into media attribution model
– Combine raw data from online purchase path, website entry
survey and offline sales with offline media placement data in
traditional (econometric) media attribution model

> Personalised URLs for direct mail
ChrisBartens.company.com > redirect to > company.com?
utm_id=neND&
Demographics=M|35&
CustomerSegment=A1&
CustomerValue=High&
CustomerSince=2001&
ProductHistory=A6&
NextBestOffer=A7&
ChurnRisk=Low
[...]

> Search call to action for offline

> Econometric media modelling
Use of traditional econometric
modelling to measure the
impact of communications on
sales for offline channels where
it cannot be measured directly
through smart calls to action
online (and thus cookie level
purchase path data).

> Tracking offline sales online
 Email click-through
– Include offline sales flag in 1st email click-through URL after
offline sale to track an ‘assisted offline sales’ conversion
 First login after purchase
– Similar to the above method, however offline sales flag
happens via JavaScript parameter defined on 1st login
 Unique phone numbers
– Assign unique website numbers to responses from specific
channels, search terms or even individual visitors to match
offline call center results back to online activity
 Website entry survey for purchase intent
– Survey website visitors to at least measure purchase
intent in case actual offline sales cannot be tracked

> Offline sales driven by online
Fulfilment,
CRM, etc
Confirmation
email, 1st login
Advertising
campaign
Website
research
Phone
sales
Retail
sales
Online
sales
Cookie
Online sales
confirmation
Virtual sales
confirmation

> Email click-through identification
http://www.company.com/email-landing-page.html?
utm_id=neNCu&
CustomerID=12345&
Demographics=M|35&
CustomerSegment=A1&
CustomerValue=High&
ProductHistory=A6&
NextBestOffer=A7&
ChurnRisk=Low
[...]

> Login landing and exit pages
Customer data exposed in page or URL on login or logout
CustomerID=12345&
Demographics=M|35&
CustomerSegment=A1&
CustomerValue=High&
ProductHistory=A6&
NextBestOffer=A7&
ChurnRisk=Low
[...]

> Combining data sources
Website behavioural data
Campaign response data
Customer profile data
+ The whole is greater
than the sum of its parts

> Transactions plus behaviours
+
CRM Profile
one-off collection of demographical data
age, gender, address, etc
customer lifecycle metrics and key dates
profitability, expiration, etc
predictive models based on data mining
propensity to buy, churn, etc
historical data from previous transactions
average order value, points, etc
Updated Occasionally
Site Behaviour
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 promotion responses
emails, internal search, etc
Updated Continuously

> Customer profiling in action
Using website and email responses
to learn a little bite more about
subscribers at every
touch point to keep
refining profiles
and messages.

> Unique visitor overestimation
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
overestimated visitors by up to 7.6 times whilst a cookie-based
approach overestimated visitors by up to 2.3 times.
Source: White Paper, RedEye, 2007

> Maximise identification points
160%
140%
120%
100%
80%
60%
40%
20%
−−− Probability of identification through Cookies
0 4 8 12 16 20 24 28 32 36 40 44 48
Weeks

> Combining targeting platforms
On-site
targeting
Off-site
targeting
CRM

101011010010010010101111010010010101010100001011111001010101
010100101011001100010100101001101101001101001010100111001010
010010101001001010010100100101001111101010100101001001001010
> Re-marketing

> Importance of online experience
The consumer decision process is changing from linear to circular.
Consideration
set now grows
during online
research phase
which increases
importance of
user experience
during that phase
Online research

> Increase revenue by 10-20%

APPLY NOW

> Network wide re-targeting
Product A
Product B
prospect
Product A
prospect
Product A
customer
Product B Product C
Product C
prospect
Product B
prospect
Product B
customer
Product A
prospect
Product C
prospect
Product C
customer

> Network wide re-targeting
Group wide campaign with approximate impression targets by product rather than hard budget limitations
Product B
prospect
Product A
prospect
Product A
customer
Product C
prospect
Product B
prospect
Product B
customer
Product A
prospect
Product C
prospect
Product C
customer

> Story telling or ad-sequencing
$
Introducer
Message 1
Message 1
Message 1
Message 2
Message 3
Message 2 Message 3
Message 4
Message 4
Message 2 Message 3 Message 4
Product A
Product B
Product C

> Ad-sequencing in action
Marketing is about
telling stories and
stories are not static
but evolve over time
Ad-sequencing can help to
evolve stories over time the
more users engage with ads

> Targeting: Quality vs. quantity
30% new visitors with no
previous website history
aside from campaign or
referrer data of which
maybe 50% is useful
30% repeat visitors with
referral data and some
website history allowing
50% to be segmented by
content affinity
30% existing customers with extensive
profile including transactional history of
which maybe 50% can actually be
identified as individuals
10% serious
prospects
with limited
profile data

> ANZ home page targeting
ANZ home page
re-targeting and
merchandising
combined with
landing page
optimisation
delivered an
increase in offer
response and
conversion rates
with an overall
project ROI of
578%

Exercise: Re-targeting matrix

> Exercise: Re-targeting matrix
Purchase
Cycle
Segmentation based on: Search keywords,
display ad clicks and website behaviour Data
Points
Default,
awareness
Default
Research,
consideratio
n
Product
view, etc
Purchase
intent
Checkout,
chat, etc
Existing
customer
Login, email
click, etc

> Exercise: Re-targeting matrix
Purchase
Cycle
Points
Default Product A Product B
Default,
awareness
Acquisition
message D1
Acquisition
message A1
Acquisition
message B1
Default
Research,
consideratio
n
Acquisition
message D2
Acquisition
message A2
Acquisition
message B2
Product
view, etc
Purchase
intent
Acquisition
message D3
Acquisition
message A3
Acquisition
message B3
Checkout,
chat, etc
Existing
customer
Cross-sell
message D4
Cross-sell
message A4
Cross-sell
message B4
Login, email
click, etc

Google: “enable remarketing google analytics”

> Unique phone numbers
2 out of 3 callers
hang up as they
cannot get their
information fast
enough.
Unique phone
numbers can
help improve
call experience.

 1 unique phone number
– Phone number is considered part of the brand
– Media origin of calls cannot be established
– Added value of website interaction unknown
 2-10 unique phone numbers
– Different numbers for different media channels
– Exclusive number(s) reserved for website use
– Call origin data more granular but not perfect
– Difficult to rotate and pause numbers

 10+ unique phone numbers
– Different numbers for different media channels
– Different numbers for different product categories
– Different numbers for different conversion steps
– Call origin becoming useful to shape call script
– Feasible to pause numbers to improve integrity
 100+ unique phone numbers
– Different numbers for different website visitors
– Call origin and time stamp enable individual match
– Call conversions matched back to search terms

> Website call center integration
Purchase
Cycle
Points
Default Product A Product B
Default,
awareness
1300 000 001 1300 000 005 1300 000 009 Default
Research,
consideratio
n
1300 000 002 1300 000 006 1300 000 010
Product
view, etc
Purchase
intent
1300 000 003 1300 000 007 1300 000 011
Checkout,
chat, etc
Existing
customer
1300 000 004 1300 000 008 1300 000 012
Login, email
click, etc

101011010010010010101111010010010101010100001011111001010101
010100101011001100010100101001101101001101001010100111001010
010010101001001010010100100101001111101010100101001001001010
> About OptimaHub

> Combine data across channels
Behavioural data
Web / Apps / Email / Display / Phone / Social / etc
3rd party data vendors
Geo-demographics / income / social influence / etc
+
“The whole is greater
than the sum of its parts.”
Transactional data
Product holding / Lifetime value / CRM profile / etc
Repeat customers
Customers
Prospects

> The big data marketing platform
 Wide range of OptimaHub Splunk apps enables identification and
tracking of customers across channels on an individual user level
 Easy implementation and maintenance if used in conjunction
with the SuperTag container tag and mobile app SDKs
 Standardised DataCollector data format enables population of
pre-built OptimaHub Splunk dashboards without the need for
any additional configuration or complex ETL processes
 The Splunk big data platform delivers data storage, data mining
and analysis as well as data visualisation, reporting and alerts in
one (which is unique among BI platforms)
 Splunk will scale with your business

> Core OptimaHub applications
WebAnalytics – Website clickstream analysis
AppAnalytics – Mobile app clickstream analysis
SocialAnalytics – Social media activity analysis
CallAnalytics – Phone call activity analysis
RetailAnalytics – Point of sale activity analysis
SingleView – Cross-channel single customer view
MediaAttribution – Cross-channel ROI analysis

> Holy grail: Next best message
Data source
(channel 1)
Data source
(channel 2)
Data source
(channel N)
DataCollector
(standardised
data format)
R-Scripts
(predict next
best message)
OptimaHub apps
(Splunk storage,
analysis, etc)
CRM tools
(SalesForce,
Capsule, etc)
DataExchange
(API integrations
synching data)
Campaign tools
(Urban Airship,
MailChimp, etc)

OptimaHub MediaAttribution

> Additional OptimaHub apps
ErrorAnalytics – Website JavaScript error
analysis
SpeedAnalytics – Web page load speed
analysis
DisplayAnalytics – Display ad view-ability
analysis
AffiliateAnalytics – Affiliate tracking solution

> OptimaHub SuperTag integration
 Easy to implement, only requires …
– Central install of container tag and mobile SDKs
– Switching of call provider (keep same numbers)
– Katana1 Splunk cloud hosting available as well
 Easy to configure and maintain
– User friendly drag and drop user interface
– Integrations between reports and applications

Multi-channel Analytics by Datalicious

More Related Content

What's hot

Similar to Multi-channel Analytics by Datalicious

More from Datalicious

Recently uploaded

Multi-channel Analytics by Datalicious

Editor's Notes