SlideShare ist ein Scribd-Unternehmen logo
1 von 58
Downloaden Sie, um offline zu lesen
 




                  Segmenta(on:	
  	
  
The	
  Shadowy	
  Side	
  of	
  Persona	
  Development	
  
                            UPA	
  2012	
  
                                    	
  
                    David	
  A.	
  Siegel	
  Ph.D.	
  
                   Dray	
  &	
  Associates,	
  Inc.	
  
                   Minneapolis,	
  MN	
  	
  USA	
  
                                         	
  	
  	
  	
  	
  	
  
                   david.siegel@dray.com	
  	
  	
  
                       www.dray.com	
  
                      +1	
  612	
  377	
  1980	
  
                                       	
  


                                        	
  


                                                      u     Copyright,	
  Dray	
  &	
  Associates,	
  Inc.,	
  2012	
  
                                                                                Copyright 2012
 
                       Segmentation                                	
  
                                	
  
                                	
  
                                	
  
       Market	
  Segmentation	
  	
  	
  	
  	
  	
  	
  	
  	
  User	
  Classification	
  
	
  



                                              2                                    Copyright 2012
Interlocking	
  Challenges:	
  




            Who?	
                What?	
  




                         3                    Copyright 2012
Shadowy?	
  

          4    Copyright 2012
•    Colors = dimensions
                                                                    •    Can you align them all?
                                                                    •    The most successful are
                                                                         those willing to break a
                                                                         partial alignment and start
                                                                         from scratch




Goal:	
  	
  	
  
Ø  Make	
  explicit	
  choices	
  and	
  tradeoffs,	
  whether	
  working	
  with	
  and	
  exis<ng	
  
     segmenta<on,	
  or	
  proposing	
  a	
  classifica<on	
  scheme	
  of	
  your	
  own	
  
	
  
Themes:	
  
Ø  Segmenta<on	
  as	
  a	
  subtype	
  of	
  classifica<on	
  
Ø  Classifica<on	
  =	
  selec<ng,	
  defining,	
  priori<zing,	
  and	
  combining	
  dimensions	
  to	
  
     usefully	
  divide	
  up	
  a	
  mul<-­‐dimensional	
  space	
  
Ø  Influenced	
  by	
  subjec<ve	
  choices	
  and	
  prone	
  to	
  distor<ons,	
  whether	
  done	
  
     casually	
  or	
  through	
  the	
  most	
  high-­‐powered	
  sta<s<cal	
  analysis	
  
                                                   6                                       Copyright 2012
Ø  What	
  makes	
  a	
  useful	
  classifica<on?	
  
	
  
Ø  Tensions	
  between	
  marke<ng	
  and	
  UX	
  segments	
  

Ø  The	
  paradox	
  of	
  “precision”	
  

Ø  Pros	
  and	
  cons	
  of	
  	
  

       •  Demographics	
  

       •  Occupa<onal	
  Roles	
  
	
  
       •  Psychographics	
  
	
  
     •  Behavior	
  
     	
  
Ø  Tensions	
  between	
  marke<ng	
  and	
  UX	
  segments	
  


                                              7                    Copyright 2012
Unusually	
  Clean	
  Clusters	
  




               8                     Copyright 2012
Coherence	
  Within	
  Clusters
                              	
  




               9                     Copyright 2012
Differen<a<on	
  Among	
  Clusters
                                	
  




                9                      Copyright 2012
Personas-­‐-­‐Landmarks	
  Within	
  Clusters
                                            	
  




                       11                          Copyright 2012
Dimensions	
  of	
  Difference	
  Are	
  Not	
  Givens	
  
         -­‐-­‐Even	
  when	
  they	
  describe	
  seemingly	
  obvious	
  differences	
  




                                    What	
  is	
  this?	
  



                                             12                                   Copyright 2012
Now	
  what	
  is	
  it?	
  




13                    Copyright 2012
Now?	
  




14              Copyright 2012
Now?	
  




15              Copyright 2012
The dimensions we perceive and identify depend
on
Ø  Context of comparison
Ø  What have we sampled
Ø  What distinctions we perceive or assume to be
    relevant

E.g., if our purpose was to evaluate agricultural
products in terms of potential for industrialized
production, we might have classified differently
                         16                  Copyright 2012
Informa<on	
  is	
  a	
  difference	
  that	
  
                                                       makes	
  a	
  difference.	
  	
  	
  
                                                                   	
           	
              	
         	
         	
  
                                                                   	
           	
  -­‐-­‐Gregory	
  Bateson	
  




Source: http://www.nndb.com/people/169/000100866       /
Segmenta<on	
  needs	
  to	
  point	
  to	
  different	
  ac<ons	
  that	
  are	
  available	
  to	
  us,	
  on	
  
the	
  basis	
  of	
  predicted	
  differences	
  in	
  response	
  from	
  different	
  audiences	
  or	
  
users.	
  
	
  
                                                           17                                        Copyright 2012
Different	
  differences	
  make	
  a	
  difference,	
  depending	
  on	
  what	
  
different	
  ac<ons	
  we	
  are	
  focusing	
  on.
                                                	
  
    	
  




                                          18                              Copyright 2012
Classifica<on	
  Variable	
  1	
  
                        	
  
                                              Descriptor	
                       Descriptor	
  
                	
                           Dimensions	
                       Dimensions	
  
                	
                                	
                                 	
  
          Classifica<on	
                       Ac<on	
                            Ac<on	
  
           Variable	
  2	
                   Implica<ons	
                      Implica<ons	
  
                                              Descriptor	
                       Descriptor	
  	
  
                                             Dimensions	
                       Dimensions	
  
                                                  	
                                 	
  
                                               Ac<on	
                            Ac<on	
  
                                             Implica<ons	
                      Implica<ons	
  
Ø  Not	
  necessarily	
  2	
  x	
  2,	
  or	
  even	
  factorial	
  
Ø  Choice	
  of	
  classifica<on	
  variables	
  usually	
  based	
  on	
  what	
  we	
  think	
  makes	
  cleanest	
  
    split,	
  is	
  easiest	
  to	
  detect,	
  or	
  summarizes	
  the	
  profile	
  of	
  descriptors	
  
Ø  But	
  descriptors	
  could	
  be	
  turned	
  into	
  classifiers,	
  depending	
  on	
  what	
  maers	
  

                                                           19                                            Copyright 2012
Paradox	
  of	
  Precision:	
  The	
  “Zoom	
  In”	
  Problem	
  
                                                                        Ø  Zoom in = more detailed, granular
                                                                            description
                                                                             •  More dimensions
                                                                             •  More distinctions
                                                                             •  More subgroups




Ø  Perceived	
  as	
  more	
  precise,	
  more	
  
    convincing	
  
Ø  But	
  (all	
  things	
  else	
  being	
  equal)	
  finer	
  
    grained	
  dis<nc<ons	
  become	
  more	
  
    fuzzy,	
  boundaries	
  blur	
  
Ø  A	
  law	
  of	
  nature!	
  

                                                                   20                         Copyright 2012
Case	
  in	
  point:	
  Let’s	
  zoom	
  in	
  here	
  


                        Non	
  
                                                                     Customers	
  
                     customers	
  




	
        	
  	
                               Opportunity	
     At	
  Risk	
  


                                     Aachment	
  


                                       21                                    Copyright 2012
Aachment	
  

Anything	
  we	
  do	
  to	
  improve	
  the	
  ra<o	
  of	
  people	
  in	
  our	
  sample	
  
that	
  we	
  are	
  interested	
  in	
  will	
  exclude	
  some	
  of	
  them,	
  and	
  reduce	
  
our	
  ability	
  to	
  know	
  how	
  they	
  relate	
  to	
  the	
  popula<on	
  as	
  a	
  whole	
  
                                                 22                                    Copyright 2012
With	
  drill-­‐down,	
  subgroups	
  can	
  cut	
  across	
  segments
                                                                     	
  

   Seg.	
  A	
        Seg.	
  B	
          Seg.	
  C	
        Seg.	
  D	
  




                                      23                            Copyright 2012
Case	
  Example:	
  Segments	
  based	
  on	
  abtudes	
  did	
  differ	
  in	
  composi<on.	
  But….	
  

            Seg.	
  A	
            Seg.	
  B	
           Seg.	
  C	
            Seg.	
  D	
  



  …the	
  groupings	
  
  across	
  segments	
  
  were	
  more	
  
  coherent	
  and	
  
  dis<nct	
  re:	
  usage	
  
  paern	
  




                                                   24                                 Copyright 2012
w	
  rapidly	
  
ns	
  mul<ply,	
  
n	
  simple	
  
rip<ons	
  




                     ‹#›
Overall	
  SESS	
  


w	
  rapidly	
  
ns	
  mul<ply,	
  
n	
  simple	
  
rip<ons	
  




                                           ‹#›
Overall	
  SESS	
  


w	
  rapidly	
       Age	
  
ns	
  mul<ply,	
  
n	
  simple	
  
rip<ons	
  




                                           ‹#›
Overall	
  SESS	
  


w	
  rapidly	
         Age	
  
ns	
  mul<ply,	
  
n	
  simple	
  
rip<ons	
  
                     Ethnicity	
  



                                             ‹#›
Overall	
  SESS	
  


w	
  rapidly	
         Age	
  
ns	
  mul<ply,	
  
n	
  simple	
  
rip<ons	
  
                     Ethnicity	
  



                                             ‹#›
Overall	
  SESS	
  
                                               Orienta<on	
  to	
  self-­‐service	
  

w	
  rapidly	
         Age	
  
ns	
  mul<ply,	
  
                         Net	
  Worth	
  
n	
  simple	
  
                                             Disposable	
  Income	
  
rip<ons	
  
                     Ethnicity	
  
                                                  Source	
  of	
  influence	
  

                                                                  Importance	
  of	
  Iden<ty	
  
                                                    ‹#›
Dimensions	
  apply	
  to	
  all,	
  but	
  are	
  called	
  out	
  only	
  where	
  most	
  
dis<nc<ve,	
  heightening	
  percep<on	
  of	
  difference	
  


                                         Affluent	
                        Others	
  
                    	
  
                                  • Highly	
  Influenced	
  by	
  
               	
                 family?	
                       • Highly	
  Influenced	
  
           Hispanics	
            • High	
  SES?	
                by	
  family	
  
                                  • Manage	
  own	
               	
  
                                  finances	
  on	
  line?	
  

               	
                                                 	
  
                                  • Manage	
  own	
                            ?	
  
             Others	
             finances	
  on	
  line	
  




                                                    31                                     Copyright 2012
Segments	
  Summarizing	
  Overall	
  Difference	
  in	
  Profile	
  on	
  Mul<ple	
  
                                Dimensions     	
  




Ø  Some	
  dimensions	
  differen<ate	
  more	
  strongly	
  than	
  others.	
  	
  	
  
Ø  Smaller	
  differences	
  should	
  be	
  weighted	
  less	
  
Ø  But	
  ofen	
  all	
  the	
  differences	
  become	
  equal	
  parts	
  of	
  the	
  descrip<on	
  	
  	
  	
  
                                                            32                                   Copyright 212
                                                                                                    Copyright 2012
Some<mes,	
  it	
  may	
  look	
  like	
  we	
  can	
  make	
  precise	
  dis<nc<ons	
  based	
  on	
  
small	
  differences,	
  only	
  because	
  large	
  samples	
  make	
  them	
  sta<s<cally	
  
significant.	
  	
  But	
  do	
  those	
  differences	
  maer?	
  

 Math	
  scores:	
  Yes,	
  the	
  
 distribu<ons	
  are	
  different	
  
 (assuming	
  large	
  N).	
  	
  But	
  if	
  you	
  
 made	
  dichotomous	
  decisions	
  
 based	
  on	
  gender,	
  (e.g.,	
  pubng	
  
 girls	
  in	
  low	
  math	
  group	
  and	
  
 boys	
  in	
  high	
  math	
  group)	
  you	
  
 could	
  be	
  wrong	
  large	
  %	
  of	
  
 cases.	
  	
  


                                                                    Source: http://www.wanoah.co.uk/?p=37

	
  The	
  larger	
  the	
  sample	
  it	
  takes	
  to	
  find	
  a	
  sta<s<cally	
  significant	
  difference,	
  the	
  less	
  
likely	
  it	
  is	
  to	
  have	
  a	
  prac<cal	
  significance!

                                                               33                                            Copyright 2012
Striving	
  for	
  Precision	
  on	
  Mul<ple	
  Dimensions	
  




Two	
  views,	
  	
  
same	
  points	
  in	
  space	
  à	
  
	
  
	
  
	
  


                                      34                   Copyright 2012
Back	
  to	
  our	
  unnaturally	
  clean	
  clusters:	
  




                                    35                       Copyright 2012
Imagine	
  they	
  are	
  really	
  in	
  3	
  dimensions,	
  but	
  we	
  have	
  only	
  viewed	
  
them	
  from	
  one	
  angle	
  (i.e.,	
  only	
  focusing	
  on	
  2	
  dimensions)	
  




                                          36                                      Copyright 2012
Now	
  we	
  rotate.	
  Same	
  points	
  different	
  views—clusters	
  smear	
  out.	
  	
  	
  




                                               37                                   Copyright 2012
In	
  these	
  dimensions,	
  	
  clusters	
  are	
  broader,	
  have	
  different	
  members,	
  
personas	
  not	
  as	
  “well	
  placed”	
  to	
  represent	
  them.	
  




   Op<mizing	
  groupings	
  on	
  some	
  dimensions,	
  tends	
  to	
  “smear”	
  them	
  on	
  

                                                    38                                     Copyright 2012
Now	
  imagine	
  if	
  we	
  had	
  started	
  off	
  with	
  a	
  more	
  realis<c	
  set	
  of	
  clusters
—just	
  slight	
  varia<ons	
  in	
  density	
  –	
  because	
  on	
  many	
  of	
  the	
  
dimensions	
  we	
  care	
  about,	
  people	
  don’t	
  fall	
  into	
  such	
  discreet	
  groups.	
  




                                                     39                                       Copyright 2012
Is	
  vanilla	
  ice	
  cream	
  
                                   	
  
  more	
  like	
  chocolate	
  milk	
   	
  
     or	
  banana	
  yogurt?     	
  




                                                    ?


Needed:	
  a	
  way	
  of	
  combining	
  differences	
  
on	
  mul<ple	
  dimensions	
  into	
  a	
  judgment	
  
of	
  overall	
  similarity	
  and	
  difference.	
  	
  

                                               40          Copyright 2012
Ø  Sta<s<cal	
  approaches	
  to	
  building	
  clusters	
  usually	
  try	
  to	
  manage	
  problem	
  of	
  
     over-­‐op<mizing	
  on	
  some	
  dimensions	
  and	
  smearing	
  on	
  others	
  
 Ø  Use	
  “distance”	
  to	
  represent	
  “difference”	
  	
  


                              Heavy	
  user	
  of	
  very	
  
                                few	
  features	
  	
  
                                                                                                    Cluster	
  2	
  
                                                                                                 Group	
  A	
  =	
  Heavy	
  
Y	
  =	
  overall	
                                                                               users	
  of	
  many	
  
usage	
  (me                                                                                        features	
  	
  
                                                                  Which	
  
                                                                 group	
  is	
  
                                                                 this	
  one	
  
                                                                most	
  like?	
  
                        Group	
  B	
  =1	
  ight	
  to	
  medium	
  
                        Cluster	
   	
  L
                          users	
  of	
  few	
  features	
  	
  

                                                                X	
  =	
  number	
  of	
  features	
  used	
  
                                                                 41                                              Copyright 2012
But	
  how	
  do	
  we	
  measure	
  distance?	
  You	
  get	
  to	
  choose,	
  e.g.:	
  
  Euclidean:	
  Hypotenuse	
  of	
  difference	
  on	
  x	
  and	
  difference	
  on	
  y	
  
  	
  	
  	
  	
  ΔX	
  +	
  ΔY:	
  Sum	
  of	
  the	
  differences	
  on	
  each	
  separate	
  dimension	
  
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  Both	
  make	
  intui<ve	
  sense,	
  but	
  give	
  different	
  results!	
  

                                                                                                       The	
  two	
  	
  methods	
  
 	
  	
      	
  	
     	
  	
      	
  	
         	
  	
      	
  	
          	
  	
  
                                                                                                       assign	
  the	
  point	
  to	
  
 	
  	
      	
  	
     	
  	
      	
  	
         	
  	
      	
  	
          	
  	
                  different	
  clusters.	
  
                                                  Euclidean	
  distance	
  ≈	
  4.6	
  
                                                           ΔX	
  +	
  ΔY	
  distance	
  ≈	
  6.5	
  
 	
  	
      	
  	
     	
  	
      	
  	
        	
  	
              	
  	
    	
  	
  
                                                                                                       Also,	
  the	
  dimensions	
  
 	
  	
      	
  	
     	
  	
      	
  	
         	
  	
      	
  	
          	
  	
  
                                                                                                       should	
  be	
  weighted	
  
                                                              Centroid	
  of	
  Group	
  A	
           differently	
  based	
  on:	
  
 	
  	
      	
  	
     	
  	
      	
  	
         	
  	
      	
  	
          	
  	
                  Ø  Are	
  they	
  scaled	
  the	
  
                                                                                                           same?	
  
 	
  	
      	
  	
     	
  	
      	
  	
         	
  	
      	
  	
          	
  	
  
                                                                                                       Ø  Are	
  they	
  measured	
  
                                               Euclidean	
  distance	
  ≈	
  5.94	
  
 	
  	
      	
  	
     	
  	
      	
  	
  
                                                 ΔX	
  +	
  Δ	
  Y	
  distance	
  ≈	
  6	
  
                                                 	
  	
          	
       	
  	
                           equally	
  reliably?	
  
                                                                                                       Ø  Are	
  they	
  equally	
  
 	
  	
      	
  	
     	
  	
      	
  	
         	
  	
      	
  	
          	
  	
  
                                                                                                           good	
  predictors	
  of	
  
    Centroid	
  of	
  Group	
  B	
                                                                         something	
  we	
  care	
  
 	
  	
      	
  	
     	
  	
      	
  	
         	
  	
      	
  	
          	
  	
  

                                                                                                           about?	
  	
  
 	
  	
      	
  	
     	
  	
      	
  	
         	
  	
      	
  	
          	
  	
  


                                                                          42                                              Copyright 2012
Exaggera<ng	
  Dis<nc<veness	
  



                                                These	
  look	
  dis<nct,	
  but	
  most	
  of	
  
                                                 Cluster	
  2embers	
  have	
  a	
  lot	
  in	
  
                                                    their	
  m 	
  
                                                          common	
  on	
  one	
  or	
  both	
  
                                                                             dimensions	
  	
  

Y	
  




           Cluster	
  1	
  

                                   X
                                           43                              Copyright 2012
Prac<cal	
  Criteria	
  for	
  Priori<zing,	
  Weigh<ng	
  and	
  Combining	
  
       Dimensions	
  

Ø        How	
  efficiently	
  they	
  let	
  you	
  divide	
  the	
  sample	
  into	
  categories	
  	
  
Ø        Whether	
  there	
  is	
  a	
  clear	
  breakpoint	
  or	
  threshold	
  effect	
  on	
  other	
  variables	
  
Ø        Ease	
  of	
  defini<on,	
  measurement	
  
Ø        Ease	
  of	
  loca<ng	
  real	
  representa<ves	
  when	
  you	
  want	
  to	
  study	
  group	
  in	
  
          more	
  depth	
  
Ø        Amount	
  of	
  varia<on	
  on	
  the	
  dimension	
  
Ø        Amount	
  of	
  independent	
  informa<on	
  added,	
  how	
  much	
  heterogeneity	
  
          the	
  dimension	
  accounts	
  for	
  
Ø        Usefulness	
  as	
  proxy	
  for	
  harder	
  to	
  measure	
  variables	
  
Ø        Availability	
  of	
  external	
  informa<on	
  sources	
  for	
  es<ma<ng	
  prevalence	
  
Ø        Power	
  as	
  a	
  predictor	
  of	
  differen<al	
  response	
  	
  
Ø        Prac<cal	
  ability	
  to	
  act	
  differen<ally	
  depending	
  on	
  where	
  on	
  the	
  
          dimension	
  people	
  fall	
  
	
                                                           44                                             Copyright 2012

	
  
Demographic	
  Segmenta<on	
  




               45                Copyright 2012
Demographic	
  Segmenta<on	
  	
  
Ø  Ofen	
  cri<cized	
  as	
  selec<on	
  criteria	
  for	
  usability	
  studies	
  
Ø  But	
  demographic	
  variables	
  have	
  some	
  advantages	
  
       •    Rela<vely	
  easily	
  defined,	
  measured,	
  detected,	
  sized	
  	
  
       •    Easy	
  to	
  locate	
  real	
  representa<ves	
  when	
  	
  
       •    Amount	
  of	
  varia<on	
  can	
  be	
  great	
  
       •    Informa<on	
  added	
  at	
  lile	
  cost,	
  makes	
  them	
  good	
  proxies	
  	
  
       •    Many	
  products	
  designed	
  for	
  targeted	
  demographics	
  	
  
       •    Many	
  aspects	
  of	
  life	
  may	
  correlate	
  with	
  demographic	
  dis<nc<ons,	
  so	
  
            can	
  have	
  power	
  as	
  a	
  predictor	
  of	
  differen<al	
  response,	
  needs	
  	
  
       •    Prac<cal	
  ability	
  to	
  act	
  differen<ally	
  toward	
  them	
  for	
  messaging,	
  sales	
  
            channels,	
  etc.	
  
       •    First	
  level	
  filter	
  when	
  you	
  don’t	
  yet	
  know	
  enough	
  to	
  be	
  more	
  nuanced	
  
	
  
	
  
                                                          46                                          Copyright 2012
Occupa<onal	
  Segmenta<on
 Occupa<onal	
  Segmenta<on	
  

Ø    Profession
Ø    Abstract, higher order category (e.g., “knowledge worker,”
      “entrepreneur”
Ø    General functional area: operations, customer service, finance, IT
Ø    Specific roles
Ø    Hierarchy: “Executive,” “Manager,” “Supervisor,” “Front line worker”
Ø    Context focused: Industry or industry type, company size, business
      model, organizational structure




                                       48                       Copyright 2012
Occupa<onal	
  Segmenta<on:	
  Issues	
  
Ø  Varying degrees of standardization in nomenclature,
    function, and job design
Ø  Can your domain knowledge, focus, and sample size
    compensate for the “zoom in” problem?
Ø  Functional labels can be very difficult to define:
  •  What is a “knowledge worker”?
  •  What is a “power user”?




                                   49              Copyright 2012
u  What	
  is	
  a	
  “knowledge	
  worker?”
                                                                         	
  



                                                                         	
  
	
  	
  	
  	
  	
  	
  	
  	
  Controller	
  (finance) 	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  	
  Logis<cs	
  manager	
  
u  	
  •  Focus:	
  high-­‐level	
  processes	
  to	
  
                           	
                                                                             •  Focus:	
  tac(cal,	
  opera(onal	
  
               manage	
  financial	
  risk	
  
          	
                                                                             	
  
          •  Priority:	
  Preven(on	
  of	
  low	
                                       •  Priority:	
  Increase	
  efficiency,	
  
               probability	
  events	
                                                        ensure	
  smooth	
  opera(on	
  
          	
                                                                             	
  
          •  Decisions	
  based	
  on	
                                                  •  Needs	
  quan(ta(ve	
  data	
  to	
  
               professional	
  judgment,	
                                                    manage	
  processes,	
  look	
  for	
  
               knowledge	
  of	
  best	
  prac(ce	
                                           improvement	
  opportuni(es	
  

          •  Sets	
  policy	
  for	
  long	
  term	
                                     •  Manages	
  processes	
  in	
  real	
  
                                                                                            (me	
  
                                                                            50                                                        Copyright 2012
Psychographic	
  Segmenta<on	
  

Ø  Abtudes,	
  preferences,	
  values	
  
Ø  Intended	
  to	
  predict	
  “resonance”	
  for	
  messaging	
  
Ø  Also	
  ofen	
  emphasized	
  in	
  personas	
  for	
  broad,	
  
      generalizable	
  implica<ons	
  
Ø    How	
  strongly	
  do	
  they	
  relate	
  to	
  or	
  predict	
  usage	
  
      paern	
  or	
  other	
  behaviors?	
  
Ø    Are	
  they	
  really	
  more	
  “stable”	
  than	
  behaviors?	
  	
  
Ø    How	
  hard	
  are	
  they	
  to	
  measure	
  reliably	
  and	
  validly	
  
Ø    Self	
  report	
  versus	
  behavioral	
  self-­‐iden<fica<on	
  

                                           51                              Copyright 2012
Behavioral	
  Self-­‐Iden<fica<on	
               	
  
                                           	
  

What	
  can	
  you	
  say	
  about	
  psychographics	
  (e.g.,	
  preferences)	
  of	
  
                 people	
  who	
  gather	
  In	
  these	
  venues?  	
  




                                          52                                    Copyright 2012
People	
  Who	
  Choose	
  These	
  Periodicals?	
  




                         53                            Copyright 2012
Behavioral	
  Segmenta<on	
  
Ø  Self-­‐reported	
  versus	
  observed	
  
Ø  Purchasing	
  behaviors	
  	
  
Ø  Usage	
  behaviors:	
  Amount?	
  Variety?	
  Qualita<ve	
  
    paern?	
  
Ø  Expressed	
  behavioral	
  inten<ons:	
  	
  
      •  How	
  immediate?	
  
      •  Evidence	
  of	
  preliminary	
  steps	
  to	
  confirm?	
  	
  
Ø  Evaluate	
  degree	
  of	
  demonstrated	
  associa<on	
  with	
  
        behavior	
  of	
  ul<mate	
  interest	
  


                                              54                           Copyright 2012
u  	
  	
  
 
                                   	
  
         Marke<ng	
  Segments	
  &	
  UX	
  Categories:	
  The	
  Ideal
                                                                      	
  

                                                                                              Time	
  
                                                                              Non-­‐
                                                                             Target	
  	
                          Target	
  	
  




                                                                                                UX	
  delivers	
  promised	
  value	
  
                                                                                                (and	
  more)	
  à	
  sa<sfac<on,	
  
                                                                                                            reten<on	
  

 Targeted	
  value	
  messaging	
            Purchase	
  decision	
  
increases	
  concentra<on	
  of	
     process	
  filters	
  out	
  most	
  
     poten<al	
  buyers	
             of	
  non-­‐target	
  popula<on	
  



                                                         55                                                            Copyright 2012
 
                                           	
  
                 Marke<ng	
  Segments	
  &	
  UX	
  Categories:	
  The	
  Ideal
                                                                              	
  
Warning:	
  	
                                                                        Time	
  
Ø  This	
  is	
  most	
  likely	
  when	
  Market	
  Segmenta<on	
  and	
  UX	
  
      categoriza<on	
  map	
  to	
  each	
  other	
  	
  
Ø    But	
  market	
  segmenta<on	
  guides	
  strategies	
  for	
  ini<al	
  
      filtering,	
  rather	
  than	
  ongoing	
  experience,	
  so	
  relevant	
  and	
  
      available	
  dis<nc<ons	
  in	
  ac<on	
  may	
  be	
  different	
  
Ø    UX	
  has	
  to	
  provide	
  extended	
  sa<sfac<on	
  over	
  a	
  range	
  of	
  
      encounters	
  for	
  each	
  user	
  
Ø    UX	
  has	
  more	
  at	
  stake	
  in	
  each	
  touch	
  point,	
  because	
  goal	
  is	
  
      engagement	
  for	
  an	
  already-­‐filtered	
  audience	
  
Ø    Therefore,	
  UX	
  may	
  introduce	
  deeper	
  and/or	
  transverse	
  
      dis<nc<ons	
  essaging	
   filters	
  out	
  most	
  of	
  non-­‐target	
  
        Targeted	
  value	
  m
       increases	
  concentra<on	
  of	
  
                                           Purchase	
  decision	
  process	
     UX	
  delivers	
  promised	
  value	
  
                                                                                 (and	
  more)	
  à	
  sa<sfac<on,	
  
              poten<al	
  buyers	
                  popula<on	
                                   reten<on	
  


                                                           56                                             Copyright 2012
Tips	
  
Ø  Method	
  triangula<on:	
  	
  
    Ø  Start	
  with	
  criterion	
  groups	
  (differences	
  you	
  really	
  care	
  about)	
  and	
  then	
  look	
  for	
  
        differen<ators.	
  	
  	
  
    Ø  Start	
  with	
  possible	
  differen<ators	
  and	
  test	
  to	
  see	
  if	
  they	
  do	
  predict	
  differences	
  
        you	
  really	
  care	
  about.	
  
Ø  Test	
  dis<nc<ons	
  among	
  segments	
  that	
  people	
  already	
  believe	
  
    in	
  to	
  validate	
  that	
  they	
  really	
  do	
  predict	
  something	
  important	
  
    and	
  ac<onable	
  	
  
Ø  Par<al	
  alignment	
  on	
  a	
  few	
  variables	
  of	
  different	
  types	
  may	
  be	
  
    more	
  robust	
  and	
  useful	
  than	
  than	
  op<mizing	
  for	
  “clean”	
  
    dis<nc<ons	
  	
  
Ø  Priori<ze	
  dimensions	
  based	
  on	
  both	
  prac<cal	
  and	
  conceptual	
  
    tradeoffs	
  


                                                             57                                              Copyright 2012
More	
  Tips
                                                      	
  
Ø  Test	
  dis<nc<ons	
  across	
  mul<ple	
  studies,	
  or	
  do	
  cross-­‐
      valida<on	
  within	
  your	
  sample	
  by	
  splibng	
  it.	
  
Ø    Consider	
  impact	
  of	
  variables	
  one	
  at	
  a	
  <me	
  rather	
  than	
  only	
  in	
  
      combina<ons,	
  to	
  reduce	
  risk	
  of	
  illusory	
  precision	
  
Ø    Try	
  to	
  work	
  within	
  exis<ng	
  segments,	
  but	
  be	
  prepared	
  to	
  
      show	
  how	
  different	
  contexts	
  may	
  make	
  transverse	
  segments	
  
      more	
  or	
  less	
  relevant	
  	
  
Ø    Studying	
  pre-­‐defined	
  segments	
  one	
  at	
  a	
  may	
  blind	
  you	
  to	
  
      subgroups	
  that	
  are	
  similar	
  across	
  segments-­‐include	
  
      contras<ng	
  hypothesized	
  segments	
  into	
  samples	
  within	
  or	
  
      across	
  studies	
  
Ø    Don’t	
  expect	
  the	
  “average”	
  differences	
  of	
  segments	
  to	
  show	
  
      up	
  in	
  small	
  samples.	
  
                                                  58                                     Copyright 2012
David	
  A.	
  Siegel	
  Ph.D.	
  
Dray	
  &	
  Associates,	
  Inc.	
  
Minneapolis,	
  MN	
  	
  USA	
  
                  	
  	
  	
  	
  	
  	
  
  david.siegel@dray.com	
  	
  	
  
      www.dray.com	
  
     +1	
  612	
  377	
  1980	
  

Weitere ähnliche Inhalte

Andere mochten auch

Andere mochten auch (12)

CoagSurveyIntro
CoagSurveyIntroCoagSurveyIntro
CoagSurveyIntro
 
My vacations
My vacationsMy vacations
My vacations
 
Java server pages
Java server pagesJava server pages
Java server pages
 
Wordpress Manual Document
Wordpress Manual DocumentWordpress Manual Document
Wordpress Manual Document
 
Penanganan Limbah Padat IV
Penanganan Limbah Padat IVPenanganan Limbah Padat IV
Penanganan Limbah Padat IV
 
UX 101 - TajRiba | UX Month Nairobi
UX 101 - TajRiba | UX Month NairobiUX 101 - TajRiba | UX Month Nairobi
UX 101 - TajRiba | UX Month Nairobi
 
Java Basics
Java BasicsJava Basics
Java Basics
 
Penanganan Limbah Padat III
Penanganan Limbah Padat IIIPenanganan Limbah Padat III
Penanganan Limbah Padat III
 
PHP slides
PHP slidesPHP slides
PHP slides
 
Penanganan Limbah Padat I
Penanganan Limbah Padat IPenanganan Limbah Padat I
Penanganan Limbah Padat I
 
Penanganan Limbah Padat II
Penanganan Limbah Padat IIPenanganan Limbah Padat II
Penanganan Limbah Padat II
 
ADO.NET
ADO.NETADO.NET
ADO.NET
 

Ähnlich wie Segmentation - The Shadowy Side of Persona Development

[Webinar Slides] Align Your Marketing Mix to Your Buyers’ Journey: Solution S...
[Webinar Slides] Align Your Marketing Mix to Your Buyers’ Journey: Solution S...[Webinar Slides] Align Your Marketing Mix to Your Buyers’ Journey: Solution S...
[Webinar Slides] Align Your Marketing Mix to Your Buyers’ Journey: Solution S...Optify
 
Cogent Company.Business Discovery
Cogent Company.Business DiscoveryCogent Company.Business Discovery
Cogent Company.Business DiscoveryMarc Hoppers
 
Webinar Deck: GICs vs. Service Providers: Who is Winning?
Webinar Deck: GICs vs. Service Providers: Who is Winning?Webinar Deck: GICs vs. Service Providers: Who is Winning?
Webinar Deck: GICs vs. Service Providers: Who is Winning?Everest Group
 
Webinar - The Connected Company
Webinar  - The Connected CompanyWebinar  - The Connected Company
Webinar - The Connected CompanyDachis Group
 
The MarkeTech Group - Scientific Method Webinar
The MarkeTech Group - Scientific Method WebinarThe MarkeTech Group - Scientific Method Webinar
The MarkeTech Group - Scientific Method WebinarThe MarkeTech Group
 
Keeping Your Project on Track Using the DEADLINES Model
Keeping Your Project on Track Using the DEADLINES ModelKeeping Your Project on Track Using the DEADLINES Model
Keeping Your Project on Track Using the DEADLINES ModelMarigold_Consulting
 
Differentiate or Die, Secrets of Silicon Valley - Presentation by Bob Wright
Differentiate or Die, Secrets of Silicon Valley - Presentation by Bob WrightDifferentiate or Die, Secrets of Silicon Valley - Presentation by Bob Wright
Differentiate or Die, Secrets of Silicon Valley - Presentation by Bob WrightProductNation/iSPIRT
 
How To Give A Good Presentation -- Getting Your Audience To Listen!
How To Give A Good Presentation -- Getting Your Audience To Listen!How To Give A Good Presentation -- Getting Your Audience To Listen!
How To Give A Good Presentation -- Getting Your Audience To Listen!Blue Elephant Consulting
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010samuelkew
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010BHewlett
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010joshirving
 
Granger Reis Brochure
Granger Reis BrochureGranger Reis Brochure
Granger Reis Brochureluke_small15
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010SVowles
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010ToriRogers
 
Content Measurement: Quantifying Content Contribution - Content2Conversion Co...
Content Measurement: Quantifying Content Contribution - Content2Conversion Co...Content Measurement: Quantifying Content Contribution - Content2Conversion Co...
Content Measurement: Quantifying Content Contribution - Content2Conversion Co...G3 Communications
 

Ähnlich wie Segmentation - The Shadowy Side of Persona Development (20)

[Webinar Slides] Align Your Marketing Mix to Your Buyers’ Journey: Solution S...
[Webinar Slides] Align Your Marketing Mix to Your Buyers’ Journey: Solution S...[Webinar Slides] Align Your Marketing Mix to Your Buyers’ Journey: Solution S...
[Webinar Slides] Align Your Marketing Mix to Your Buyers’ Journey: Solution S...
 
Cogent Company.Business Discovery
Cogent Company.Business DiscoveryCogent Company.Business Discovery
Cogent Company.Business Discovery
 
Webinar Deck: GICs vs. Service Providers: Who is Winning?
Webinar Deck: GICs vs. Service Providers: Who is Winning?Webinar Deck: GICs vs. Service Providers: Who is Winning?
Webinar Deck: GICs vs. Service Providers: Who is Winning?
 
Reputation resiliency drj 3.28.2012 final
Reputation resiliency drj 3.28.2012 finalReputation resiliency drj 3.28.2012 final
Reputation resiliency drj 3.28.2012 final
 
Webinar - The Connected Company
Webinar  - The Connected CompanyWebinar  - The Connected Company
Webinar - The Connected Company
 
The MarkeTech Group - Scientific Method Webinar
The MarkeTech Group - Scientific Method WebinarThe MarkeTech Group - Scientific Method Webinar
The MarkeTech Group - Scientific Method Webinar
 
Keeping Your Project on Track Using the DEADLINES Model
Keeping Your Project on Track Using the DEADLINES ModelKeeping Your Project on Track Using the DEADLINES Model
Keeping Your Project on Track Using the DEADLINES Model
 
Visual Literacy Week 2 (of 6) Slides
Visual Literacy Week 2 (of 6) SlidesVisual Literacy Week 2 (of 6) Slides
Visual Literacy Week 2 (of 6) Slides
 
ConceptOfSolutioners
ConceptOfSolutionersConceptOfSolutioners
ConceptOfSolutioners
 
Differentiate or Die, Secrets of Silicon Valley - Presentation by Bob Wright
Differentiate or Die, Secrets of Silicon Valley - Presentation by Bob WrightDifferentiate or Die, Secrets of Silicon Valley - Presentation by Bob Wright
Differentiate or Die, Secrets of Silicon Valley - Presentation by Bob Wright
 
How To Give A Good Presentation -- Getting Your Audience To Listen!
How To Give A Good Presentation -- Getting Your Audience To Listen!How To Give A Good Presentation -- Getting Your Audience To Listen!
How To Give A Good Presentation -- Getting Your Audience To Listen!
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010
 
Granger Reis Brochure
Granger Reis BrochureGranger Reis Brochure
Granger Reis Brochure
 
Gr Brochure
Gr BrochureGr Brochure
Gr Brochure
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010
 
Gr Brochure Aug 2010
Gr Brochure Aug 2010Gr Brochure Aug 2010
Gr Brochure Aug 2010
 
Content Measurement: Quantifying Content Contribution - Content2Conversion Co...
Content Measurement: Quantifying Content Contribution - Content2Conversion Co...Content Measurement: Quantifying Content Contribution - Content2Conversion Co...
Content Measurement: Quantifying Content Contribution - Content2Conversion Co...
 
Exxon slr preso 10 11-12
Exxon slr preso 10 11-12Exxon slr preso 10 11-12
Exxon slr preso 10 11-12
 

Kürzlich hochgeladen

Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherRemote DBA Services
 
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...Principled Technologies
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityPrincipled Technologies
 
HTML Injection Attacks: Impact and Mitigation Strategies
HTML Injection Attacks: Impact and Mitigation StrategiesHTML Injection Attacks: Impact and Mitigation Strategies
HTML Injection Attacks: Impact and Mitigation StrategiesBoston Institute of Analytics
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processorsdebabhi2
 
Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businesspanagenda
 
GenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdfGenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdflior mazor
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfsudhanshuwaghmare1
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024The Digital Insurer
 
AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAndrey Devyatkin
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...Martijn de Jong
 
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Strategize a Smooth Tenant-to-tenant Migration and Copilot TakeoffStrategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoffsammart93
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MIND CTI
 
Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024The Digital Insurer
 
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data DiscoveryTrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data DiscoveryTrustArc
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyKhushali Kathiriya
 
Top 10 Most Downloaded Games on Play Store in 2024
Top 10 Most Downloaded Games on Play Store in 2024Top 10 Most Downloaded Games on Play Store in 2024
Top 10 Most Downloaded Games on Play Store in 2024SynarionITSolutions
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...apidays
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FMESafe Software
 
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsTop 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsRoshan Dwivedi
 

Kürzlich hochgeladen (20)

Strategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a FresherStrategies for Landing an Oracle DBA Job as a Fresher
Strategies for Landing an Oracle DBA Job as a Fresher
 
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
HTML Injection Attacks: Impact and Mitigation Strategies
HTML Injection Attacks: Impact and Mitigation StrategiesHTML Injection Attacks: Impact and Mitigation Strategies
HTML Injection Attacks: Impact and Mitigation Strategies
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
 
Why Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire businessWhy Teams call analytics are critical to your entire business
Why Teams call analytics are critical to your entire business
 
GenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdfGenAI Risks & Security Meetup 01052024.pdf
GenAI Risks & Security Meetup 01052024.pdf
 
Boost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdfBoost Fertility New Invention Ups Success Rates.pdf
Boost Fertility New Invention Ups Success Rates.pdf
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of Terraform
 
2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...2024: Domino Containers - The Next Step. News from the Domino Container commu...
2024: Domino Containers - The Next Step. News from the Domino Container commu...
 
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Strategize a Smooth Tenant-to-tenant Migration and Copilot TakeoffStrategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
Strategize a Smooth Tenant-to-tenant Migration and Copilot Takeoff
 
MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024
 
Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024
 
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data DiscoveryTrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
TrustArc Webinar - Unlock the Power of AI-Driven Data Discovery
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : Uncertainty
 
Top 10 Most Downloaded Games on Play Store in 2024
Top 10 Most Downloaded Games on Play Store in 2024Top 10 Most Downloaded Games on Play Store in 2024
Top 10 Most Downloaded Games on Play Store in 2024
 
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
Apidays Singapore 2024 - Building Digital Trust in a Digital Economy by Veron...
 
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers:  A Deep Dive into Serverless Spatial Data and FMECloud Frontiers:  A Deep Dive into Serverless Spatial Data and FME
Cloud Frontiers: A Deep Dive into Serverless Spatial Data and FME
 
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live StreamsTop 5 Benefits OF Using Muvi Live Paywall For Live Streams
Top 5 Benefits OF Using Muvi Live Paywall For Live Streams
 

Segmentation - The Shadowy Side of Persona Development

  • 1.   Segmenta(on:     The  Shadowy  Side  of  Persona  Development   UPA  2012     David  A.  Siegel  Ph.D.   Dray  &  Associates,  Inc.   Minneapolis,  MN    USA               david.siegel@dray.com       www.dray.com   +1  612  377  1980       u  Copyright,  Dray  &  Associates,  Inc.,  2012   Copyright 2012
  • 2.   Segmentation         Market  Segmentation                  User  Classification     2 Copyright 2012
  • 3. Interlocking  Challenges:   Who?   What?   3 Copyright 2012
  • 4. Shadowy?   4 Copyright 2012
  • 5. •  Colors = dimensions •  Can you align them all? •  The most successful are those willing to break a partial alignment and start from scratch Goal:       Ø  Make  explicit  choices  and  tradeoffs,  whether  working  with  and  exis<ng   segmenta<on,  or  proposing  a  classifica<on  scheme  of  your  own     Themes:   Ø  Segmenta<on  as  a  subtype  of  classifica<on   Ø  Classifica<on  =  selec<ng,  defining,  priori<zing,  and  combining  dimensions  to   usefully  divide  up  a  mul<-­‐dimensional  space   Ø  Influenced  by  subjec<ve  choices  and  prone  to  distor<ons,  whether  done   casually  or  through  the  most  high-­‐powered  sta<s<cal  analysis   6 Copyright 2012
  • 6. Ø  What  makes  a  useful  classifica<on?     Ø  Tensions  between  marke<ng  and  UX  segments   Ø  The  paradox  of  “precision”   Ø  Pros  and  cons  of     •  Demographics   •  Occupa<onal  Roles     •  Psychographics     •  Behavior     Ø  Tensions  between  marke<ng  and  UX  segments   7 Copyright 2012
  • 7. Unusually  Clean  Clusters   8 Copyright 2012
  • 8. Coherence  Within  Clusters   9 Copyright 2012
  • 9. Differen<a<on  Among  Clusters   9 Copyright 2012
  • 11. Dimensions  of  Difference  Are  Not  Givens   -­‐-­‐Even  when  they  describe  seemingly  obvious  differences   What  is  this?   12 Copyright 2012
  • 12. Now  what  is  it?   13 Copyright 2012
  • 13. Now?   14 Copyright 2012
  • 14. Now?   15 Copyright 2012
  • 15. The dimensions we perceive and identify depend on Ø  Context of comparison Ø  What have we sampled Ø  What distinctions we perceive or assume to be relevant E.g., if our purpose was to evaluate agricultural products in terms of potential for industrialized production, we might have classified differently 16 Copyright 2012
  • 16. Informa<on  is  a  difference  that   makes  a  difference.                    -­‐-­‐Gregory  Bateson   Source: http://www.nndb.com/people/169/000100866 / Segmenta<on  needs  to  point  to  different  ac<ons  that  are  available  to  us,  on   the  basis  of  predicted  differences  in  response  from  different  audiences  or   users.     17 Copyright 2012
  • 17. Different  differences  make  a  difference,  depending  on  what   different  ac<ons  we  are  focusing  on.     18 Copyright 2012
  • 18. Classifica<on  Variable  1     Descriptor   Descriptor     Dimensions   Dimensions         Classifica<on   Ac<on   Ac<on   Variable  2   Implica<ons   Implica<ons   Descriptor   Descriptor     Dimensions   Dimensions       Ac<on   Ac<on   Implica<ons   Implica<ons   Ø  Not  necessarily  2  x  2,  or  even  factorial   Ø  Choice  of  classifica<on  variables  usually  based  on  what  we  think  makes  cleanest   split,  is  easiest  to  detect,  or  summarizes  the  profile  of  descriptors   Ø  But  descriptors  could  be  turned  into  classifiers,  depending  on  what  maers   19 Copyright 2012
  • 19. Paradox  of  Precision:  The  “Zoom  In”  Problem   Ø  Zoom in = more detailed, granular description •  More dimensions •  More distinctions •  More subgroups Ø  Perceived  as  more  precise,  more   convincing   Ø  But  (all  things  else  being  equal)  finer   grained  dis<nc<ons  become  more   fuzzy,  boundaries  blur   Ø  A  law  of  nature!   20 Copyright 2012
  • 20. Case  in  point:  Let’s  zoom  in  here   Non   Customers   customers         Opportunity   At  Risk   Aachment   21 Copyright 2012
  • 21. Aachment   Anything  we  do  to  improve  the  ra<o  of  people  in  our  sample   that  we  are  interested  in  will  exclude  some  of  them,  and  reduce   our  ability  to  know  how  they  relate  to  the  popula<on  as  a  whole   22 Copyright 2012
  • 22. With  drill-­‐down,  subgroups  can  cut  across  segments   Seg.  A   Seg.  B   Seg.  C   Seg.  D   23 Copyright 2012
  • 23. Case  Example:  Segments  based  on  abtudes  did  differ  in  composi<on.  But….   Seg.  A   Seg.  B   Seg.  C   Seg.  D   …the  groupings   across  segments   were  more   coherent  and   dis<nct  re:  usage   paern   24 Copyright 2012
  • 24. w  rapidly   ns  mul<ply,   n  simple   rip<ons   ‹#›
  • 25. Overall  SESS   w  rapidly   ns  mul<ply,   n  simple   rip<ons   ‹#›
  • 26. Overall  SESS   w  rapidly   Age   ns  mul<ply,   n  simple   rip<ons   ‹#›
  • 27. Overall  SESS   w  rapidly   Age   ns  mul<ply,   n  simple   rip<ons   Ethnicity   ‹#›
  • 28. Overall  SESS   w  rapidly   Age   ns  mul<ply,   n  simple   rip<ons   Ethnicity   ‹#›
  • 29. Overall  SESS   Orienta<on  to  self-­‐service   w  rapidly   Age   ns  mul<ply,   Net  Worth   n  simple   Disposable  Income   rip<ons   Ethnicity   Source  of  influence   Importance  of  Iden<ty   ‹#›
  • 30. Dimensions  apply  to  all,  but  are  called  out  only  where  most   dis<nc<ve,  heightening  percep<on  of  difference   Affluent   Others     • Highly  Influenced  by     family?   • Highly  Influenced   Hispanics   • High  SES?   by  family   • Manage  own     finances  on  line?       • Manage  own   ?   Others   finances  on  line   31 Copyright 2012
  • 31. Segments  Summarizing  Overall  Difference  in  Profile  on  Mul<ple   Dimensions   Ø  Some  dimensions  differen<ate  more  strongly  than  others.       Ø  Smaller  differences  should  be  weighted  less   Ø  But  ofen  all  the  differences  become  equal  parts  of  the  descrip<on         32 Copyright 212 Copyright 2012
  • 32. Some<mes,  it  may  look  like  we  can  make  precise  dis<nc<ons  based  on   small  differences,  only  because  large  samples  make  them  sta<s<cally   significant.    But  do  those  differences  maer?   Math  scores:  Yes,  the   distribu<ons  are  different   (assuming  large  N).    But  if  you   made  dichotomous  decisions   based  on  gender,  (e.g.,  pubng   girls  in  low  math  group  and   boys  in  high  math  group)  you   could  be  wrong  large  %  of   cases.     Source: http://www.wanoah.co.uk/?p=37  The  larger  the  sample  it  takes  to  find  a  sta<s<cally  significant  difference,  the  less   likely  it  is  to  have  a  prac<cal  significance! 33 Copyright 2012
  • 33. Striving  for  Precision  on  Mul<ple  Dimensions   Two  views,     same  points  in  space  à         34 Copyright 2012
  • 34. Back  to  our  unnaturally  clean  clusters:   35 Copyright 2012
  • 35. Imagine  they  are  really  in  3  dimensions,  but  we  have  only  viewed   them  from  one  angle  (i.e.,  only  focusing  on  2  dimensions)   36 Copyright 2012
  • 36. Now  we  rotate.  Same  points  different  views—clusters  smear  out.       37 Copyright 2012
  • 37. In  these  dimensions,    clusters  are  broader,  have  different  members,   personas  not  as  “well  placed”  to  represent  them.   Op<mizing  groupings  on  some  dimensions,  tends  to  “smear”  them  on   38 Copyright 2012
  • 38. Now  imagine  if  we  had  started  off  with  a  more  realis<c  set  of  clusters —just  slight  varia<ons  in  density  –  because  on  many  of  the   dimensions  we  care  about,  people  don’t  fall  into  such  discreet  groups.   39 Copyright 2012
  • 39. Is  vanilla  ice  cream     more  like  chocolate  milk     or  banana  yogurt?   ? Needed:  a  way  of  combining  differences   on  mul<ple  dimensions  into  a  judgment   of  overall  similarity  and  difference.     40 Copyright 2012
  • 40. Ø  Sta<s<cal  approaches  to  building  clusters  usually  try  to  manage  problem  of   over-­‐op<mizing  on  some  dimensions  and  smearing  on  others   Ø  Use  “distance”  to  represent  “difference”     Heavy  user  of  very   few  features     Cluster  2   Group  A  =  Heavy   Y  =  overall   users  of  many   usage  (me features     Which   group  is   this  one   most  like?   Group  B  =1  ight  to  medium   Cluster    L users  of  few  features     X  =  number  of  features  used   41 Copyright 2012
  • 41. But  how  do  we  measure  distance?  You  get  to  choose,  e.g.:   Euclidean:  Hypotenuse  of  difference  on  x  and  difference  on  y          ΔX  +  ΔY:  Sum  of  the  differences  on  each  separate  dimension                      Both  make  intui<ve  sense,  but  give  different  results!   The  two    methods                               assign  the  point  to                               different  clusters.   Euclidean  distance  ≈  4.6   ΔX  +  ΔY  distance  ≈  6.5                               Also,  the  dimensions                               should  be  weighted   Centroid  of  Group  A   differently  based  on:                               Ø  Are  they  scaled  the   same?                               Ø  Are  they  measured   Euclidean  distance  ≈  5.94                   ΔX  +  Δ  Y  distance  ≈  6             equally  reliably?   Ø  Are  they  equally                               good  predictors  of   Centroid  of  Group  B   something  we  care                               about?                                 42 Copyright 2012
  • 42. Exaggera<ng  Dis<nc<veness   These  look  dis<nct,  but  most  of   Cluster  2embers  have  a  lot  in   their  m   common  on  one  or  both   dimensions     Y   Cluster  1   X 43 Copyright 2012
  • 43. Prac<cal  Criteria  for  Priori<zing,  Weigh<ng  and  Combining   Dimensions   Ø  How  efficiently  they  let  you  divide  the  sample  into  categories     Ø  Whether  there  is  a  clear  breakpoint  or  threshold  effect  on  other  variables   Ø  Ease  of  defini<on,  measurement   Ø  Ease  of  loca<ng  real  representa<ves  when  you  want  to  study  group  in   more  depth   Ø  Amount  of  varia<on  on  the  dimension   Ø  Amount  of  independent  informa<on  added,  how  much  heterogeneity   the  dimension  accounts  for   Ø  Usefulness  as  proxy  for  harder  to  measure  variables   Ø  Availability  of  external  informa<on  sources  for  es<ma<ng  prevalence   Ø  Power  as  a  predictor  of  differen<al  response     Ø  Prac<cal  ability  to  act  differen<ally  depending  on  where  on  the   dimension  people  fall     44 Copyright 2012  
  • 44. Demographic  Segmenta<on   45 Copyright 2012
  • 45. Demographic  Segmenta<on     Ø  Ofen  cri<cized  as  selec<on  criteria  for  usability  studies   Ø  But  demographic  variables  have  some  advantages   •  Rela<vely  easily  defined,  measured,  detected,  sized     •  Easy  to  locate  real  representa<ves  when     •  Amount  of  varia<on  can  be  great   •  Informa<on  added  at  lile  cost,  makes  them  good  proxies     •  Many  products  designed  for  targeted  demographics     •  Many  aspects  of  life  may  correlate  with  demographic  dis<nc<ons,  so   can  have  power  as  a  predictor  of  differen<al  response,  needs     •  Prac<cal  ability  to  act  differen<ally  toward  them  for  messaging,  sales   channels,  etc.   •  First  level  filter  when  you  don’t  yet  know  enough  to  be  more  nuanced       46 Copyright 2012
  • 47.  Occupa<onal  Segmenta<on   Ø  Profession Ø  Abstract, higher order category (e.g., “knowledge worker,” “entrepreneur” Ø  General functional area: operations, customer service, finance, IT Ø  Specific roles Ø  Hierarchy: “Executive,” “Manager,” “Supervisor,” “Front line worker” Ø  Context focused: Industry or industry type, company size, business model, organizational structure 48 Copyright 2012
  • 48. Occupa<onal  Segmenta<on:  Issues   Ø  Varying degrees of standardization in nomenclature, function, and job design Ø  Can your domain knowledge, focus, and sample size compensate for the “zoom in” problem? Ø  Functional labels can be very difficult to define: •  What is a “knowledge worker”? •  What is a “power user”? 49 Copyright 2012
  • 49. u  What  is  a  “knowledge  worker?”                    Controller  (finance)                          Logis<cs  manager   u   •  Focus:  high-­‐level  processes  to     •  Focus:  tac(cal,  opera(onal   manage  financial  risk       •  Priority:  Preven(on  of  low   •  Priority:  Increase  efficiency,   probability  events   ensure  smooth  opera(on       •  Decisions  based  on   •  Needs  quan(ta(ve  data  to   professional  judgment,   manage  processes,  look  for   knowledge  of  best  prac(ce   improvement  opportuni(es   •  Sets  policy  for  long  term   •  Manages  processes  in  real   (me   50 Copyright 2012
  • 50. Psychographic  Segmenta<on   Ø  Abtudes,  preferences,  values   Ø  Intended  to  predict  “resonance”  for  messaging   Ø  Also  ofen  emphasized  in  personas  for  broad,   generalizable  implica<ons   Ø  How  strongly  do  they  relate  to  or  predict  usage   paern  or  other  behaviors?   Ø  Are  they  really  more  “stable”  than  behaviors?     Ø  How  hard  are  they  to  measure  reliably  and  validly   Ø  Self  report  versus  behavioral  self-­‐iden<fica<on   51 Copyright 2012
  • 51. Behavioral  Self-­‐Iden<fica<on       What  can  you  say  about  psychographics  (e.g.,  preferences)  of   people  who  gather  In  these  venues?   52 Copyright 2012
  • 52. People  Who  Choose  These  Periodicals?   53 Copyright 2012
  • 53. Behavioral  Segmenta<on   Ø  Self-­‐reported  versus  observed   Ø  Purchasing  behaviors     Ø  Usage  behaviors:  Amount?  Variety?  Qualita<ve   paern?   Ø  Expressed  behavioral  inten<ons:     •  How  immediate?   •  Evidence  of  preliminary  steps  to  confirm?     Ø  Evaluate  degree  of  demonstrated  associa<on  with   behavior  of  ul<mate  interest   54 Copyright 2012 u     
  • 54.     Marke<ng  Segments  &  UX  Categories:  The  Ideal   Time   Non-­‐ Target     Target     UX  delivers  promised  value   (and  more)  à  sa<sfac<on,   reten<on   Targeted  value  messaging   Purchase  decision   increases  concentra<on  of   process  filters  out  most   poten<al  buyers   of  non-­‐target  popula<on   55 Copyright 2012
  • 55.     Marke<ng  Segments  &  UX  Categories:  The  Ideal   Warning:     Time   Ø  This  is  most  likely  when  Market  Segmenta<on  and  UX   categoriza<on  map  to  each  other     Ø  But  market  segmenta<on  guides  strategies  for  ini<al   filtering,  rather  than  ongoing  experience,  so  relevant  and   available  dis<nc<ons  in  ac<on  may  be  different   Ø  UX  has  to  provide  extended  sa<sfac<on  over  a  range  of   encounters  for  each  user   Ø  UX  has  more  at  stake  in  each  touch  point,  because  goal  is   engagement  for  an  already-­‐filtered  audience   Ø  Therefore,  UX  may  introduce  deeper  and/or  transverse   dis<nc<ons  essaging   filters  out  most  of  non-­‐target   Targeted  value  m increases  concentra<on  of   Purchase  decision  process   UX  delivers  promised  value   (and  more)  à  sa<sfac<on,   poten<al  buyers   popula<on   reten<on   56 Copyright 2012
  • 56. Tips   Ø  Method  triangula<on:     Ø  Start  with  criterion  groups  (differences  you  really  care  about)  and  then  look  for   differen<ators.       Ø  Start  with  possible  differen<ators  and  test  to  see  if  they  do  predict  differences   you  really  care  about.   Ø  Test  dis<nc<ons  among  segments  that  people  already  believe   in  to  validate  that  they  really  do  predict  something  important   and  ac<onable     Ø  Par<al  alignment  on  a  few  variables  of  different  types  may  be   more  robust  and  useful  than  than  op<mizing  for  “clean”   dis<nc<ons     Ø  Priori<ze  dimensions  based  on  both  prac<cal  and  conceptual   tradeoffs   57 Copyright 2012
  • 57. More  Tips   Ø  Test  dis<nc<ons  across  mul<ple  studies,  or  do  cross-­‐ valida<on  within  your  sample  by  splibng  it.   Ø  Consider  impact  of  variables  one  at  a  <me  rather  than  only  in   combina<ons,  to  reduce  risk  of  illusory  precision   Ø  Try  to  work  within  exis<ng  segments,  but  be  prepared  to   show  how  different  contexts  may  make  transverse  segments   more  or  less  relevant     Ø  Studying  pre-­‐defined  segments  one  at  a  may  blind  you  to   subgroups  that  are  similar  across  segments-­‐include   contras<ng  hypothesized  segments  into  samples  within  or   across  studies   Ø  Don’t  expect  the  “average”  differences  of  segments  to  show   up  in  small  samples.   58 Copyright 2012
  • 58. David  A.  Siegel  Ph.D.   Dray  &  Associates,  Inc.   Minneapolis,  MN    USA               david.siegel@dray.com       www.dray.com   +1  612  377  1980