Dr. Ekman’s Customer Base 2017
Charlotte Lopp
January 26, 2018
Results and Discussion
## Warning: package 'dplyr' was built under R version 3.4.2## Warning: package 'tidyr' was built under R version 3.4.2## Warning: package 'Rcpp' was built under R version 3.4.2
Figure 1: Figure 1A shows the relative mean access count among the different tools; Figure 1B shows the relative purchase count of each tool.
I created a two part figure in order to show amount of engagement relative to the number of tools sold. We can see that despite being the lowest selling tool, FACS test has the highest mean access count. The rest of the tools have mean access counts that are so low that I created tables in order to investigate the numbers more precisely.
## Email Tool Access.Count
## 1 bonettiraphael@gmail.com FACS Test 10063
## 2 cm.box@icloud.com FACS Test 2448
## 3 studiog.messina@tiscali.it FACS Test 1449
## 4 535265294@qq.com FACS Test 936
## 5 osamura.m@gmail.com FACS Test 726Figure 2: This figure shows the top 5 access counts for the FACS test.
## Email Tool Access.Count
## 1 caochoaq@gmail.com METT 87
## 2 johnpdoody2@yahoo.co.uk METT 44
## 3 yagana.yakubu@asburyseminary.edu METT 42
## 4 ediazl@up.edu.mx METT 40
## 5 paul.christi.kim@asburyseminary.edu METT 29Figure 3: This figure shows the top 5 access counts for the METT test.
## Email Tool Access.Count
## 1 arthur.sher@icloud.com SETT 67
## 2 adavis44@utk.edu SETT 53
## 3 rationalshawn@gmail.com SETT 23
## 4 danielfarrer10@gmail.com SETT 22
## 5 mywong@hsmc.edu.hk SETT 22Figure 4: This figure shows the top 5 access counts for the SETT test.
## Email Tool Access.Count
## 1 Khaled.Salih@icloud.com RETT: Family 7
## 2 j-lilja@hotmail.com RETT: Family 6
## 3 teunzijlmans@gmail.com RETT: Family 4
## 4 riccraig@protonmail.com RETT: Family 4
## 5 deivis_fsilva@yahoo.com.br RETT: Family 4Figure 5: This figure shows the top 5 access counts for the RETT: Family test.
## Email Tool Access.Count
## 1 move_it321@yahoo.com RETT: Law Enforcement 10
## 2 clamabu@gmail.com RETT: Law Enforcement 9
## 3 ancient.eb@gmail.com RETT: Law Enforcement 9
## 4 rouradnik@gmail.com RETT: Law Enforcement 7
## 5 epotter@pottercounseling.com RETT: Law Enforcement 6Figure 6: This figure shows the top 5 access counts for the RETT: Law Enforcement test.
## Email Tool Access.Count
## 1 zhiyan.zhang@shms-mail.ch RETT: Workplace 7
## 2 erdobber@hotmail.com RETT: Workplace 6
## 3 bethel@gmail.com RETT: Workplace 5
## 4 elisa.lazier@libero.it RETT: Workplace 5
## 5 therbkelly@hotmail.com RETT: Workplace 4Figure 7: This figure shows the top 5 access counts for the RETT: Workplace test.
From these six tables, we can see that FACS had exponentially higher engagement than the rest of the tools. METT and SETT are at least in double digit access count numbers which is good considering that they are the highest selling tools. The rest of the tools (the RETT sets) are all pretty much in single digit access counts with the exception of one user for RETT: Law Enforcement who accessed the tool 10 times.
## X2017.01.01...2017.12.31 Units Dollars
## 1 Total 2278 114351.80
## 2 summer 213 10562.20
## 3 truth 184 9269.60
## 4 newyear 161 8227.60
## 5 fall4savings 128 6276.20
## 6 holiday 119 5516.20
## 7 savemore 116 5364.80
## 8 spooky 113 5352.80
## 9 universal 102 5154.40
## 10 winter 97 4630.60
## 11 fall 93 4601.00
## 12 expression 84 4091.00
## 13 spring 78 3677.80
## 14 lietome 69 3408.20
## 15 science 59 2965.00
## 16 eyes 56 2886.60
## 17 crush 48 2272.40
## 18 blackfriday 47 2390.60
## 19 learn 46 2106.80
## 20 washington 46 2154.80
## 21 emotional 45 2081.00
## 22 holidays 43 2151.40
## 23 appreciation 39 2138.00
## 24 usa1776 37 1880.00
## 25 fool 36 1968.60
## 26 ty25rtfb 35 1746.50
## 27 light 32 1547.60
## 28 clover 32 1637.60
## 29 mask 32 1405.60
## 30 december 25 1139.00
## 31 cyber 21 1197.80
## 32 relief 9 404.20
## 33 m316jjltd 7 280.60
## 34 upgrade_2017 5 747.00
## 35 welcomeback 4 143.75
## 36 hadnagy20 4 1011.40
## 37 sv&23!9 4 254.25
## 38 asbury20 2 489.80
## 39 anger25 2 99.50
## 40 pk!30 1 19.80
## 41 #facs_10_! 1 700.00
## 42 rain 1 59.80
## 43 bl-hadnagy12 1 100.00
## 44 highbaugh20 1 240.00Figure 8: This figure shows all 44 coupon codes and their relative performance in decreasing order in both units and dollars.
We can see here that almost all of the highest performing coupon codes were holiday related with the exception of code “truth.”
Figure 9: This figure shows the number of sales we had by the billing country.
I think it is very impressive the range of countries that are represented in this figure because in order to even be represented, someone had to purchase it from there. We can see here that our highest engagement by far is the United States. However, there is considerable engagement in other countries but it is hard to discern which countries they are due to the scale of the chart. Therefore, I created a table in order to report explicitly the name of the top 20 countries.
## # A tibble: 20 x 2
## billing_country n
## <chr> <int>
## 1 US 1197
## 2 AU 165
## 3 DE 158
## 4 CA 128
## 5 GB 108
## 6 IT 77
## 7 NL 56
## 8 CH 53
## 9 BR 50
## 10 FR 49
## 11 JP 43
## 12 AT 39
## 13 RU 37
## 14 MX 36
## 15 PL 31
## 16 SE 28
## 17 ES 26
## 18 SG 22
## 19 IL 19
## 20 DK 18Figure 10: This figure shows the top 20 countries in terms of sales.
One thing to note is that this figure is not representative of overall engagement but solely sales. It is logical to assume that they would be relatively correlative and therefore made deductions and extrapolations from this graph but it’s worth keeping this in mind. For billing country identification that might be unfamiliar (http://www.nationsonline.org/oneworld/country_code_list.html): AU = Australia; DE = Germany; CA = Canada; GB = Great Britain; IT = Italy; NL = Netherlands; CH = Switzerland; BR = Brazil; FR = France; JP = Japan; AT = Austria; RU = Russian Federation; MX = Mexico; PL = Poland; SE = Sweden; ES = Spain; SG = Singapore; IL = Israel; DK = Denmark
## # A tibble: 2,626 x 2
## # Groups: billing_company [584]
## ID billing_company
## <int> <chr>
## 1 38688
## 2 46043
## 3 43979
## 4 46408
## 5 45368
## 6 41506
## 7 46498
## 8 47375
## 9 37956
## 10 43401 Ping An Technology
## # ... with 2,616 more rowsFigure 11: This figure shows the billing company for sales.
As you can see here, there are only some companies reported - not every sale has a company reported. Looking at the billing company is only a circuitous way to try and find out what industry the client is in but it is highly inefficient to look up each company reported - particularly on this scale and especially considering the return we get which may or may not be reliable. A much better way to answer the question of how the clients are using our product and which clients are using our product (in terms of industry targeting) is through directly asking the client via a drop down menu or a list of options to select from. This way we would have completley representative data and it would be accurate. A potential way to posit this question would be “What are you using this product for?”
## Warning: Removed 1 rows containing missing values (geom_col).
Figure 12: This figure shows the number of sales by age.
I considered ommitting this graph but decided to keep it just to bring attention to why this figure looks so strange. The data we have in terms of age is completely unreliable. There was a considerable number of reported ages between 100-200 years old which is incredibly dubious. Also, when trying to clean up the data, I went through the Excel spreadsheet filtering with ‘10’ in order to more efficiently find those that are listed as 100 or older but then realized the number of 10-year-olds that apparently bought this tool. I also find this questionable. I don’t know if age is something we want to remotely pursue but the data we have right now for it is unreliable and therefore unusable.