Digital Communications with Peter Shafer – Episode 59

Data Leadership Lessons
Data Leadership Lessons
Digital Communications with Peter Shafer - Episode 59
/

Watch this episode on YouTube: https://youtu.be/OCUQo9nou3E

This week we welcome Peter Shafer. Peter is the Vice President of Sales and Marketing at Everest Communications. Everest is a digital communications firm that provides counsel and program execution support to companies in the areas of analytics, social media strategy, and digital reputation repair. We have an awesome conversation around a wide variety of marketing research topics!

Save 20% on your first order at the DATAVERSITY Training Center with promo code “AlgminDL” – https://training.dataversity.net/?utm_source=algmindl_res

Connect with Anthony J. Algmin on LinkedIn – https://www.linkedin.com/in/anthonyjalgmin

Data Leadership Lessons Home https://DataLeadershipLessons.com

About our Guest:

Peter Shafer is the Vice President of Sales and Marketing at Everest Communications. Everest is a digital communications firm that provides counsel and program execution support to companies in the areas of analytics, social media strategy, and digital reputation repair.

Having worked for prestigious polling organizations such as Gallup and Harris, as well as large global PR firms, Peter is the ideal person to shed light on how using data effectively can vastly improve your digital marketing campaigns.

Your audience is going to love his advice – whether it is made up of high-level executives, marketing experts, researchers, or business owners yearning to effectively communicate their messages and build long-term engagement with their brand.

Peter believes strongly that context, content, and collaboration are the keys to success in today’s competitive digital environment. He wants to share insights with listeners to enable them to start building better digital strategies that will not only work… they will exceed their expectations!

Peter on LinkedIn: https://www.linkedin.com/in/peteshafer/

Everest Communications: https://everestcomms.com/

Everest on Facebook: https://www.facebook.com/EverestCommsPage

Episode Transcript:

1
00:00:05,338 –> 00:00:09,092
anthony_algmin: Welcome to the Data Leadership Lessons Podcast. I’m your host Anthony J.

2
00:00:09,342 –> 00:00:12,779
anthony_algmin: Algmin. Data is everywhere in our businesses and it takes leadership to make

3
00:00:12,862 –> 00:00:16,132
anthony_algmin: the most of it. We bring you the people, stories and lessons to help you

4
00:00:16,299 –> 00:00:17,217
anthony_algmin: become a data leader.

5
00:00:17,217 –> 00:00:21,187
anthony_algmin: Today on Data Leadership Lessons, we welcome, Peter Shafer. Peter is the vice

6
00:00:21,337 –> 00:00:25,024
anthony_algmin: President of sales and marketing at Everest Communications. Everest is a

7
00:00:25,108 –> 00:00:28,628
anthony_algmin: digital communications firm that provides counsel and program execution support

8
00:00:28,762 –> 00:00:28,812
anthony_algmin: support to

9
00:00:29,079 –> 00:00:33,083
anthony_algmin: two companies in the areas of analytics, social media strategy and digital

10
00:00:33,249 –> 00:00:35,719
anthony_algmin: reputation repair. Peter. Welcome to the show.

11
00:00:35,719 –> 00:00:38,088
peter_shafer: Anthony, great to meet you. Thank you very much for having me

12
00:00:38,688 –> 00:00:41,324
anthony_algmin: So like we do with all our first time. Guess, just take a moment and tell

13
00:00:41,491 –> 00:00:44,928
anthony_algmin: the audience a bit more about your career before Ever’s communications, and

14
00:00:45,078 –> 00:00:47,814
anthony_algmin: how that journey led you to doing what you do now.

15
00:00:48,014 –> 00:00:52,719
peter_shafer: sure. I actually was on two parallel paths for most of my career. One is on

16
00:00:52,802 –> 00:00:57,123
peter_shafer: the data side. I’ve worked with Gallup Poll, Harris Poll, several market

17
00:00:57,440 –> 00:01:00,894
peter_shafer: research companies on actually the gathering of data, mostly public

18
00:01:01,211 –> 00:01:04,964
peter_shafer: opinion, and in the second side has been on the communications in public

19
00:01:05,198 –> 00:01:08,651
peter_shafer: relation side. So a lot of the work that we did was actually taking the data

20
00:01:08,802 –> 00:01:12,889
peter_shafer: and then making it ready, explaining what the implications of the data are.

21
00:01:13,206 –> 00:01:17,927
peter_shafer: What the D is saying? what directional things we can draw from that? So I’ve

22
00:01:18,078 –> 00:01:21,848
peter_shafer: worked in a number of uh, P. R firms throughout my career and that’s actually

23
00:01:21,998 –> 00:01:25,852
peter_shafer: how I met the people at Evers Communications. and Uh, that’s how we started to

24
00:01:25,852 –> 00:01:29,923
peter_shafer: work together a couple years ago and then I came on staff full time with them.

25
00:01:30,090 –> 00:01:34,410
peter_shafer: But Um, So that’s that’s my background. Um, you know, D has been a part of

26
00:01:34,727 –> 00:01:39,282
peter_shafer: almost every role that I’ve had since I came out of college and Um, and most

27
00:01:39,365 –> 00:01:43,203
peter_shafer: of it again has been around market research and polling, so it’s uh, been been

28
00:01:43,369 –> 00:01:46,806
peter_shafer: interesting to see the ships in public opinion and what that looks like over

29
00:01:46,973 –> 00:01:49,042
peter_shafer: the last uh, twenty five plus years.

30
00:01:50,760 –> 00:01:55,331
anthony_algmin: Interesting, So I last week, uh, I had a conversation with someone who runs

31
00:01:55,398 –> 00:02:00,120
anthony_algmin: a digital marketing agency in Chicago, and he came out of the financial uh

32
00:02:00,453 –> 00:02:03,723
anthony_algmin: space, and and it done a lot of like hedge fund management, and and things

33
00:02:03,807 –> 00:02:07,010
anthony_algmin: like that and that that financial analysts stuff. And and that the parallels

34
00:02:07,160 –> 00:02:12,448
anthony_algmin: really surprised me, because the kinds of a analysis that you do is not

35
00:02:12,599 –> 00:02:15,802
anthony_algmin: terribly dissimilar like there’s a lot of similar kinds of of number

36
00:02:16,052 –> 00:02:19,005
anthony_algmin: crunching. But it got me thinking and I’m glad that we’re having you on the

37
00:02:19,005 –> 00:02:22,759
anthony_algmin: show today because you’re in a slightly different but related type of area.

38
00:02:23,560 –> 00:02:27,647
anthony_algmin: And it, it got me thinking around like this whole notion of polling and

39
00:02:27,730 –> 00:02:31,885
anthony_algmin: surveys, and and how we are trying to understand this world around us Right

40
00:02:32,202 –> 00:02:37,891
anthony_algmin: and I, I’m curious like there’s so many areas that as a person who likes to

41
00:02:37,974 –> 00:02:41,561
anthony_algmin: work with data like, and that’s probably so you know most of the audience

42
00:02:41,811 –> 00:02:46,049
anthony_algmin: for the show. We all enjoy working and crunching numbers. to some extent or

43
00:02:46,132 –> 00:02:50,520
anthony_algmin: another. We at least appreciate what the Uh data can do for for an

44
00:02:50,603 –> 00:02:55,158
anthony_algmin: organization, But what is really different and interesting amongst each of

45
00:02:55,241 –> 00:02:59,479
anthony_algmin: us, it seems is that we get pulled into a direction. We find this passion

46
00:02:59,646 –> 00:03:03,733
anthony_algmin: for a particular area, particular domain or particular kind of analysis, a

47
00:03:03,800 –> 00:03:08,204
anthony_algmin: particular kind of technology that we just naturally gravitate to. So, can

48
00:03:08,371 –> 00:03:12,609
anthony_algmin: you talk about why? for you, you gravitated to this area

49
00:03:13,726 –> 00:03:17,564
anthony_algmin: versus potentially something else, And and maybe what it is about it that

50
00:03:17,647 –> 00:03:18,765
anthony_algmin: you particularly enjoy?

51
00:03:19,282 –> 00:03:24,087
peter_shafer: Sure, um, I, I actually still ll into it by mistake. I guess Um. I actually

52
00:03:24,404 –> 00:03:28,725
peter_shafer: was a client who was using polling data and marketing data to make decisions

53
00:03:29,042 –> 00:03:34,247
peter_shafer: around Uh, advertising placement, And and this is back pre social media, Um.

54
00:03:34,731 –> 00:03:39,919
peter_shafer: and Uh. One of my clients actually approached me about coming in house for

55
00:03:40,169 –> 00:03:45,441
peter_shafer: Gallop Poll and working with them and the the, The One area that that really I

56
00:03:45,441 –> 00:03:49,846
peter_shafer: was very passionate about and continue to be passionate about is using the

57
00:03:49,929 –> 00:03:53,533
peter_shafer: polling data to really understand what is motivating and driving the public

58
00:03:53,933 –> 00:03:59,839
peter_shafer: towards a particular Uh, opinion of particular set of behaviors, Um. And and

59
00:04:00,173 –> 00:04:03,610
peter_shafer: just to continue to be curious as to why that’s evolving, and why that’s

60
00:04:03,843 –> 00:04:08,481
peter_shafer: changing, Um to during my time at the Gaop Pole and at Harris Pole, I spent a

61
00:04:08,481 –> 00:04:12,168
peter_shafer: lot of time looking at voter behavior, a lot of time looking at consumer

62
00:04:12,318 –> 00:04:16,572
peter_shafer: behavior and in, you know, in and how and why people moved around, And you

63
00:04:16,639 –> 00:04:20,893
peter_shafer: know there was always some nugget that we would pull out, Um. But the second

64
00:04:21,127 –> 00:04:24,964
peter_shafer: part that was, they’re very passionate about is helping other people who are

65
00:04:25,131 –> 00:04:29,052
peter_shafer: not very familiar with data, understand the implications of what the data is

66
00:04:29,118 –> 00:04:34,807
peter_shafer: saying, and and what the underlying reasons are for potential shifts in Um in

67
00:04:34,891 –> 00:04:38,962
peter_shafer: that data. And you know it’s I. I think one of the things that you know I joke

68
00:04:39,128 –> 00:04:43,199
peter_shafer: about is that today everybody’s a consultant in some form of fashion, so they

69
00:04:43,449 –> 00:04:47,520
peter_shafer: all automatically have, like you said, those set of answers that that they

70
00:04:47,687 –> 00:04:52,558
peter_shafer: hold true to, and that they find data to support those answers. Um, you know,

71
00:04:52,892 –> 00:04:56,329
peter_shafer: in in my business, I have to go in with kind of more of a clean clean slate,

72
00:04:56,479 –> 00:05:00,083
peter_shafer: and and kind of an open mind about what we think the data is going to say,

73
00:05:00,400 –> 00:05:04,971
peter_shafer: versus trying to get data to prove what we already think is our our world

74
00:05:05,288 –> 00:05:10,093
peter_shafer: view. Um, So it’s it is? it is pitrient. In fact, I’ve just um finished a blog

75
00:05:10,410 –> 00:05:15,048
peter_shafer: post Uh yesterday about some of the polling data from the elections in

76
00:05:15,131 –> 00:05:21,120
peter_shafer: Virginia and New Jersey, And you know and how that is is actually creating

77
00:05:21,604 –> 00:05:24,974
peter_shafer: decisions by these candidates in real time. Um, where they place their

78
00:05:25,124 –> 00:05:29,529
peter_shafer: advertising, what messaging they use, what issues they, they jump on, and Um,

79
00:05:29,846 –> 00:05:33,766
peter_shafer: or they using their campaign, Uh, advertising. So it you, there is a real real

80
00:05:33,850 –> 00:05:38,171
peter_shafer: world application that I like you get to actually see your work in In in, Um,

81
00:05:38,404 –> 00:05:40,973
peter_shafer: you know, play out in real time or real life. So

82
00:05:42,125 –> 00:05:45,812
anthony_algmin: Yeah, well it. it’s kind of one of the basic tenants that I’ve I’ve long

83
00:05:45,878 –> 00:05:49,732
anthony_algmin: advocated for in in data analytics. and as we’ building I build platform to

84
00:05:49,732 –> 00:05:52,852
anthony_algmin: build systems for organizations. And one of the things that I was coach on

85
00:05:52,919 –> 00:05:57,490
anthony_algmin: is you know you need to have an ability to

86
00:05:58,925 –> 00:06:03,813
anthony_algmin: analyze and act in a relatively similar amount of time,

87
00:06:04,363 –> 00:06:08,918
anthony_algmin: And that is if you are making decisions that manifest over months. you don’t

88
00:06:09,168 –> 00:06:14,774
anthony_algmin: need real time analytics at all. If you are literally making a decision this

89
00:06:14,924 –> 00:06:19,011
anthony_algmin: morning for something that happened yesterday. You’re going to want some

90
00:06:19,162 –> 00:06:23,733
anthony_algmin: things that are pretty quick. If you are a traditor. You need to have micros

91
00:06:24,450 –> 00:06:29,572
anthony_algmin: type of reaction times because that’s how quickly you are interpreting and

92
00:06:29,639 –> 00:06:31,174
anthony_algmin: adjusting to that data. So the

93
00:06:31,174 –> 00:06:34,927
anthony_algmin: pulling date is really interesting because you are probably in a similar

94
00:06:35,328 –> 00:06:40,450
anthony_algmin: type of pattern of like You want things as close to now as possible, Because

95
00:06:41,401 –> 00:06:45,638
anthony_algmin: if it starts to get a time delay, you are going to be working with

96
00:06:45,888 –> 00:06:50,443
anthony_algmin: inherently imperfect information That could hurt more than even having a sub

97
00:06:50,693 –> 00:06:52,211
anthony_algmin: optimal analytics approach.

98
00:06:52,211 –> 00:06:56,082
peter_shafer: e. exactly. and you know you, you bring up two very important points about the

99
00:06:56,082 –> 00:07:00,653
peter_shafer: immacy of data Now versus you know, back in the in. The uh, you know, I

100
00:07:00,653 –> 00:07:04,407
peter_shafer: shouldn’t say dark ages of data collection or polling, but Um, you know,

101
00:07:05,291 –> 00:07:09,762
peter_shafer: having access online to so many millions of people and being able to get those

102
00:07:10,012 –> 00:07:14,967
peter_shafer: immediate reads are certainly possible today and and are being used. Um. One

103
00:07:15,051 –> 00:07:19,288
peter_shafer: of the issues that I think has said that the research industry struggled with

104
00:07:19,689 –> 00:07:23,926
peter_shafer: is that they have not updated the types of questions that they are asking, so

105
00:07:24,010 –> 00:07:27,930
peter_shafer: they’re taking questions that you know have been proven mathematically to to

106
00:07:28,247 –> 00:07:32,652
peter_shafer: really be uh, very good, and and get and elicit the reactions that you want.

107
00:07:32,902 –> 00:07:33,903
peter_shafer: Um. But

108
00:07:34,804 –> 00:07:38,958
peter_shafer: just transferring them over to new technology doesn’t necessarily give you a a

109
00:07:39,041 –> 00:07:42,812
peter_shafer: better view over the world, So you, Yeah, you get the data immediately, but

110
00:07:43,045 –> 00:07:46,566
peter_shafer: you’re asking a a tired and maybe older question. Um, and

111
00:07:46,649 –> 00:07:50,253
peter_shafer: I think you, you know you probably have picked up on that, is that If I ask,

112
00:07:50,403 –> 00:07:53,923
peter_shafer: you know whether the country’s on the right track are wrong track. Um, you

113
00:07:54,006 –> 00:07:59,212
peter_shafer: know at which is a a very typical question in polling Um, You know that may

114
00:07:59,362 –> 00:08:03,199
peter_shafer: resonate differently today than it would even ten years ago. So updating that

115
00:08:03,282 –> 00:08:07,286
peter_shafer: type of question is important. but the the, the mechanisms exactly are right,

116
00:08:07,687 –> 00:08:12,725
peter_shafer: You, we can get. Now you know, Uh, this is a a ballpark, but you can get as

117
00:08:12,808 –> 00:08:16,245
peter_shafer: many as three hundred responses in less than an hour for a particular

118
00:08:16,812 –> 00:08:19,999
peter_shafer: question. Whether it’s on a pricing issue, whether it’s on a reputational

119
00:08:20,166 –> 00:08:23,686
peter_shafer: issue, Whether it’s on an event that happened in your reaction to that event

120
00:08:23,920 –> 00:08:28,174
peter_shafer: And that does help marketers. It helps Uh. people in in in know my area of

121
00:08:28,241 –> 00:08:32,879
peter_shafer: crisis communication are in. in, you know reputational management, Um, to get

122
00:08:33,129 –> 00:08:37,683
peter_shafer: Uh to get an immediate read and to use that as Um, a baseline for additional

123
00:08:37,683 –> 00:08:38,684
peter_shafer: action.

124
00:08:39,168 –> 00:08:42,288
anthony_algmin: Yeah, it. it brings up because you just mentioned something that I want

125
00:08:42,522 –> 00:08:45,241
anthony_algmin: touch on before I get to the biggest question that’s been in my mind since

126
00:08:45,725 –> 00:08:50,680
anthony_algmin: since we started talking. But the first thing is is when we think about you

127
00:08:50,763 –> 00:08:54,684
anthony_algmin: know something that has such a rapid response. So I’ll use an example. I

128
00:08:54,700 –> 00:08:55,701
anthony_algmin: just bought a new car

129
00:08:56,118 –> 00:08:59,405
anthony_algmin: and like a lot of people are buying a new car. I was on the message board

130
00:08:59,488 –> 00:09:03,726
anthony_algmin: for that car model and I’ reading all about it or whatever, and I knew I had

131
00:09:03,893 –> 00:09:07,480
anthony_algmin: a skewed perspective to some extent because the people that are more

132
00:09:07,647 –> 00:09:12,518
anthony_algmin: inclined to post on that board either have an exceptionally good experience

133
00:09:12,852 –> 00:09:15,087
anthony_algmin: or more likely an exceptionally bad experience,

134
00:09:15,721 –> 00:09:19,959
anthony_algmin: which will sway my perception of overall quality and likelihood of issues

135
00:09:20,126 –> 00:09:25,248
anthony_algmin: and things like that. And so when you have those kinds of of opt in types of

136
00:09:25,398 –> 00:09:28,367
anthony_algmin: responsees like you said Hey, we might get three other responses very

137
00:09:28,534 –> 00:09:34,290
anthony_algmin: quickly To me. I wonder how do you manage or control for that self selection

138
00:09:35,174 –> 00:09:40,046
anthony_algmin: issue when certain groups are going to be more inclined to be more vocal

139
00:09:40,529 –> 00:09:42,365
anthony_algmin: than than others. how do you even control for that

140
00:09:42,365 –> 00:09:44,800
peter_shafer: you know that’s That’s a great question. and I’ll I’ll talk about it in two

141
00:09:44,967 –> 00:09:49,438
peter_shafer: different ways. One thing that that one explosion in data has been on on what

142
00:09:49,522 –> 00:09:52,808
peter_shafer: we call the qualitative side, which is the message boards and things like

143
00:09:52,892 –> 00:09:56,812
peter_shafer: that. You know where people are actually sharing their Um. experiences their

144
00:09:56,963 –> 00:10:02,401
peter_shafer: opinions in in in a more Um and an unstructured way. So it’s not scalable

145
00:10:02,652 –> 00:10:05,838
peter_shafer: Things like that. The the one technique that that a lot of companies are now

146
00:10:06,172 –> 00:10:11,043
peter_shafer: starting to use is taking those message boards and taking the commentary, put

147
00:10:11,210 –> 00:10:16,565
peter_shafer: them into um. artificial intelligence tools, and start sorting out patterns of

148
00:10:16,799 –> 00:10:20,086
peter_shafer: of different words, and you know, try to match up sentiment. Certainly, it’s

149
00:10:20,169 –> 00:10:24,724
peter_shafer: not a perfect science, but it is a better way to to go about. you know. back

150
00:10:24,890 –> 00:10:29,045
peter_shafer: in the old day you would have a coder take a file of the verbatums from a

151
00:10:29,211 –> 00:10:33,049
peter_shafer: survey, you know, open ended question, and, and do exactly the same thing. And

152
00:10:33,132 –> 00:10:37,770
peter_shafer: now obviously the technology is, Is there? Um? the the balance to that, Though

153
00:10:37,853 –> 00:10:41,691
peter_shafer: Anthony is exactly what you said is that there is polling data almost on every

154
00:10:41,924 –> 00:10:46,879
peter_shafer: topic and every information. and in there those are structured and and kind of

155
00:10:47,129 –> 00:10:52,001
peter_shafer: um. Uh, what’s the word? I’m looking for more mathematically uh, acceptable

156
00:10:52,318 –> 00:10:56,889
peter_shafer: methodologies to actually get to what you’re talking about. So, if if on your

157
00:10:56,972 –> 00:11:01,844
peter_shafer: new car that you know, let’s say it’s Um, you know, brand ▁x, has a you know

158
00:11:01,927 –> 00:11:05,281
peter_shafer: seventy two percent customer satisfaction rate, and then you go to the message

159
00:11:05,448 –> 00:11:09,602
peter_shafer: board and you see you know that that you know, Maybe it’s fifty fifty in terms

160
00:11:09,685 –> 00:11:12,805
peter_shafer: of whether people like it or not, you at least have now two data points to

161
00:11:12,805 –> 00:11:16,559
peter_shafer: kind of to say one is quantitative, the other’s qualitative, And then you can

162
00:11:16,809 –> 00:11:20,746
peter_shafer: make a decision from there from my standpoint from the analysis, And and this

163
00:11:20,746 –> 00:11:21,747
peter_shafer: is what we help

164
00:11:21,764 –> 00:11:25,918
peter_shafer: executives. With is that they often get both of those sets of data and can’t

165
00:11:26,085 –> 00:11:29,688
peter_shafer: make sense of. Well, what about this? What about this? Um. I will tell you

166
00:11:29,772 –> 00:11:34,810
peter_shafer: that in, in some regards that the default now is more towards the qualitative

167
00:11:35,211 –> 00:11:39,048
peter_shafer: information just because there’s so much of it versus the quantity of

168
00:11:39,198 –> 00:11:42,968
peter_shafer: information which there is a finite set of Um. And that’s one of the big

169
00:11:43,119 –> 00:11:44,253
peter_shafer: differences now. So

170
00:11:45,571 –> 00:11:50,126
anthony_algmin: And it makes some sense that you know this is a problem that can be solved

171
00:11:50,292 –> 00:11:54,130
anthony_algmin: to an extent at least at the pattern level at the aggregate level. Once you

172
00:11:54,280 –> 00:11:58,134
anthony_algmin: learn some of the best practices around sifting through some of that

173
00:11:58,684 –> 00:12:04,123
anthony_algmin: inconsistency, then that applies in many situations, and so as a as an

174
00:12:04,290 –> 00:12:08,127
anthony_algmin: expert pollar, as a person who does this professionally, you’re going to be

175
00:12:08,210 –> 00:12:11,647
anthony_algmin: able to navigate that with tools and techniques that are probably not as

176
00:12:12,281 –> 00:12:17,486
anthony_algmin: accessible if not available to general business people like myself. for

177
00:12:17,486 –> 00:12:21,290
peter_shafer: uh, no, that’s that’s true. And and you know the the, The science is evolving

178
00:12:21,440 –> 00:12:24,560
peter_shafer: all the time, Because obviously there’s just so much data that you can start

179
00:12:24,810 –> 00:12:27,930
peter_shafer: working with. You had mentioned something earlier that I. I. I, you know, kind

180
00:12:28,013 –> 00:12:30,733
peter_shafer: of joke with with you know, math is math, and

181
00:12:30,966 –> 00:12:35,287
peter_shafer: statistics are statistics. And so there’s a lot you can do within that. but

182
00:12:35,521 –> 00:12:40,242
peter_shafer: ▁ultimately, what we’re in the business of is math. And so when we come up

183
00:12:40,326 –> 00:12:45,047
peter_shafer: with these designs and and’ working with these large data sets, Um, you know,

184
00:12:45,131 –> 00:12:48,567
peter_shafer: it comes down to statistical modeling and math. And and these are all tools

185
00:12:48,734 –> 00:12:53,122
peter_shafer: that have been around for for a long time, and a fine tuned. so we feel like.

186
00:12:53,606 –> 00:12:59,528
peter_shafer: In in many regards, Um, we may be able to give both directional information

187
00:13:00,012 –> 00:13:05,367
peter_shafer: but also data to support that in a way that is certainly not perfect. but it

188
00:13:05,518 –> 00:13:12,158
peter_shafer: is going to be more um real world than just making a gut decision on it. Um,

189
00:13:12,408 –> 00:13:17,763
peter_shafer: you know, and and and we again, like as you know, in my work, we work on

190
00:13:18,564 –> 00:13:22,968
peter_shafer: individual brands, we work on mega brands. We were on global brands, and in

191
00:13:23,052 –> 00:13:27,373
peter_shafer: that case, then you’re piecing together a much bigger puzzle, and what we’re

192
00:13:27,523 –> 00:13:32,411
peter_shafer: really looking for at that point, Are you know trends that we see within the

193
00:13:32,411 –> 00:13:36,732
peter_shafer: data that help at least make the decision or make the action point, as you

194
00:13:36,799 –> 00:13:40,402
peter_shafer: mention clearer to the decision makers, So

195
00:13:41,904 –> 00:13:42,905
anthony_algmin: Yeah,

196
00:13:44,039 –> 00:13:46,759
anthony_algmin: and and and yeah, that’s that’s interest. I didn’ even have a questionested,

197
00:13:46,842 –> 00:13:51,013
anthony_algmin: read because I was, so. I was like, Um, but so let’s let’s ship gear because

198
00:13:51,030 –> 00:13:52,031
anthony_algmin: I want to get to the big

199
00:13:52,031 –> 00:13:55,167
anthony_algmin: question or the the big thing that I wanted to say. As a result of of this

200
00:13:55,251 –> 00:13:58,604
anthony_algmin: conversation. Then we’ll come back to Uh, some of the techniques and and

201
00:13:58,687 –> 00:14:00,773
anthony_algmin: some of the more advanced stuff. Because I think like we’ve just been

202
00:14:00,840 –> 00:14:05,644
anthony_algmin: talking about. This is the. the. You know the current. You’ best to read the

203
00:14:05,644 –> 00:14:09,398
anthony_algmin: most complex stuff that’s out there and it’s fascinating, but it may also be

204
00:14:09,882 –> 00:14:13,002
anthony_algmin: a little bit advanced for folks that haven’t really spent a lot of time

205
00:14:13,085 –> 00:14:15,487
anthony_algmin: talking about this. and I will tell you a little bit about something like.

206
00:14:15,638 –> 00:14:20,125
anthony_algmin: We all have these moments in our careers and in our lives where we have

207
00:14:20,359 –> 00:14:23,812
anthony_algmin: something that we learn or something that we experience, and that we never

208
00:14:24,046 –> 00:14:29,251
anthony_algmin: look at the world quite the same way ever again, and for me in this space I

209
00:14:29,318 –> 00:14:33,005
anthony_algmin: will never forget many years ago, sadly, many years ago. Now, um, I had the

210
00:14:33,088 –> 00:14:36,525
anthony_algmin: opportunity to to get an M B, A. And and like my M. B, A, probably like I

211
00:14:36,525 –> 00:14:40,045
anthony_algmin: could probably just write like a short book of like all the clips and little

212
00:14:40,212 –> 00:14:43,249
anthony_algmin: things, little nuggets that are still in my brain from that entire you know,

213
00:14:43,332 –> 00:14:48,370
anthony_algmin: two and a half years. Um, but one of the moments that forever has terrified

214
00:14:48,854 –> 00:14:54,126
anthony_algmin: me ever since is when I took a marketing research class and I learned like

215
00:14:54,443 –> 00:15:00,449
anthony_algmin: I’ve long known that sometimes surveys, sometimes analysis. You know they.

216
00:15:00,683 –> 00:15:03,969
anthony_algmin: They come with a preconceived notion of the answer in mind,

217
00:15:04,119 –> 00:15:07,806
anthony_algmin: and then everything kind of fits to support that this kind of data. I’ve

218
00:15:07,890 –> 00:15:11,894
anthony_algmin: I’ve I’ve coined. the term data Justified is where we have our idea and then

219
00:15:11,961 –> 00:15:16,448
anthony_algmin: we look for the data to to you know, support that idea. That’s just bad like

220
00:15:16,599 –> 00:15:20,369
anthony_algmin: we get it like that’s just bad analysis. That’s bad scientific method or

221
00:15:20,452 –> 00:15:23,722
anthony_algmin: what have you? But what really terror me in this marketing research class?

222
00:15:24,206 –> 00:15:27,559
anthony_algmin: Not only did I start to learn about all the techniques and stuff that focus

223
00:15:27,893 –> 00:15:29,078
anthony_algmin: groupsy, and,

224
00:15:30,279 –> 00:15:36,368
anthony_algmin: but I learned just how bad most surveys are and how much

225
00:15:37,569 –> 00:15:45,010
anthony_algmin: unintended slant or bias creeps into these things, and that at that point I

226
00:15:45,327 –> 00:15:49,798
anthony_algmin: never could look at a survey the same way again. And it, it terrifies me,

227
00:15:50,366 –> 00:15:54,687
anthony_algmin: people that are intending and thinking that they are doing great market

228
00:15:55,170 –> 00:16:00,042
anthony_algmin: research are completely introducing bias That makes that entire effort

229
00:16:01,410 –> 00:16:06,765
anthony_algmin: unworthy to be done at all. And so from an expert opinion I want to

230
00:16:07,082 –> 00:16:10,686
anthony_algmin: understand from you how do you deal with that and even how do you even just

231
00:16:10,769 –> 00:16:12,771
anthony_algmin: manage like the survey monkeys That I’m sure,

232
00:16:12,921 –> 00:16:18,527
anthony_algmin: like everybody else you get. and it’s like you. You must not enjoy some of

233
00:16:18,527 –> 00:16:21,613
peter_shafer: well, I. I, You know what. You just brought up something and I’m I’m not going

234
00:16:21,680 –> 00:16:29,204
peter_shafer: to uh name names, but Um, the the. what the D i y survey industry has blown

235
00:16:29,371 –> 00:16:35,210
peter_shafer: up, which is great, I guess, but it has created a lot of horrendous research

236
00:16:35,844 –> 00:16:40,165
peter_shafer: and you, and, and some really really poor decisions. Um, and and there’s

237
00:16:40,399 –> 00:16:44,887
peter_shafer: unfortunately not a great way to police that. And and you know, and and I,

238
00:16:45,204 –> 00:16:50,242
peter_shafer: because it’s now so easy and it’s so cheap, even some really sophisticated and

239
00:16:50,242 –> 00:16:54,563
peter_shafer: and what I would you know? term you know. Really? really? Just you know, sharp

240
00:16:54,813 –> 00:17:00,402
peter_shafer: research minds are defaulting to using that for just these casual one off, or

241
00:17:00,652 –> 00:17:05,607
peter_shafer: you know, quick down and dirty type research projects, Um, one of the

242
00:17:05,691 –> 00:17:09,361
peter_shafer: fallacies that people get caught in And and you mentioned it a little bit, is

243
00:17:09,528 –> 00:17:14,166
peter_shafer: that the more responses I have, that means the error rate is less. and that

244
00:17:14,249 –> 00:17:17,286
peter_shafer: means I’m I’m you know, stronger. But if you still have a crappy question,

245
00:17:17,453 –> 00:17:21,206
peter_shafer: you’re going to get a crappy answer. And that’s and regardless what they have

246
00:17:21,206 –> 00:17:24,727
peter_shafer: to, two thousand or five thousand. Um, And that’s the part that I’m seeing is

247
00:17:24,810 –> 00:17:30,165
peter_shafer: that you’re getting really really biased questions being asked that don’t

248
00:17:30,649 –> 00:17:34,169
peter_shafer: number one. Don’t give the respondent a real sense of what it is They’re

249
00:17:34,403 –> 00:17:38,640
peter_shafer: they’re trying to do. The Second is you give them the opportunity to actually

250
00:17:39,608 –> 00:17:44,163
peter_shafer: give a more biased answer. Um, you know, I, I, I don’t. I mean. This is

251
00:17:44,480 –> 00:17:48,801
peter_shafer: probably too far in the weeds, but you know I’ve seen now a number of surveys

252
00:17:48,884 –> 00:17:52,971
peter_shafer: come back with these. what what are called binary responses? It’s yes, no. Um,

253
00:17:53,288 –> 00:17:56,725
peter_shafer: and those are the only two options that you have. Um. You know people are more

254
00:17:56,809 –> 00:18:01,363
peter_shafer: sophisticated that th therere used to set uh ranges of of you know degrees of

255
00:18:02,164 –> 00:18:05,684
peter_shafer: the way they feel or whatever. And and yes, no, it’s is really a hard line on

256
00:18:05,768 –> 00:18:08,654
peter_shafer: some of these things, and I’ve seen the re emergence of those questions just

257
00:18:08,720 –> 00:18:13,358
peter_shafer: because they’re easycause’. Quick and Um, and I think for marketers they can

258
00:18:13,442 –> 00:18:16,078
peter_shafer: then say Well, you know, it’s sixty two percent Yes, and that’s why we’re

259
00:18:16,245 –> 00:18:19,998
peter_shafer: going in this direction. Um. one of the things that that we used to, and this

260
00:18:20,082 –> 00:18:23,285
peter_shafer: is back. You know. there used to be a lot of research on research that was

261
00:18:23,452 –> 00:18:27,289
peter_shafer: done, Meaning that Um. and there’s a. There’s a technique called clustering

262
00:18:27,773 –> 00:18:32,244
peter_shafer: where you ask the same question using different versions of that question and

263
00:18:32,411 –> 00:18:37,683
peter_shafer: you look at how the uh answers get clustered, and the um.

264
00:18:38,734 –> 00:18:42,638
peter_shafer: The, the, the science behind it was pretty simple. Is that, the more you have

265
00:18:42,805 –> 00:18:47,759
peter_shafer: clustered in the middle, the kind of less strong. That question was. Um,

266
00:18:48,093 –> 00:18:49,127
peter_shafer: because it. you know it.

267
00:18:49,611 –> 00:18:52,648
peter_shafer: The, for you know, the the way we would describe it is we don’t you know you,

268
00:18:52,798 –> 00:18:55,918
peter_shafer: you wantnna, you don’t want to see a lot of threes. That’s kind of a neutral

269
00:18:56,084 –> 00:18:59,371
peter_shafer: position. So if they clustered in the middle, that meant the question really

270
00:18:59,438 –> 00:19:03,292
peter_shafer: didn’t kind of get to the issue. If you saw the clustering at the at the

271
00:19:03,358 –> 00:19:07,446
peter_shafer: extremes or closer to the extremes, then you knew you had a question that was

272
00:19:07,529 –> 00:19:14,002
peter_shafer: going to be a, or allow the respondent to at least give a more Um. in, you

273
00:19:14,086 –> 00:19:19,124
peter_shafer: know, a a more natural or more truthful answer. Um, and that you know in in

274
00:19:19,208 –> 00:19:21,927
peter_shafer: one regard what you were talking about the message boards, kind of a

275
00:19:21,927 –> 00:19:25,764
peter_shafer: reflection of that is that it’s the two extremes and univeersus the people in

276
00:19:25,848 –> 00:19:30,569
peter_shafer: the middle. Um. But one of the things was, and this is used a lot in the

277
00:19:30,652 –> 00:19:34,973
peter_shafer: market research side. Uh, Um, and especially on ad testing and and message

278
00:19:35,123 –> 00:19:41,213
peter_shafer: testing. Um, is that those two extremes allow you to create different messages

279
00:19:41,363 –> 00:19:45,601
peter_shafer: for different target audiences, so that you can you know you can say all right

280
00:19:45,767 –> 00:19:48,654
peter_shafer: Here are the three messages, six messages. We’re going to use and we’re going

281
00:19:48,654 –> 00:19:51,840
peter_shafer: to use it with this audience. This audience. That’s where the magic has

282
00:19:51,924 –> 00:19:55,694
peter_shafer: started to come back for some of the clustering analysis because it now gives

283
00:19:55,844 –> 00:19:58,814
peter_shafer: you okay. This is going to play well with this audience. This is not going to

284
00:19:58,814 –> 00:20:03,118
peter_shafer: play well with this audience. So, but that’s that’s you know again. There’s no

285
00:20:03,285 –> 00:20:09,524
peter_shafer: way to um to really prevent against the bias. The one thing that I w. i. That,

286
00:20:09,691 –> 00:20:14,479
peter_shafer: where the bias I think now has shifted a little bit too, Is that because fewer

287
00:20:14,730 –> 00:20:18,734
peter_shafer: and fewer people are taking surveys or the same people are taking more

288
00:20:18,967 –> 00:20:23,288
peter_shafer: surveys? Is that you know you have this phenomena of a professional survey

289
00:20:23,522 –> 00:20:27,693
peter_shafer: taker and they know you know theoretically how to maneuver. Um. That’s put

290
00:20:27,843 –> 00:20:32,014
peter_shafer: more pressure on the researchers to create waiting schemes on the back end

291
00:20:32,164 –> 00:20:36,802
peter_shafer: when they’re doing their analysis to balance that out. So the balancing part

292
00:20:36,969 –> 00:20:40,639
peter_shafer: of it is now more on the back end than it is on the front end, and that’s you

293
00:20:40,639 –> 00:20:45,043
peter_shafer: know again. A, a. sometimes a factor of math. Um, we were talking about this

294
00:20:45,127 –> 00:20:48,880
peter_shafer: in public polling because you know that that. that’s become a big issue

295
00:20:49,281 –> 00:20:53,201
peter_shafer: especially for telephone surveys. Because you know most people aren’t And it’s

296
00:20:53,452 –> 00:20:57,606
peter_shafer: it’s you know. The waiting on the back end is trying to sort out and, and you

297
00:20:57,606 –> 00:21:01,526
peter_shafer: know, make sure that Um that it is. ▁quote. unquote as representative as it

298
00:21:01,610 –> 00:21:02,728
peter_shafer: can be. so

299
00:21:03,161 –> 00:21:08,050
anthony_algmin: yeah, you just mentioned phone, and that just made me think like there’s got

300
00:21:08,133 –> 00:21:09,318
anthony_algmin: to be a challenge with

301
00:21:10,519 –> 00:21:15,641
anthony_algmin: how the mechanisms people used to communicate evolve. because I won’t answer

302
00:21:15,958 –> 00:21:18,360
anthony_algmin: my phone even if I know the person calling me.

303
00:21:18,443 –> 00:21:22,047
anthony_algmin: Certainly not getting to answer it. If I don’t know the person calling me, I

304
00:21:22,130 –> 00:21:23,966
anthony_algmin: know that that’s kind of a trend that

305
00:21:25,083 –> 00:21:27,653
anthony_algmin: you. you won’t get. You can’t pick up the phone and get a representative

306
00:21:27,803 –> 00:21:30,839
anthony_algmin: sample of anybody at this point. Um you, knowing that,

307
00:21:30,839 –> 00:21:33,358
peter_shafer: yeah, I mean it’s it it it takes. I. I’ll give you this example. It used to be

308
00:21:33,692 –> 00:21:38,013
peter_shafer: that for every one survey response you would get, you’d need to make six phone

309
00:21:38,163 –> 00:21:42,968
peter_shafer: calls Now. it’s one to probably about every seventy, Um. And so it takes now

310
00:21:43,285 –> 00:21:46,571
peter_shafer: instead of you know, twenty four hours to forty eight hours is taking a week

311
00:21:46,805 –> 00:21:53,679
peter_shafer: to get to it. Um. And and then you’re still relying on the Um. the respondent

312
00:21:54,479 –> 00:21:57,683
peter_shafer: telling you that, Yeah, this is that I’m I’m exxer That you know they’re

313
00:21:57,766 –> 00:22:01,203
peter_shafer: They’re basically confirming the information that you have. So it’s still not

314
00:22:01,370 –> 00:22:06,008
peter_shafer: that you. There’s still some gaps in in that. Um. It used to be random digit

315
00:22:06,158 –> 00:22:08,727
peter_shafer: dial Because everybody had a landline or at least ninety eight

316
00:22:08,810 –> 00:22:12,481
peter_shafer: percent of the population. Did you could randomize those telephone numbers and

317
00:22:12,647 –> 00:22:16,401
peter_shafer: and have a good cell phones? It’s a lot different because you have you know

318
00:22:16,802 –> 00:22:19,921
peter_shafer: they. they’ different logistics. Um, and there’ different laws around and

319
00:22:20,088 –> 00:22:23,525
peter_shafer: regulations around. You know who you can call and who you can’t. The one big

320
00:22:23,692 –> 00:22:28,480
peter_shafer: trend though is, I think a lot of market researchers have basically said this

321
00:22:28,647 –> 00:22:32,884
peter_shafer: is you know. Through marketing, this is our target audience, so we’re okay

322
00:22:33,051 –> 00:22:36,571
peter_shafer: with the bias of talking to this target audience because that’s actually who

323
00:22:36,638 –> 00:22:41,293
peter_shafer: our customer is. So the default is we’re just going to embrace the bias and

324
00:22:41,610 –> 00:22:48,083
peter_shafer: use that to f tune or to create new product, Uh, lines or new brands, Because

325
00:22:48,333 –> 00:22:53,205
peter_shafer: we now know who we want to talk to, Um, So th that that? that’s an intentional

326
00:22:53,372 –> 00:22:54,573
peter_shafer: bias that has creeped in

327
00:22:59,094 –> 00:23:00,362
anthony_algmin: So I want to go back.

328
00:23:00,612 –> 00:23:03,965
anthony_algmin: So we were talking about, and you, You may have mentioned this in terms of

329
00:23:04,683 –> 00:23:07,969
anthony_algmin: the the back and analysis cause. That may be part of your answer here. but

330
00:23:07,969 –> 00:23:12,374
anthony_algmin: I’m I’m curious about this. I was thinking about. you, say work for an

331
00:23:12,374 –> 00:23:15,243
anthony_algmin: organization and they do their performancews or whatever at the five point

332
00:23:15,494 –> 00:23:21,883
anthony_algmin: scale And they really really try hard to. Uh, you have be the average, you

333
00:23:21,967 –> 00:23:25,170
anthony_algmin: know, they. They really try to anchor towards that middle. And what have

334
00:23:25,320 –> 00:23:29,241
anthony_algmin: you? But then another five point scale. I use the example of the lifts,

335
00:23:29,324 –> 00:23:34,846
anthony_algmin: Andoubvers and Netflixes out there, where if it’s not a five, it’s horrible

336
00:23:35,080 –> 00:23:39,000
anthony_algmin: like it. It’s basically created this binary scale where especially like the

337
00:23:39,000 –> 00:23:42,053
anthony_algmin: oobs and list know, Netflix has done some things to evolve the To and up

338
00:23:42,120 –> 00:23:48,126
anthony_algmin: down. Um, But the uh, the the the ride shares fascinate me because I feel so

339
00:23:48,443 –> 00:23:53,965
anthony_algmin: compelled to rate somebody a five because I don’t want them to lose their

340
00:23:54,199 –> 00:23:59,254
anthony_algmin: job over the four. I give them right and so like, I literally yesterday had

341
00:23:59,404 –> 00:24:04,442
anthony_algmin: to take a ride and took a ride. And the driver like was a wanna be Nascar

342
00:24:04,693 –> 00:24:08,363
anthony_algmin: person. He cut ten minutes off a thirty minute ride. I’m like, How do you

343
00:24:08,363 –> 00:24:12,367
anthony_algmin: even do that? mathematically? it. It was just amazing and this car felt like

344
00:24:12,451 –> 00:24:15,654
anthony_algmin: it was going to fall apart and I was terrified for my life. It was the first

345
00:24:15,971 –> 00:24:19,724
anthony_algmin: lift ride. I’d take it in a year and I’m like I’m like this is horrible And

346
00:24:19,724 –> 00:24:22,444
anthony_algmin: then it got to the point where it’s time to rate them and I’m like, Oh,

347
00:24:22,611 –> 00:24:25,730
anthony_algmin: guess I’ll give them a five. You got me there and it was like like this is

348
00:24:25,814 –> 00:24:28,450
anthony_algmin: the wrongg thing, but I can’t beat the system here.

349
00:24:29,484 –> 00:24:32,204
anthony_algmin: Is it broken or am I just overthinking it?

350
00:24:32,437 –> 00:24:36,808
peter_shafer: no, no, I, I, I, um, I, you know this. I hadn’t thought about it from a broken

351
00:24:37,042 –> 00:24:40,729
peter_shafer: standpoint. I think it needs some significant upgrades. And and you just

352
00:24:40,812 –> 00:24:45,767
peter_shafer: mentioned something that I. I. I think we overlook Um, and I think Po posters

353
00:24:45,934 –> 00:24:50,489
peter_shafer: even, and market researchers overlook is that there are so many different

354
00:24:51,039 –> 00:24:55,527
peter_shafer: values assigned on a one to five scale, So you get in a lift. And you think

355
00:24:56,011 –> 00:25:00,165
peter_shafer: that that far right star, the five you know like that’s excellent or whatever.

356
00:25:00,732 –> 00:25:03,451
peter_shafer: And and but you don’t know whether it’s excellent or not, you just think five

357
00:25:03,602 –> 00:25:08,173
peter_shafer: stars. so implicitly you’re like five stars means excellent. The scaling that

358
00:25:08,240 –> 00:25:12,644
peter_shafer: we have used traditionally used to be pretty fine night. I mean, it had a very

359
00:25:13,445 –> 00:25:17,599
peter_shafer: um, deliberate and have a very serious implication for how you interpreted the

360
00:25:17,599 –> 00:25:20,802
peter_shafer: numbers. Now the scaling is all over the map. I mean, even if you was one of

361
00:25:20,802 –> 00:25:23,438
peter_shafer: five ski, you can describe, you know, bad,

362
00:25:24,573 –> 00:25:29,678
peter_shafer: bad, bad, good, Fantastic, and that you know, so I think you’re you’re. you?

363
00:25:29,678 –> 00:25:30,729
peter_shafer: You’ve hit on something that

364
00:25:31,846 –> 00:25:36,985
peter_shafer: is, is. Um, Both, for the researcher is a struggle. but for the respondent you

365
00:25:36,985 –> 00:25:37,986
peter_shafer: know

366
00:25:38,803 –> 00:25:42,173
peter_shafer: you’re you know your condition to be like Okay. I’m going to do a five. I’m

367
00:25:42,240 –> 00:25:45,927
peter_shafer: goingnna do a five. I’m goingnna do a five. And and the other part, too, that

368
00:25:46,094 –> 00:25:49,681
peter_shafer: I think is when we talked about this earlier, is that a lot of researchers

369
00:25:49,848 –> 00:25:54,402
peter_shafer: write the question to be rational, when in a lot of ways your response is

370
00:25:54,653 –> 00:25:56,888
peter_shafer: completely irrrational. It has something to do with

371
00:25:56,972 –> 00:26:00,492
peter_shafer: something completely different. So how you felt, Not your opinion about it.

372
00:26:00,542 –> 00:26:01,543
peter_shafer: Um. I,

373
00:26:02,410 –> 00:26:06,398
peter_shafer: One thing about the political polling, which, Um, you know, our firm does or

374
00:26:06,648 –> 00:26:10,085
peter_shafer: works with companies that do that, Um, and

375
00:26:10,669 –> 00:26:11,670
peter_shafer: the

376
00:26:12,404 –> 00:26:18,093
peter_shafer: the numbers aren’t have not picked up the intensity, the frustration, the

377
00:26:18,159 –> 00:26:22,647
peter_shafer: anger that voters are seeing. And and or that that they’re ex, they’re They’re

378
00:26:22,814 –> 00:26:26,167
peter_shafer: demonstrating in their behavior, but also they’re saying in you know, on

379
00:26:26,167 –> 00:26:30,972
peter_shafer: message boards and things like that. And because, let’s say it’s a. you know,

380
00:26:31,039 –> 00:26:35,210
peter_shafer: the scores, the three point five or whatever is that we will water down those

381
00:26:35,443 –> 00:26:39,364
peter_shafer: results Because it’s but it’s not measuring the intensity And I think that’s

382
00:26:39,531 –> 00:26:43,685
peter_shafer: where there. there could be a big room for improvement. Is designing questions

383
00:26:43,768 –> 00:26:48,239
peter_shafer: to measure intensity, not just measure opinion. Um, and I think that’s and and

384
00:26:48,406 –> 00:26:50,558
peter_shafer: or or you know. And and that’s in on. you know,

385
00:26:50,809 –> 00:26:54,729
peter_shafer: branding research that’s on anything, is that you know. Um, and you, probably,

386
00:26:55,213 –> 00:26:59,050
peter_shafer: since you know you’re you know you travel. You know you’ll get. Did this

387
00:26:59,367 –> 00:27:02,887
peter_shafer: experience change my opinion about the brand? Well, most of the experiences

388
00:27:02,971 –> 00:27:05,924
peter_shafer: you would have are transactional unless you have a really good one or a really

389
00:27:06,007 –> 00:27:09,044
peter_shafer: bad one. It’s kind of. uh. You know it’s a neutral question and the

390
00:27:09,127 –> 00:27:12,013
peter_shafer: researchers are hoping that it’s neutral because that gives them the

391
00:27:12,080 –> 00:27:14,966
peter_shafer: justification. Hey, we’re doing our job or’re doing this. You know, if you

392
00:27:15,050 –> 00:27:18,320
peter_shafer: call a call center or things like that and rate the person after the call,

393
00:27:18,720 –> 00:27:21,690
peter_shafer: that’s you know. Like you said, from a performance standpoint, they love

394
00:27:21,923 –> 00:27:26,411
peter_shafer: threes because that you know, that’s kind of the great equalizer, Um. but I do

395
00:27:26,411 –> 00:27:27,379
peter_shafer: think that that’s

396
00:27:27,779 –> 00:27:28,780
peter_shafer: the

397
00:27:29,614 –> 00:27:34,252
peter_shafer: how people are using the data on the back end, Um, both in terms of the

398
00:27:34,319 –> 00:27:38,673
peter_shafer: calculations, but also in terms of the interpretation that has shifted. I

399
00:27:38,673 –> 00:27:39,657
peter_shafer: think really

400
00:27:39,657 –> 00:27:42,560
peter_shafer: greatly, and I, as you know, I mentioned before, I think we all think our of

401
00:27:42,644 –> 00:27:46,247
peter_shafer: ourselves as you know, data consultants in one regard, because we’ you know we

402
00:27:46,331 –> 00:27:51,686
peter_shafer: see so much of it. Um. but um, you know the how they’re using it in the

403
00:27:51,686 –> 00:27:56,408
peter_shafer: intention that they’re in uh, ascribing to the different statistics. Um, you

404
00:27:56,491 –> 00:28:00,245
peter_shafer: know that that’s very different and that I think that inconsistency I think

405
00:28:00,328 –> 00:28:04,332
peter_shafer: confuses you know you and I and anybody else who who you know gets a call or

406
00:28:04,482 –> 00:28:08,486
peter_shafer: takes a survey or you know, wants to you know. Provide our opinion.

407
00:28:10,205 –> 00:28:14,209
anthony_algmin: Yeah, I think. I mean, I think it’s fascinating to think about how the back

408
00:28:14,609 –> 00:28:19,247
anthony_algmin: end is trying to error correct for these

409
00:28:20,365 –> 00:28:24,853
anthony_algmin: inconsistencies, or or like, To your point, how strongly people feel about

410
00:28:25,086 –> 00:28:28,206
anthony_algmin: the number that they’re giving Or what have you? I’m sure doing analysis

411
00:28:28,440 –> 00:28:32,293
anthony_algmin: like Hey, Anthony always gives a five on anything. If he gives a one on

412
00:28:32,360 –> 00:28:35,163
anthony_algmin: something. Maybe we take that more seriously than the person who gives

413
00:28:35,330 –> 00:28:39,167
anthony_algmin: eighty percent ones and very rarely gives anything else you know.

414
00:28:39,334 –> 00:28:43,405
anthony_algmin: And so I see how that could be done just purely from a a mathematic basis.

415
00:28:43,571 –> 00:28:46,374
anthony_algmin: And that’s that’s really interesting. and there is one thing in some of the

416
00:28:46,441 –> 00:28:49,160
anthony_algmin: the topic listings that we had that. Im. I’m interested in talking with you

417
00:28:49,244 –> 00:28:52,530
anthony_algmin: because we talked a lot about like message boards, and and some of that

418
00:28:52,764 –> 00:28:56,684
anthony_algmin: qualitative side of things, Um. and and one of the things, and I have some

419
00:28:56,768 –> 00:29:00,205
anthony_algmin: exposure to this in different parts, but I’m certainly no expert on it. Um.

420
00:29:00,772 –> 00:29:03,324
anthony_algmin: talking about how structuring unstructured data.

421
00:29:03,641 –> 00:29:09,481
anthony_algmin: So how do we or how do you in your Um, in your work take and quantify,

422
00:29:09,647 –> 00:29:13,318
anthony_algmin: Assume there’s a quantification component to this. How do you structure that

423
00:29:13,485 –> 00:29:18,373
anthony_algmin: unstructured data so that you can bring that and understand some of that

424
00:29:18,523 –> 00:29:22,444
anthony_algmin: nuance to help in the effort of correcting some of these biasy is

425
00:29:22,444 –> 00:29:24,529
anthony_algmin: intentional or not that we ▁ultimately will have

426
00:29:24,529 –> 00:29:27,599
peter_shafer: yeah, th. it’s a. It’s a great question. Because it, and and it’s a struggle.

427
00:29:27,849 –> 00:29:31,920
peter_shafer: Can you know it still can? Um, it still can be fraught with errors and things

428
00:29:32,086 –> 00:29:36,891
peter_shafer: like that. I mean the, the basic way to do it is to look at all of the

429
00:29:37,208 –> 00:29:40,879
peter_shafer: information that you have that surrounds that unstructured data. So for

430
00:29:40,962 –> 00:29:45,283
peter_shafer: example time stamps are pretty popular. Are there any Um? Are there any

431
00:29:45,450 –> 00:29:50,238
peter_shafer: attribute data available? Um. that would help contact? So for example, on the

432
00:29:50,238 –> 00:29:55,043
peter_shafer: message board, you know we have. let’s say a thousand comments. We know that

433
00:29:55,360 –> 00:29:59,547
peter_shafer: fifty of them came from Minnesota. Two hundred came from Illinois. You know we

434
00:29:59,547 –> 00:30:00,548
peter_shafer: can start looking

435
00:30:00,798 –> 00:30:05,203
peter_shafer: at at some of the the ancillary data that’s collected along with this to begin

436
00:30:05,520 –> 00:30:10,091
peter_shafer: to structure it out. and then we will start looking for patterns within the

437
00:30:10,241 –> 00:30:14,879
peter_shafer: unstructured data that would give us clues as to so. for example, if we see

438
00:30:15,129 –> 00:30:19,450
peter_shafer: the word, you know good a lot. that would be a signal to us that there might

439
00:30:19,534 –> 00:30:23,371
peter_shafer: be something. And then we take a look if we look at for patterns within the

440
00:30:24,172 –> 00:30:28,960
peter_shafer: the, The natural language that people are using to describe it. Um. The one of

441
00:30:29,043 –> 00:30:33,364
peter_shafer: the techniques I know that we’ve we’ve used is Um. and I, I’ve done some work

442
00:30:33,598 –> 00:30:37,519
peter_shafer: in Um. The entertainment industry. There is a correlation between. the longer

443
00:30:37,685 –> 00:30:42,807
peter_shafer: your review is, the better the film was, so that you know. So, if you look at

444
00:30:42,891 –> 00:30:46,961
peter_shafer: word count alone, that’s one. but if you look at, you know, Okay, this person

445
00:30:47,445 –> 00:30:51,849
peter_shafer: had a A more. So you know there. There are different techniques to look at the

446
00:30:51,849 –> 00:30:56,170
peter_shafer: patterns of of how that data came. that. Now that uh, artificial intelligence

447
00:30:56,404 –> 00:31:00,892
peter_shafer: is becoming, you know so much more, you can do a lot more. And what’s happened

448
00:31:01,042 –> 00:31:04,562
peter_shafer: in is now. Thank goodness that everybody’ saved all this data that they’ve

449
00:31:04,729 –> 00:31:09,684
peter_shafer: collected. They’ve gone back into those datas sets, Run these analysis and

450
00:31:09,767 –> 00:31:13,371
peter_shafer: then created these models, hoping to say, Here’s the You know, the next

451
00:31:13,521 –> 00:31:17,292
peter_shafer: algorithm that’s going to be able to be used to do this. I’ll I’ll use it. Um.

452
00:31:17,525 –> 00:31:22,330
peter_shafer: There’s a major pharmacy pharmac pharmacy company that used to work with Um,

453
00:31:22,480 –> 00:31:27,368
peter_shafer: and we did a project with them using a a real initial version of a I, where it

454
00:31:27,452 –> 00:31:31,039
peter_shafer: was two hundred and fifty thousand responses from about five million people

455
00:31:31,289 –> 00:31:35,293
peter_shafer: you know. And then they were all typed in. We segmented it A by mobile phone

456
00:31:35,526 –> 00:31:39,764
peter_shafer: by p. C. where they were coming from time stamps. How quickly you know? for

457
00:31:39,931 –> 00:31:43,201
peter_shafer: example, did you respond immediately after the transaction? Did it take you a

458
00:31:43,284 –> 00:31:47,522
peter_shafer: day? Did it take you to? We clustered all of the data within you know these

459
00:31:47,839 –> 00:31:51,759
peter_shafer: categories. And and that’s that was the umbrella structure that we used was

460
00:31:51,926 –> 00:31:55,847
peter_shafer: those. And then we went and dug in and said Okay, Everybody who said that, uh,

461
00:31:55,930 –> 00:32:00,802
peter_shafer: or reacted within fifteen minutes of leaving a particular store, We were able

462
00:32:00,885 –> 00:32:04,572
peter_shafer: to look at that and you know, so you? there. Were you know it? It really just

463
00:32:04,889 –> 00:32:08,810
peter_shafer: sat down and said, This is the data map we want to use or hear the things that

464
00:32:08,893 –> 00:32:13,364
peter_shafer: are important to us. Um, you know length of response, or you know any of this,

465
00:32:13,765 –> 00:32:18,720
peter_shafer: and again the science is evolving. There’s no one correct way to do it. But

466
00:32:18,970 –> 00:32:23,691
peter_shafer: when you start looking at what you have, you can go say. I really think this

467
00:32:23,841 –> 00:32:27,679
peter_shafer: would be a a good way to look at it, And then you can run it. It doesn’t mean

468
00:32:27,762 –> 00:32:30,798
peter_shafer: you have to use it, but there’s more flexibility in being able to do that

469
00:32:30,882 –> 00:32:32,000
peter_shafer: analysis now and before

470
00:32:33,251 –> 00:32:36,921
anthony_algmin: Yeah, I, I like to think of like going back and looking at old data kind of

471
00:32:37,088 –> 00:32:41,409
anthony_algmin: reminds it. Like how we’re able to go and remaster old films Like this is

472
00:32:41,559 –> 00:32:47,649
anthony_algmin: why we say if when when a dou capture data capture insights record better,

473
00:32:47,882 –> 00:32:50,284
anthony_algmin: you can figure out what you’re going to do later. But if you miss the

474
00:32:50,451 –> 00:32:53,888
anthony_algmin: opportunity to get that data and to get that source information

475
00:32:54,339 –> 00:32:55,340
peter_shafer: exactly

476
00:32:55,340 –> 00:32:59,160
anthony_algmin: in that moment it goes away forever. And so that’s really, and I really like

477
00:32:59,410 –> 00:33:05,166
anthony_algmin: too, the the role of meta data and some of that structured contextual data

478
00:33:05,566 –> 00:33:10,204
anthony_algmin: around the unstructured text in that message Cause you right there. We know

479
00:33:10,288 –> 00:33:13,007
anthony_algmin: where that person is from. We know when they posted

480
00:33:13,174 –> 00:33:17,729
anthony_algmin: this. These are these are, are structured data, meta data around the thing

481
00:33:17,879 –> 00:33:20,365
anthony_algmin: that you really care about, which is their opinion or what have you? But

482
00:33:20,365 –> 00:33:21,349
anthony_algmin: even things like the

483
00:33:22,283 –> 00:33:25,486
anthony_algmin: that. That’s something that I never thought of right. Like the length of the

484
00:33:25,486 –> 00:33:30,441
anthony_algmin: the answer gives us a signal that we can then use for interpretation And

485
00:33:30,525 –> 00:33:33,478
anthony_algmin: that’s really. That’s really cool. and I mean like nobody out there, don’t

486
00:33:33,644 –> 00:33:37,398
anthony_algmin: use that in isolation and say, Oh, our. Our Ne promoter scores have gone

487
00:33:37,565 –> 00:33:41,002
anthony_algmin: through the roof because everybody’s posting long messages. No, No, so, but

488
00:33:41,169 –> 00:33:44,205
anthony_algmin: there’s but that’s where I think you know. Just knowing a little bit.

489
00:33:44,372 –> 00:33:47,158
anthony_algmin: Granted, it can be dangerous. but knowing a little bit kind of opens up this

490
00:33:47,241 –> 00:33:50,445
anthony_algmin: world. And this is why we do this show. Is that it, it gives people a taste

491
00:33:50,611 –> 00:33:53,331
anthony_algmin: of something that maybe they hadn’t thought about before. Maybe it is it’s

492
00:33:53,398 –> 00:33:56,050
anthony_algmin: something that they didn’t have much exposure to or didn’t have that

493
00:33:56,134 –> 00:33:58,052
anthony_algmin: opportunity to take a class and learn about these things.

494
00:33:58,119 –> 00:34:01,489
anthony_algmin: So I really appreciate that and the last minute or two that that we have.

495
00:34:02,123 –> 00:34:08,045
anthony_algmin: Um, What advice do you have for those organizations where they have just

496
00:34:08,763 –> 00:34:13,334
anthony_algmin: realized? Oh man, we should be doing more with this and I don’t even know

497
00:34:13,568 –> 00:34:17,004
anthony_algmin: where to go. What what would you say? Good places to start, things to think

498
00:34:17,121 –> 00:34:20,324
peter_shafer: you know? Uh, it’s a great question. In fact, I’m facing it with a couple of

499
00:34:20,408 –> 00:34:25,763
peter_shafer: clients right now. Um, the first thing is to know that you’re not alone and

500
00:34:25,847 –> 00:34:28,566
peter_shafer: that there are other people within your organization that are struggling with

501
00:34:28,566 –> 00:34:33,838
peter_shafer: the same question. Um, what I have told several of my clients is just even to

502
00:34:34,005 –> 00:34:38,893
peter_shafer: form like a three five person task force just to take an inventory and an

503
00:34:38,960 –> 00:34:42,647
peter_shafer: audit of what you actually think you have, Because what you think you have is

504
00:34:42,730 –> 00:34:45,933
peter_shafer: probably different than what you actually have, and getting that outside

505
00:34:46,167 –> 00:34:50,972
peter_shafer: perspective, or at least that secondary internal perspective is really pretty

506
00:34:51,122 –> 00:34:58,012
peter_shafer: good. Um, The the next is what you had said earlier and that is, don’t stop

507
00:34:58,412 –> 00:35:04,719
peter_shafer: data flow. keep collecting it. Keep using it. keep, um. Keep the pipes open

508
00:35:05,052 –> 00:35:08,406
peter_shafer: for allowing that data because you don’t want to get into a situation where

509
00:35:08,639 –> 00:35:13,361
peter_shafer: you do cut off critical pieces of information coming in. Um. And so I, I would

510
00:35:13,444 –> 00:35:17,048
peter_shafer: say you know. I’m I’m working with a company right now. Who wanted to? You

511
00:35:17,048 –> 00:35:21,052
peter_shafer: know? Because of the P. I rules, Um wanted to cut off and I said No, No,

512
00:35:21,119 –> 00:35:22,487
peter_shafer: that’s not pi. I. You can

513
00:35:22,570 –> 00:35:25,690
peter_shafer: continue to do that so you know, please keep collecting it because it’s

514
00:35:25,923 –> 00:35:31,612
peter_shafer: important for what what we need. Um. The third thing is to just allow yourself

515
00:35:32,079 –> 00:35:36,801
peter_shafer: the time to think about what you. what what you want out of the data. Because

516
00:35:37,051 –> 00:35:42,089
peter_shafer: I think in, I was thinking about this before we were talking. so many ecomerce

517
00:35:42,240 –> 00:35:46,244
peter_shafer: sites allow you to go in and you know, basically ▁query, I want to fly from

518
00:35:46,327 –> 00:35:49,530
peter_shafer: Chicago to Baltimore, and you know, type it in and they get a price or

519
00:35:49,614 –> 00:35:53,201
peter_shafer: whatever with the type of data that we work with. It’s not a ▁query system

520
00:35:53,451 –> 00:35:57,288
peter_shafer: You’ve got to really think about. Okay, I want to look at I Wa, to look at

521
00:35:57,438 –> 00:36:02,009
peter_shafer: Netflix subscribers who live in the Midwest, who at least watched two hours of

522
00:36:02,093 –> 00:36:05,613
peter_shafer: Netflix a week, and and have watched The Crown in the last six months. Well,

523
00:36:05,680 –> 00:36:09,283
peter_shafer: there are four attributes right there That. If you don’t you know, that’s the

524
00:36:09,367 –> 00:36:13,054
peter_shafer: kind of thing that I think people really need to start taking a harder look

525
00:36:13,120 –> 00:36:18,092
peter_shafer: at, or or piecing together. Does you shopping or does shopping at Barnes and

526
00:36:18,159 –> 00:36:23,047
peter_shafer: Noble actually translate into anything related to any other, say,

527
00:36:23,281 –> 00:36:27,201
peter_shafer: Entertainment category gaming category, And I think that’s the thing is that

528
00:36:27,218 –> 00:36:28,219
peter_shafer: the

529
00:36:28,886 –> 00:36:32,406
peter_shafer: we don’t spend enough time asking those questions. We think. Oh, because the

530
00:36:32,573 –> 00:36:37,128
peter_shafer: person shops at Kroger This is. this is a. This is a Kroger shopper, and it’s

531
00:36:37,211 –> 00:36:40,882
peter_shafer: really not that monolithic and you know so challenging that challenging some

532
00:36:40,882 –> 00:36:44,802
peter_shafer: of the assumptions around it. Um, the fourth is, don’t skimp on investing in

533
00:36:44,802 –> 00:36:45,786
peter_shafer: it. Um, I think

534
00:36:45,853 –> 00:36:50,241
peter_shafer: a lot because data is so cheap and there’s so much of it is that there’s not

535
00:36:50,408 –> 00:36:53,928
peter_shafer: enough money being spent on mining it, and and and, and like you said,

536
00:36:54,328 –> 00:36:58,733
peter_shafer: spending enough time on structuring it or putting it into a platform that can

537
00:36:58,883 –> 00:37:03,604
peter_shafer: help you manipulate the data and work with it in a in a constructive way. Um,

538
00:37:03,921 –> 00:37:08,893
peter_shafer: you know, I, I. I. I see too many mistakes being made by saying, Oh, we’re

539
00:37:08,960 –> 00:37:12,647
peter_shafer: just going to do. We’re just going to use Excel to figure this out, and and

540
00:37:12,730 –> 00:37:15,850
peter_shafer: ▁xel’s got a lot of power, no doubt about it, but you know some of these, It

541
00:37:15,933 –> 00:37:18,085
peter_shafer: takes more sophisticated tools, so

542
00:37:18,603 –> 00:37:22,206
anthony_algmin: Absolutely well, Peter. we were all at a time. Thank you so much. just ama.

543
00:37:22,924 –> 00:37:23,958
anthony_algmin: The insights of this are are

544
00:37:24,041 –> 00:37:27,078
anthony_algmin: fantastic, so thank you for sharing that with us and and be it on the show

545
00:37:27,078 –> 00:37:29,614
peter_shafer: no, thank you so much, Anthony. Really pleasure to talk to you and have a

546
00:37:29,614 –> 00:37:30,798
peter_shafer: great rest of your day. thanks.

547
00:37:31,399 –> 00:37:34,285
anthony_algmin: Thanks you too, and thank you all for joining us today. You’ll find more

548
00:37:34,452 –> 00:37:37,805
anthony_algmin: information and links in the show notes. Dive deeper with my book at Data

549
00:37:37,972 –> 00:37:41,809
anthony_algmin: DataLeadershipBook.Com and use Promo Code “ALGMINDL” at the DATAVERSITY Online

550
00:37:42,043 –> 00:37:44,845
anthony_algmin: Trainding center for twenty percent off your first purchase. And if you

551
00:37:45,012 –> 00:37:47,798
anthony_algmin: enjoy our show and would love your own, but don’t know where to start, visit

552
00:37:47,965 –> 00:37:51,969
anthony_algmin: Algmin.com to learn how we make having your own video podcast as easy as

553
00:37:52,053 –> 00:37:55,640
anthony_algmin: joining a call and sending an email. Stay safe during these unusual times,

554
00:37:55,723 –> 00:37:56,774
anthony_algmin: and go make an impact

Scroll to top