Sei sulla pagina 1di 19

[

 Media  a(ribu,on  ]  
How  to  measure  cross-­‐channel  media  
a3ribu5on  and  access  impact  of  
banner  view-­‐throughs  
[  Overview  ]  
  Challenge:  establish  a  generic  model  to  accurately  
a3ribute  conversion  credits  to  each  channel  and  
analyse  the  true  impact  of  banner  view-­‐throughs  in  
par5cular.  
  Key  ques5ons:  Is  my  company  a3ribu5ng  the  right  
amount  of  conversions  to  each  channel  or  some  
channels  such  as  display  ads  geAng  to  much  credit?  
What  would  be  the  op5mal  channel  mix  and  spend  
to  op5mize  conversions?  
  Desired  outcome:  Iden5fy  the  op5mal  channel  mix  
and  budget  for  each  and  if  possible  reduce  media  
spend  in  some  channels  while  maintaining  
effec5veness.  
9/18/09   ©  Datalicious  Pty  Ltd   2  
[  Methodology  part  1  ]  
  Development  of  a  single  source  of  truth  that  accurately  
tracks  all  campaign  touch  points  leading  up  to  a  
conversion  event  in  one  central  repor5ng  plaRorm  to  
enable  de-­‐duplica5on  across  mul5ple  channels  and  
calcula5on  of  true  ROAS  (Return  On  Adver5sing  Spend).  
  Recorded  campaign  touch  points  should  include  
responses  from  paid  and  organic  sources  as  well  as  
online  and  offline  channels  plus  include  clicks  and  view-­‐
throughs  to  provide  maximum  accuracy.  
  Omniture  would  be  the  plaRorm  of  choice  due  to  its    
flexibility  and  accuracy  but  sa5sfying  results  could  also  be  
achieved  using  other  plaRorms  such  as  Google  Analy5cs.  

9/18/09   ©  Datalicious  Pty  Ltd   3  


[  Single  source  of  truth  ]  

data  

ROAS  
9/18/09   ©  Datalicious  Pty  Ltd   4  
[  De-­‐duplica,on  across  channels  ]  
Instead  of  recording  
Paid     Bid    
Search   Mgmt   $   conversions  for  each  channel  in  
separate  plaRorms,  ul5mately  
leading  to  sales  being  recorded  
mul5ple  5mes  due  to  the  fact  
Banner    
Ads  
Ad    
Server   $   that  more  than  one  channel  
usually  contributes  to  a  final  
conversion,  all  campaign  touch  
points  should  be  tracked  and  
Email     Email  
$   analyzed  in  one  central  
Blast   PlaGorm  
analy5cs  plaRorm  to  enable  
de-­‐duplica5on  and  comparison  
of  performance  across  
Organic   Google   channels.  
Search   Analy,cs   $  

9/18/09   ©  Datalicious  Pty  Ltd   5  


[  De-­‐duplica,on  across  channels  ]  
Instead  of  recording  
Paid    
Search   $   conversions  for  each  channel  in  
separate  plaRorms,  ul5mately  
leading  to  sales  being  recorded  
mul5ple  5mes  due  to  the  fact  
Banner    
Ads   $   that  more  than  one  channel  
usually  contributes  to  a  final  
Central  
Analy,cs   conversion,  all  campaign  touch  
PlaGorm   points  should  be  tracked  and  
Email    
$   analyzed  in  one  central  
Blast  
analy5cs  plaRorm  to  enable  
de-­‐duplica5on  and  comparison  
of  performance  across  
Organic   channels.  
Search   $  

9/18/09   ©  Datalicious  Pty  Ltd   6  


[  Methodology  part  2  ]  
  Website  entry  survey  to  determine  the  origin  of  
conversions  currently  a3ributed  to  direct  site  visits  as  
well  as  branded  organic  and  paid  search  responses  (i.e.  
search  terms  that  contain  the  brand).  
  Direct  website  visits  and  branded  searches  should  not  
receive  credit  for  conversions  as  these  channels  most  
likely  were  not  the  source  of  the  generated  awareness.  
  In  addi5on  to  recording  view-­‐throughs  in  the  central  
campaign  stacking,  an  ad  server  exposure  test  is  
recommended  to  establish  the  incremental  increase  in  
search  traffic  and  conversions  due  to  display  campaigns  
(i.e.  banners  ads  lead  to  an  increases  in  searches  if  they  
are  not  clicked  on).  

9/18/09   ©  Datalicious  Pty  Ltd   7  


[  Website  entry  survey  ]  
Greatest  Influencer  on  Branded  Search  /  STS  
Channel   %  of  influence  
Word  of  Mouth   32%   De-­‐duped  Campaign  Report  

{
Blogging  &  Social  Media   24%   Channel   %  of  conversions  

9%   Straight  to  Site   27%  


Newspaper  Adver5sing  
14%   SEO  -­‐  Branded   15%  
Display  Adver5sing  
7%   SEM  -­‐  Branded   9%  
Email  Marke5ng  
SEO  -­‐  Generic   7%  
Paid  Search  -­‐  Generic   14%  
SEM  -­‐  Generic   14%  
Conversions  a3ributed  to  search  terms   Display  Adver5sing   7%  
that  contain  brand  keywords  and  direct   Affiliate  Marke5ng   9%  
website  visits  are  most  likely  not  the   Referrals   5%  
origina5ng  channel  that  generated  the   7%  
Email  Marke5ng  
awareness  and  as  such  conversion  
credits  should  be  re-­‐allocated.    

9/18/09   ©  Datalicious  Pty  Ltd   8  


[  Ad  server  exposure  test  ]  
1   User  qualifies  for  the  display  campaign  
(if  the  user  has  already  been  tagged  go  to  step  3)  

1st  impression  

2   Audience  Segmenta,on  
10%  of  users  in  control  group,  90%  in  exposed  group  

User  tagged  with  segment  


Measurement:  
Conversions  per  
1000  unique   Control   Exposed  
visitors   (displayed  non-­‐branded  message)   (displayed  branded  message)  

N  impressions   User  remains  in  segment  

3   Control   Exposed  
(displayed  non-­‐branded  message)   (displayed  branded  message)  

9/18/09   ©  Datalicious  Pty  Ltd   9  


[  Methodology  part  3  ]  
  Reality  check  and  defini5on  of  company  wide  cookie  
expira5on  policies  for  the  various  channels  to  
guarantee  accuracy  of  a3ribu5on  model.    
  Development  of  customized  success  a3ribu5on  
model  following  the  principles  and  methods  
suggested  by  Forrester  and  including  data  from  
campaign  response  tracking,  website  entry  survey  
and  ad  server  exposure  test.  
  One-­‐off  summary  report  of  findings  plus  automa5on  
of  media  a3ribu5on  report  delivery  via  a  customized  
online  dashboard  to  increase  ma3er  awareness  
within  the  company.  

9/18/09   ©  Datalicious  Pty  Ltd   10  


[  Success  a(ribu,on  models  ]  
Banner     Paid    
Organic   Success   Last  channel  
Search  
Ad   Search  
$100   $100   gets  all  credit  

Banner    
Paid     Email     Success   First  channel  
Ad  
$100  
Search   Blast   $100   gets  all  credit  

Paid     Banner     Affiliate     Success   All  channels  get  


Search   Ad   Referral  
$100   $100   $100   $100   equal  credit  

Print     Social     Paid     Success   All  channels  get  


Ad   Media   Search  
$33   $33   $33   $100   par,al  credit  

9/18/09   ©  Datalicious  Pty  Ltd   11  


[  Sample  campaign  stacking  report  ]  
Chart  shows  
percentage  of  
channel  touch  
points  that  lead  
to  a  conversion.  

Neither  first    
nor  last-­‐click  
measurement  
would  provide  
true  picture    

9/18/09   ©  Datalicious  Pty  Ltd   12  


[  Forrester  media  a(ribu,on  ]  
Chart  shows  an  
example  only,  
a3ribu5on  model  
needs  to  be  defined  
for  each  company  
separately  based  on  
their  individual  
success  metrics  (and  
cookie  expira5on  
policies).  

9/18/09   ©  Datalicious  Pty  Ltd   13  


101011010010010010101111010010010101010100001011111001010101
010100101011001100010100101001101101001101001010100111001010
010010101001001010010100100101001111101010100101001001001010  

[  About  Datalicious  ]  
9/18/09   ©  Datalicious  Pty  Ltd   14  
[  Con,nuous  op,miza,on  ]    

Data  

Ac,on   Insights  

9/18/09   ©  Datalicious  Pty  Ltd   15  


[  Best  prac,ce  approach  ]  

Con,nuous  op,miza,on  

Research   SEM   SEO   Tes5ng   Targe5ng  

Analy,cs  framework  

9/18/09   ©  Datalicious  Pty  Ltd   16  


[  Wide  range  of  services  ]  
Data   Insights   Ac,on  

Web  Analy,cs  Solu,ons   Keyword  Research   Search  Lead  Media  

Marke,ng  System  Integra,on   Campaign  Repor,ng   Campaign  Op,misa,on  

Cross  Channel  Media  Tracking   Segmenta,on/Data  Mining   Internal  Search  Op,misa,on  

Online  Surveys/Panels   Quan,ta,ve  Research   Targe,ng/Merchandizing  

Omniture  Specialists   Market/Consumer  Trends   A/B,  Mul,variate  Tes,ng  

Google  Analy,cs  Specialists   Compe,tor  Analysis   Staff  Training/Workshops  

9/18/09   ©  Datalicious  Pty  Ltd   17  


[  Challenging  clients  ]  

9/18/09   ©  Datalicious  Pty  Ltd   18  


Ques,ons  
cbartens@datalicious.com  

Updates  
twi3er.com/datalicious  
blog.datalicious.com  

9/18/09   ©  Datalicious  Pty  Ltd   19  

Potrebbero piacerti anche