Top Banner
© Aurélie Pols 1 Amicus brief 1 : Should you measure when a user logs out? Table of Contents: To the attention of .................................................................................................... 1 Objective of this document ........................................................................................ 1 Authors ..................................................................................................................... 2 Cited sources ............................................................................................................................................................ 2 Background information ............................................................................................ 3 Description of the data ecosystem............................................................................. 5 Involved actors ........................................................................................................................................................ 5 Vocabulary ................................................................................................................................................................. 5 Legal jargon (borrowed from EU legislation) ............................................................................................ 6 Risk and potential liability ................................................................................................................................. 8 Type of content accessed (and loggedout from) ..................................................................................... 9 Reasonable client expectation .......................................................................................................................... 9 Minimal requirements to lower risk ............................................................................................................ 10 Doomsday scenario ............................................................................................................................................. 11 Conclusion ............................................................................................................... 11 To the attention of The Digital Analytics Association, more specifically Name Company Title Email Jodi McDermott comScore President XXXXXXX Bob Page HortonWorks Vice President XXXXXXX Jim Sterne Chair of the Board XXXXXXX Mike Levin DAA Executive Director XXXXXXX Objective of this document This amicus brief is intended to support the digital analytics community with the understanding of the implications of digital measurement practices from the angle of increasing Privacy, Compliance, Ethics and Security requirements. This document is not intended to hold any legal recommendations. The purpose of this document is to foster reflections and discussions within the digital analytics community about vendors’ measurement practices, ways to tackle evolving global Privacy legislation and increased feelings of lack of trust that is felt by Internet users all over the world. 1 Amicus brief or Amicus Curiae: A person (or other entity, such as state government) who is not a party to a particular lawsuit but nevertheless has a strong interest in it may be allowed, by leave of the court, to file an amicus curiae brief, a statement of particular views on the subject matter of the lawsuit. Source: http://www.merriamwebster.com/dictionary/amicus%20curiae
12
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   1  

Amicus  brief1:  Should  you  measure  when  a  user  logs  out?    Table  of  Contents:  

To  the  attention  of  ....................................................................................................  1  

Objective  of  this  document  ........................................................................................  1  

Authors  .....................................................................................................................  2  Cited  sources  ............................................................................................................................................................  2  

Background  information  ............................................................................................  3  

Description  of  the  data  ecosystem  .............................................................................  5  Involved  actors  ........................................................................................................................................................  5  Vocabulary  .................................................................................................................................................................  5  

Legal  jargon  (borrowed  from  EU  legislation)  ............................................................................................  6  Risk  and  potential  liability  .................................................................................................................................  8  Type  of  content  accessed  (and  logged-­‐out  from)  .....................................................................................  9  Reasonable  client  expectation  ..........................................................................................................................  9  Minimal  requirements  to  lower  risk  ............................................................................................................  10  Doomsday  scenario  .............................................................................................................................................  11  

Conclusion  ...............................................................................................................  11    

To  the  attention  of  The  Digital  Analytics  Association,  more  specifically  Name   Company   Title   Email  Jodi  McDermott   comScore   President   XXXXXXX  Bob  Page   HortonWorks   Vice  President   XXXXXXX    Jim  Sterne     Chair  of  the  

Board  XXXXXXX  

Mike  Levin   DAA   Executive  Director  

XXXXXXX  

Objective  of  this  document  This  amicus  brief  is  intended  to  support  the  digital  analytics  community  with  the  understanding  of  the  implications  of  digital  measurement  practices  from  the  angle  of  increasing  Privacy,  Compliance,  Ethics  and  Security  requirements.    This  document  is  not  intended  to  hold  any  legal  recommendations.    The  purpose  of  this  document  is  to  foster  reflections  and  discussions  within  the  digital  analytics  community  about  vendors’  measurement  practices,  ways  to  tackle  evolving  global  Privacy  legislation  and  increased  feelings  of  lack  of  trust  that  is  felt  by  Internet  users  all  over  the  world.  

                                                                                                               1  Amicus  brief  or  Amicus  Curiae:  A  person  (or  other  entity,  such  as  state  government)  who  is  not  a  party  to  a  particular  lawsuit  but  nevertheless  has  a  strong  interest  in  it  may  be  allowed,  by  leave  of  the  court,  to  file  an  amicus  curiae  brief,  a  statement  of  particular  views  on  the  subject  matter  of  the  lawsuit.  Source:  http://www.merriam-­‐webster.com/dictionary/amicus%20curiae    

Page 2: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   2  

Authors  Name   Company   Country   Email  Aurélie  Pols   OX3  Analytics  S.L.   Spain   [email protected]    Peter  O’Neill   L3  Analytics   UK   XXXXXX  Benjamin  Mercier  

Barclays   UK   XXXXXX    

 

Cited  sources  Name   Company   Country   Email  Simo  Ahava   Netbooster   Finland   XXXXXX  Tahir  Fayyaz   Havas  Media   UK   XXXXXX  Doug  Hall   Conversion  Works   UK   XXXXXX      Date:  January  12th  2015  Version:  5  

Page 3: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   3  

Background  information    In  October  2014,    Simo  Ahava  from  Netbooster  Finland  wrote  an  excellent  blog  post  entitled  “#GTMtips:  Once  userID,  Always  userID”  about  the  use  of  Google  Universal  Analytics’  UserID  across  sessions.    http://www.simoahava.com/gtm-­‐tips/once-­‐userid-­‐always-­‐userid/      

   The  same  day,  Peter  O’Neill  from  L3  Analytics  in  the  UK  bounced  on  the  article  and  started  a  Twitter  conversation  about  whether  a  visitor  should  continue  to  be  identified  and  measured  after  having  expressly  logged-­‐out  from  a  website  section  or  an  application.    

   Current  perception  within  the  industry:  As  clearly  shown  through  the  feedback  to  Peter  O’Neill’s  tweet,  digital  analytics  professionals  tend  to  refer  to  vendor  documentation  and  more  specifically  their  Terms  of  Use  or  policy  in  order  to  define  the  legality  of  certain  measurement  practices.  

 

   When  the  question  is  raised  to  the  vendors  and  nothing  is  found  within  the  legal  documentation,  the  next  logical  step  is  usually    to  assure  that  the  client  is  “happy”  with  the  tracking  methods.  By  client  we  define  here  the  party  that  is  effectively  using  the  vendor’s  solution  on  their  digital  properties  for  eg.  an  ecommerce,  bank,  insurance  company…    

Page 4: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   4  

   Digital  professionals  should  however  also  take  into  consideration  “reasonable  expectations”  of  visitors  of  online  properties.  As  they  are  recommending  on  measurement  best  practices  either  on  behalf  of  their  clients,  as  external  consultants,  or  for  their  employer  as  internal  digital  analysts.  

   Which  brings  to  the  most  important  point  for  the  digital  analytics  sector  and  other  players  within  this  data  ecosystem  such  as  vendors.    While  being  considered  as  a  competitive  advantage,  their  visitor  tracking  methodology  often  lacks  transparency,  potentially  harming  their  clients  and  in  the  process  those  consultants  recommending  their  very  tools.    Additionally,  while  at  the  same  time,  vendors  are  engaged  into  new  and  parallel  features  races  in  order  to  assure  adequate  alignment  with  Privacy  requirements,  this  lack  of  transparency  often  leaves  actors  second-­‐guessing.    Here  is  an  example  of  how  KissMetrics2  apparently  auto  stitches  visitor’s  data  between  sessions,  independently  of  whether  users  logged  out  (according  to  Tahir  Fayyaz  from  Havas  Media  UK).  

   It  raises  the  question  of  whether  a  choice,  the  very  feature,  actually  exists  for  the  websites  to  define  how  the  data  about  their  clients’  behavior  is  being  stitched  together.    

     

                                                                                                               2  KISSMetrics  Finalizes  Supercookies  Settlement  by  Wendy  Davis,  MediaPost,  January  2013,  http://www.mediapost.com/publications/article/191409/kissmetrics-­‐finalizes-­‐supercookies-­‐settlement.html,  last  visited  November  5th  2014  

Page 5: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   5  

Description  of  the  data  ecosystem  

Involved  actors  

Vocabulary  • “Website  owner”  is  defined  in  this  document  as  the  company  collecting  the  

data  about  their  clients  in  order  to  optimize  their  digital  properties.  Such  a  company  could  be  a  pure  digital  player  like  an  ecommerce  property  or  online  retailer,  a  bank,  a  pharmaceutical  or  insurance  company,  etc.    

• “Customer”  is  defined  as  the  visitor  to  the  digital  properties  or  apps,  which  by  interacting  with  the  properties  leaves  data  exhausts  of  preferences  in  ways  of  clicks  and  data  introduced  through  forms  and  other  logging  methods.  

• Actors  in  between  this  relationship  are  considered  “intermediaries”,  who  hold  their  own  legal  liability  within  the  data  ecosystem,  and  are  often  either  tool  vendors  &/or  agencies.  

 More  specifically,  the  eco  system  of  actors  looks  like  this:    

 Where  data  flows,  through  intermediaries,  from  visitors  towards  the  company  collecting  the  data,  from  the  customer  to  the  website  properties  in  this  case.    Depending  upon  the  type  of  data,  sector  and  geography,  the  company  collecting  the  data,  the  customer  for  digital  analytics  agencies  and  vendors,  has  certain  responsibilities  related  to  the  data  being  collected  (and  the  person  this  data  might  be  coming  from3).  

                                                                                                               3  Avoiding  any  debate  here  about  data  ownership  in  order  to  keep  this  simple  

Page 6: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   6  

 

 In  between  the  extremes  of  these  data  flows  and  related  responsibility,  lay  tools  and  agencies,  which  take  part  in  the  data  flow  and  hence  pick  up  some  of  the  responsibility.  In  a  word,  they  may  be  liable  in  case  of  issues.  Such  issues  can  be  related  to  compliance,  security  or  more  vaguely  Privacy  issues.    Tools  or  vendors  typically  waiver  their  liability  within  this  data  eco  system  through  their  Terms  of  Use  or  Terms  and  Conditions,  where  they  stipulate  correct  and  incorrect  uses  of  their  technology  whenever  possible.    After  all,  technology  is  Privacy  neutral  and  it  would  be  impossible  for  vendors  to  imagine  every  case  scenario.    What  vendors  can  decide  is:    

1. Under  which  legislation  the  data  is  stored.    2. Which  functionalities  are  developed  to  support  business  needs,  including  

possible  security,  privacy  and  compliance  requirements.  

Legal  jargon  (borrowed  from  EU  legislation)  European  Data  Protection4  legislation  attributes  roles  and  responsibilities  related  to  data  flows.    More  specifically,  EU  Privacy  legislation  talks  of  “Data  Controllers”  and  “Data  Processors”,  or  sub-­‐processors,  in  this  data  eco  system.  

                                                                                                               4  Europe  talks  of  Data  Protection  instead  of  Privacy  legislation,  which  is  more  of  a  US  focused  topic.  The  UK  sits  in  between  as  for  now,  it’s  still  part  of  Europe.  

Page 7: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   7  

 

   Intermediaries  hold  responsibilities  in  the  data  flow,  using  the  legal  term  “Data  Processors”,  or  “Data  Sub-­‐Processors”,  in  most  cases  for  digital  analytics5.    The  responsibilities  of  a  “Data  Controller”,  the  digital  property  collecting  the  data  in  the  first  place,  is  roughly  outlined  as  follows6:  

1. Inform  participants;  2. Obtain  informed  consent;  3. Ensure  that  data  held  is  accurate;  4. Delete  personal  data  when  it  is  no  longer  needed;  5. Protect  against  unauthorized  destruction,  loss,  alteration  and  disclosure;  6. Contract  with  Data  Processors  responsibly;  7. Take  care  transferring  data  out  of  Europe;  8. If  you  collect  “special”  categories  of  data,  get  specialist  advice;  9. Deal  with  any  subject  access  requests;  10. If  the  assessment  is  high  stakes,  ensure  there  is  review  of  any  automated  

decision  making;  11. Appoint  a  data  protection  officer  and  train  the  staff;  12. Work  with  supervisory  authorities  and  respond  to  complaints.  

   

                                                                                                               5  The  main  exception  is  Google  Analytics,  who  acts  as  both  a  processor  but  also  a  controller,  which  is  why  they  don’t  want  data  that  could  potentially  identify  an  individual  within  their  tool  cf.  http://www.mindyourprivacy.com/english-­‐us-­‐role-­‐playing-­‐which-­‐one-­‐are-­‐you-­‐google-­‐analytics-­‐controller-­‐or-­‐processor/?lang=en    6  Note  that  in  the  case  of  a  vendor’s  website,  the  vendor  then  takes  on  the  role  of  “Data  controller”  for  it’s  own  digital  properties  

Page 8: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   8  

Risk  and  potential  liability  Getting  back  to  the  initial  question  of  whether  a  digital  analyst  should  continue  to  track  and  measure  once  a  client  logs  out,  the  answer  is  best  expressed  in  terms  of  risk.    What  is  wrong  about  continuing  to  track  visitors  after  a  log  out  action?      The  first  risk  is  legal,  during  the  session,  the  visitor  made  an  action  like:  “stop  identifying  or/&  tracking  me”.  If  the  visitor  continues  to  browse  the  site,  he  would  expect  to  be  treated  as  an  anonymous  visitor  and  not  be  tracked.  In  most  digital  properties,  after  logging  out,  the  site  doesn’t  display  the  visitors  name  anymore,  photos  etc.  but  still  remembers  him  and  continues  to  track  his  actions  as  if  no  logout  ever  happened.    Such  risk  can  either  be  of  a  non-­‐compliance  nature  and  therefore  the  customer  –  the  data  controller  -­‐  could  encounter  financial  fines  for  non-­‐compliance  with  the  legislation  or  such  risk  might  be  related  to  client  feelings  of  creepiness.    Indeed,  a  visitor  who  did  expressively  log  out  might  “expect”  not  to  be  tracked  anymore.  Therefore  if  this  visitor  gets  re-­‐targeted  with  promotions  related  to  unlogged  navigation,  it  might  damage  the  trust  relationship  that  stands  between  the  site  and  the  visitor.  This  is  what  we  call  Creepiness.    Additionally,  risk  is  distributed  between  the  actors  within  the  data  eco  system  as  the  data  controller  can  turn  against  a  data  processor  or  sub-­‐processor  to  claim  for  compensation  in  case  of  trouble.    The  initial  data  controller  should  go  through  the  exercise  of  balancing  its  own  risk  by  asking  the  following  questions:  

1. Is  my  company  being  non-­‐compliant  by  still  tracking  an  identified  visitor  even  though  the  visitor  did  expressly  log  out?  (an  email  address  is  considered  to  be  PII  in  all  US  states  so  let’s  consider  we  are  talking  about  an  individual  as  this  is  login)    

2. If  so,  what  is  the  probability  of  being  fined  and  for  which  maximum  amount?  

3. If  not  legal  issues,  are  there  a  potential  brand  perception  issues  that  might  arise  from  this  practice  if  word  comes  out?  

4. If  so,  what  are  the  rewards  from  still  tracking  an  individual  after  they  expressly  logged  out  compared  to  this  potential  feeling  of  creepiness?  

 For  intermediaries  like  agencies  mainly,  they  should  ask  themselves  the  same  questions  but  in  the  light  of  their  own  liability.  In  fact,  agencies  should  include  as  a  mandatory  step  of  their  relationship  with  their  customers,  an  explanation  of  what  exactly  does  the  tracking  technology  collects  as  data  and  how  visitors’  sessions  are  delimited.  According  to  the  transparency  principle  and  hopefully  with  the  help  of  the  vendors,  the  web  sites  will  be  able  to  make  an  informed  decision  about  the  best  data  strategy  to  take.  

Page 9: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   9  

Type  of  content  accessed  (and  logged-­‐out  from)  A  word  of  caution  related  to  question  2:  the  probability  of  being  fined.    Certain  sectors  and  geographies  hold  higher  probabilities  of  fines  &/or  class  actions.    In  Spain  for  example,  Telcos  are  the  favorite  target  for  Data  Protection  Agencies  while  in  Italy,  credit  agencies  should  be  more  careful.    The  US,  unlike  the  EU  (who  has  overarching  Data  Protection  legislation  for  all  sectors)  holds  specific  Privacy  related  legislation  per  sector.    The  typical  ones  are  related  to  health  (HIPPA),  children  (COPPA)  but  also  banking,  energy,  video  rentals,  etc.  etc.  and  often  talk  of  the  use  of  “sensitive”  data  (health,  financial,  sexual  orientation,  political  views,  …)  on  top  of  the  initial  classification  between  the  probability  of  identifying  an  individual  or  not.  Typically  pharma  clients,  banks  and  insurances,  digital  properties  dealing  with  children,  etc.  should  be  extra  careful  with  the  choices  they  make  related  to  their  digital  analytics  infrastructure  and  measure  practices.  

Reasonable  client  expectation  Even  if  “reasonable  client  expectation”  could  be  argued  to  answer  questions  1  and  2,  for  which  legal  analysis  would  be  necessary  depending  upon  country  and  sector,  it’s  mainly  for  question  3  and  4  that  expectations  and  perception  really  starts  playing  an  active  role.      As  mentioned  in  the  previous  section  about  types  of  content,  the  question  should  be  asked  as  to  why  a  client  would  expressly  logout  of  an  application  or  online  service.    Certain  industries  would  typically  terminate  sessions  as  the  browser  is  closed  like  airlines  while  others,  like  banks,  would  often  automatically  log  out  after  a  defined  period  of  time,  if  their  clients  don’t  do  it  after  finishing  their  transactions.  On  the  other  side  of  the  spectrum,  social  sites  like  Facebook  would  keep  the  automatic  login  active  even  when  a  window  is  closed  and  opened  up  again  within  the  same  browser.      Choices  related  to  how  to  allow  logout  in  the  first  place  are  therefore  abundant  and  will  depend  upon  each  particular  situation.  Those  logout  choices  will  be  influenced  by  the  sector  the  company  is  operating  in,  security  reasons  and  possibly  analytics  practices  if  not  region.  From  there  on  follows  that  the  choice  of  continuing  to  track  a  user  even  after  they  actively  logged  out  is  not  a  black  and  white  answer  as  it  depends,  possibly  even  on  more  factors  than  those  listed  above.    And  while  companies  will  certainly  have  internal  discussions  about  how  and  when  to  close  sessions  and  log  out,  the  same  cannot  be  said  for  analytics.  The  simple  reason  for  the  difference  is  because  tracking  can  go  undetected  from  the  trained  digital  analytics  eye.  And  you  can’t  really  ask  questions  about  what  you  can’t  see.    

Page 10: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   10  

It  therefore  often  falls  upon  the  underlying  agency  that  is  consulting  related  to  the  digital  analytics  set  up  of  the  customer  to  recommend  best  practices,  with  all  the  liability  that  this  infers  as  discussed  earlier.  

Minimal  requirements  to  lower  risk  While  the  #1  responsibility  of  a  data  controller  is  to  inform  participants,  the  question  remains  open  as  to  whether  a  Privacy  Policy  should  specify  a  data  is  being  collected  even  if  a  user  logs  out.    At  the  time  of  writing,  it  doesn’t  seem  common  practice.    While  Privacy  Policies  are  clearly  evolving  in  terms  of  transparency,  tone  and  focus,  going  this  deep  into  data  collection  details  is  far  from  common  practice.  Another  point  to  raise  would  be  about  the  type  of  data  being  collected  after  logout  as  this  data  could  remain  linked  to  a  uniquely  identified  individual  or  become  part  of  a  bucketed  type  of  anonymous  data,  if  the  tools  allowed  for  such  a  distinction.      As  an  example  it  would  be  interesting  for  those  companies  to  separate  in  the  data  governance  guidance,  the  data  that  would  be  used  by  analytics  to  produce  insights,  improve  the  navigation,  make  a  better  user  experience  etc..  from  the  data  that  is  used  by  marketing  to  (re-­‐)target  customers  from  the  data  that  is  used  by  the  business  to  increase  the  sales.  That  way  it  makes  more  options  for  internal  reflections  when  deciding  about  tracking  data  after  logout.        This  functionality  was  actually  described  by  Seth  Romanow  while  at  Microsoft  at  eMetrics  in  2007  and  he  called  it  “Personamous”:  

 

Page 11: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   11  

This  set-­‐up  was  reached  through  clever  technology  and  the  use  of  webtrends  and  Omniture  at  the  time:  2  tools  and  a  lot  of  databases  in  between.    

Doomsday  scenario  Imagine  a  health  insurer  website  where  a  visitor  is  logged  in  to  request  refunds.    Let’s  now  imagine  this  visitor  logs  out  and  looks  for  a  specialized  physician  related  to  prostate  cancer.  What  would  our  industry  do  with  this  information?    The  current  Big  Data  Privacy  debate,  initiated  by  the  then  French  Data  Protection  Authority  president  Isabelle  Falque-­‐Pierrotin,  is  whether  discrimination  might  take  place  due  to  excessive  tracking.  Would  an  insurance  company  increase  its  rates  if  you  were  to  search  for  a  prostate  cancer  physician  and  fall  within  the  likelihood  of  having  prostate  cancer  (because  you’re  male  and  are  over  50  years)?    Imagine  you’re  logged  onto  a  health  website,  you  log  out  and  look  for  Viagra.  Are  you  going  to  receive  an  automatic  email  with  discount  coupons  for  Viagra  through  some  kind  of  Marketing  Automation  program  on  your  family  email  address?  

Conclusion  There  is  no  black  and  white  answer  to  the  initial  question  posed  in  this  document:  should  you  measure  when  logged  out?    The  way  data  will  be  picked  up,  stored  and  later  re-­‐used  should  be  seen  on  a  case-­‐by-­‐case  scenario  basis  where  clearly  the  responsibility  of  our  industry  is  to  promote  “Responsible  Measure  Practices”  as  pointed  out  by  Doug  Hall  at  eMetrics  London.    Not  only  the  companies  using  the  measurement  technologies  to  better  understand  their  clients  should  be  aware  of  their  responsibilities.in  terms  of  compliance  and  consumer  feelings  of  creepiness.  The  digital  analytics  vendors  and  the  specialized  consultancies  also  have  a  part  to  play  in  the  liability  of  the  digital  data  ecosystem.    Agencies  can  hedge  their  liability  by  understanding  the  consequences  of  their  recommendations  and  asking  for  more  transparency  from  vendors  as  to  how  data  is  being  collected,  stored  and  shared.  Additionally,  they  should  not  shy  away  from  asking  professional  support  in  legal  matters  related  to  compliance  with  current  and  evolving  Privacy  legislation.    Vendors  have  been  limiting  their  liability  typically  through  their  Terms  of  Use  and  will  continue  to  do  so  in  order  to  assure  technological  neutrality.    After  all,  they  cannot  be  held  responsible  for  the  use  of  their  products.  Yet  they  should  give  the  opportunity  to  digital  analysts  to  have  the  right  features  in  place  that  would  allow  for  increased  choice  and  safer  ways  of  (re)using  the  data  being  collected.  

Page 12: Privacy & Ethics: Should you measure when a user logs out?

©  Aurélie  Pols   12  

Some  actions  can  be  taken  to  improve  the  data  privacy  without  hurting  the  vision  of  analytics.    A  solution  could  be  a  reset  of  marketing  related  measurement  after  each  logout  keeping  analytics  live.    Also,  The  Universal  Analytics  userID  feature,  as  described  by  Simo  Ahava  in  his  blog  post,  is  a  great  feature,  it  might  be  worth  asking  whether  a  second  userID  to  support  Microsoft’s  Personamous  suggestion  would  not  be  worth  considering.