Bellabeat Case Study
Alan Mark A. Bundoc - Google Data Analytics Capstone Project
30th January, 2023
Table of Contents
Scenario
Ask
- Identifying the business task
- Key stakeholders
Prepare
- Dataset quality & ROCCC analysis
- Data verification
- Analysis limitations
Process
- Data cleaning & data selection
- Data cleaning & data selection
Analyze
- Relationship between calories vs. total no. of steps
- Relationship between sedentary minutes vs. minutes slept
- Quantifying sedentary minutes
- Measuring user smart device usage
Act
1. Scenario
Bellabeat is a high-tech company founded in 2013 that manufactures health-focused smart products. The company has grown rapidly and quickly positioned itself as a tech-driven wellness company for women. Currently, they offer 5 products as follows: Bellabeat App, Leaf, Time, Spring, and Bellabeat Membership.
Despite being a relatively small-scale company, their potential
for growth in technology’s health and wellness sector can be driven by
further analyzing consumer’s behavior towards smart device products. The
insights gained from these consumer behaviors can establish Bellabeat as
a top-competing company in the global smart device market.
2. Ask
i - Identifying the business task:
Gain insights from FitBit Fitness Tracker Data to understand consumer’s fitness habits tracked by the FitBit smart device, and ultimately utilize these insights to assist Bellabeat’s marketing team to make a strategic data-driven plan for driving business growth.
ii - Key stakeholders:
- Primary: Urška Sršen & Sando Mur
- Secondary: Bellabeat Marketing Analytics Team
3. Prepare
i - Dataset Quality & ROCCC Analysis:
The dataset was obtained from FitBit Fitness Tracker Data (Kaggle) which is a CC0: Public Domain, made available through Mobius. It contains fitness data of 30 eligible FitBit users, with the dataset having a total of 18 .csv files.
To determine if the data set is acceptable, we utilize the ROCCC approach:
| Criteria | Rating (1 - Lowest, 5 - Highest) | Comment |
|---|---|---|
| Reliable | 2 | Fitness data was only collected from 30 individuals – this is the minimum sample size to achieve reasonable statistical power. |
| Orignal | 1 | Data was collected from a third party - Amazon Mechanical Turk |
| Comprehensive | 4 | Dataset comprises of a minute, hourly and daily level. Most importantly, insights can already be derived from the daily-level data for activity, intensity, calories, steps and sleep. |
| Current | 3 | Collected from Mar 2016 to May 2016, approximately 7 years ago. 2016 smart devices were most likely base-models for the current devices (2023), so data collected from these older devices already have “foundational” information (such as weight, steps taken, activity intensity, etc.) |
| Cited | 2 | Dataset was originally from zenodo.org which shows at least 1 citation. |
ii - Data verification:
Note: I will first set up my R environment by loading all the packages mentioned below to prepare for my data analysis:
tidyverse - for general data manipulation, cleaning exploration and visualization
janitor - for cleaning and transforming data
ggpmisc - for additional geoms/stats/scales
ggpubr - for additional tools for formatting plots, including themes, labels, etc.
library(tidyverse)
library(janitor)
library(ggpmisc)
library(ggplot2)
library(ggpubr)
library(lubridate)I will be focusing on the daily csv files as it should be able to provide the necessary high-level insights which Bellabeat can use to design an effective marketing strategy. The following files were selected for my data preparation:
dailyActivity_merged.csv
dailyCalories_merged.csv
dailyIntensities_merged.csv
dailySteps_merged.csv
sleepDay_merged.csv
weightLogInfo_merged.csv
Since the dataset contains numerous tables with varying size, I will be using R for a quicker scan of the dataset. It is important to first identify the parameter or category that connects the dataset together.
Running the following code below was enough for me to identify that the ID and Date are two parameters which connects the daily csv files. These parameters will be the basis for my analysis.
dailyActivity <- read_csv("~/R Projects/dailyActivity_merged.csv")## Rows: 940 Columns: 15
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): ActivityDate
## dbl (14): Id, TotalSteps, TotalDistance, TrackerDistance, LoggedActivitiesDi...
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.dailyCalories <- read_csv("~/R Projects/dailyCalories_merged.csv")## Rows: 940 Columns: 3
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): ActivityDay
## dbl (2): Id, Calories
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.dailyIntensities <- read_csv("~/R Projects/dailyIntensities_merged.csv")## Rows: 940 Columns: 10
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): ActivityDay
## dbl (9): Id, SedentaryMinutes, LightlyActiveMinutes, FairlyActiveMinutes, Ve...
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.dailySteps <- read_csv("~/R Projects/dailySteps_merged.csv")## Rows: 940 Columns: 3
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): ActivityDay
## dbl (2): Id, StepTotal
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.sleepDay <- read_csv("~/R Projects/sleepDay_merged.csv")## Rows: 413 Columns: 5
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): SleepDay
## dbl (4): Id, TotalSleepRecords, TotalMinutesAsleep, TotalTimeInBed
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.weightInfo <- read_csv("~/R Projects/weightLogInfo_merged.csv")## Rows: 67 Columns: 8
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): Date
## dbl (6): Id, WeightKg, WeightPounds, Fat, BMI, LogId
## lgl (1): IsManualReport
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.To verify the number of unique user ID in each table, the following code was run:
distinct(dailyActivity, Id)## # A tibble: 33 × 1
## Id
## <dbl>
## 1 1503960366
## 2 1624580081
## 3 1644430081
## 4 1844505072
## 5 1927972279
## 6 2022484408
## 7 2026352035
## 8 2320127002
## 9 2347167796
## 10 2873212765
## # … with 23 more rowsdistinct(dailyCalories, Id)## # A tibble: 33 × 1
## Id
## <dbl>
## 1 1503960366
## 2 1624580081
## 3 1644430081
## 4 1844505072
## 5 1927972279
## 6 2022484408
## 7 2026352035
## 8 2320127002
## 9 2347167796
## 10 2873212765
## # … with 23 more rowsdistinct(dailyIntensities, Id)## # A tibble: 33 × 1
## Id
## <dbl>
## 1 1503960366
## 2 1624580081
## 3 1644430081
## 4 1844505072
## 5 1927972279
## 6 2022484408
## 7 2026352035
## 8 2320127002
## 9 2347167796
## 10 2873212765
## # … with 23 more rowsdistinct(dailySteps, Id)## # A tibble: 33 × 1
## Id
## <dbl>
## 1 1503960366
## 2 1624580081
## 3 1644430081
## 4 1844505072
## 5 1927972279
## 6 2022484408
## 7 2026352035
## 8 2320127002
## 9 2347167796
## 10 2873212765
## # … with 23 more rowsdistinct(sleepDay, Id)## # A tibble: 24 × 1
## Id
## <dbl>
## 1 1503960366
## 2 1644430081
## 3 1844505072
## 4 1927972279
## 5 2026352035
## 6 2320127002
## 7 2347167796
## 8 3977333714
## 9 4020332650
## 10 4319703577
## # … with 14 more rowsdistinct(weightInfo, Id)## # A tibble: 8 × 1
## Id
## <dbl>
## 1 1503960366
## 2 1927972279
## 3 2873212765
## 4 4319703577
## 5 4558609924
## 6 5577150313
## 7 6962181067
## 8 8877689391Notes:
Since the weightLogInfo_merged.csv table has only 8 distinct user ID’s, we will be automatically disregarding this data due to the low number of sample size.
There are inconsistencies in the unique user ID in the tables despite the dataset mentioning that there are only 30 supposed users.
iii - Analysis limitations :
Based on the data verifications, we can see that the sample size is relatively small (n = 30). However, this meets the minimum sample size to form an effective analysis. (Source)
Since the user demographic is unknown or not specified, the data could be involving sampling bias.
The duration of the data collection is also limited (2 months) – which may not be enough to track accurate habits. As per MindOwl, it takes roughly 66 days to form a habit. (Source)
It was stated that this dataset comprised of the response of 30 FitBit users, but upon investigation, there were 33 users (based on the unique user ID). The data did not specify or account for the 3 additional users which could be from erroneous data, or duplicated data.
4. Process
i - Data cleaning & data selection:
To detect all NA and duplicate values from all 5 tables (dailyActivity, dailyCalories, dailyIntensities, dailySteps, and sleepDay), the following code below was run. No NA or duplicate values from tables dailyActivity, dailyCalories, dailyIntensities, and dailySteps were found.
dim(dailyActivity)## [1] 940 15sum(is.na(dailyActivity))## [1] 0sum(duplicated(dailyActivity))## [1] 0dim(dailyCalories)## [1] 940 3sum(is.na(dailyCalories))## [1] 0sum(duplicated(dailyCalories))## [1] 0dim(dailyIntensities)## [1] 940 10sum(is.na(dailyIntensities))## [1] 0sum(duplicated(dailyIntensities))## [1] 0dim(dailySteps)## [1] 940 3sum(is.na(dailySteps))## [1] 0sum(duplicated(dailySteps))## [1] 0For sleepDay, we could see that there are 3 duplicated records, and was cleaned using the code below:
dim(sleepDay)## [1] 413 5sum(is.na(sleepDay))## [1] 0sum(duplicated(sleepDay))## [1] 3## Deleting duplicate records
sleepDay1<- sleepDay[!duplicated(sleepDay), ]
## Rechecking if duplicate records still exist
dim(sleepDay1)## [1] 410 5Summary of removing NA and duplicate values:
| Previous Table Name | New Table Name | NA Values Exist | Duplicates Exist | Changes Made |
|---|---|---|---|---|
| dailyActivity | dailyActivity | No | No | None |
| dailyCalories | dailyCalories | No | No | None |
| dailyIntensities | dailyIntensities | No | No | None |
| dailySteps | dailySteps | No | No | None |
| sleepDay | sleepDay1 | No | Yes | Duplicates removed (3 records) |
To further narrow down which data to be used, we will look at
tables dailyActivity, dailySteps, dailyCalories, and
dailyIntensities.
For dailySteps, a visual inspection shows that data from the
table is already incorporated in the dailyActivity table. To confirm, we
will create a subset of dailyActivity using the first 3 columns, and
compare it to dailySteps.
# Comparing dailyActivity with dailySteps data
dailyActivityTEST1 <- select(dailyActivity, Id, ActivityDate, TotalSteps)
dailyActivityTEST1 <- rename(dailyActivityTEST1, ActivityDay = ActivityDate)
dailyStepsTEST <- rename(dailySteps, TotalSteps = StepTotal)
all.equal(dailyActivityTEST1, dailyStepsTEST, check.attributes = FALSE)## [1] TRUE
Similar to the dailySteps table, we will be repeating this step
with the dailyCalories table:
# Comparing dailyActivity with dailyCalories data
dailyActivityTEST2 <- select(dailyActivity, Id, ActivityDate, Calories)
dailyActivityTEST2 <- rename(dailyActivityTEST2, ActivityDay = ActivityDate)
all.equal(dailyActivityTEST2, dailyCalories, check.attributes = FALSE)## [1] TRUE
Proceeding with the same checks for dailyIntensities table:
# Comparing dailyActivity with dailyIntensities data
dailyActivityTEST3 <- select(dailyActivity, Id, ActivityDate, SedentaryMinutes,
LightlyActiveMinutes, FairlyActiveMinutes, VeryActiveMinutes,
SedentaryActiveDistance, LightActiveDistance,
ModeratelyActiveDistance, VeryActiveDistance)
dailyActivityTEST3 <- rename(dailyActivityTEST3, ActivityDay = ActivityDate)
all.equal(dailyActivityTEST3, dailyIntensities, check.attributes = FALSE)## [1] TRUE
This proves that the dailySteps, dataCalories, and
dataIntensities tables are already incorporated in dailyActivity table,
and future analysis from these three tables can be done directly via the
dailyActivity table; therefore, we will proceed to omit these three
tables.
Final tables to be used for the analysis:
dailyActivity
sleepDay1
We will further clean this data by ensuring that the 2 primary
keys, ID and Date, are in the same format & column
name:
# Cleaning dailyActivity and sleepDay1 data
clean_names(dailyActivity)## # A tibble: 940 × 15
## id activity…¹ total…² total…³ track…⁴ logge…⁵ very_…⁶ moder…⁷ light…⁸
## <dbl> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 1503960366 4/12/2016 13162 8.5 8.5 0 1.88 0.550 6.06
## 2 1503960366 4/13/2016 10735 6.97 6.97 0 1.57 0.690 4.71
## 3 1503960366 4/14/2016 10460 6.74 6.74 0 2.44 0.400 3.91
## 4 1503960366 4/15/2016 9762 6.28 6.28 0 2.14 1.26 2.83
## 5 1503960366 4/16/2016 12669 8.16 8.16 0 2.71 0.410 5.04
## 6 1503960366 4/17/2016 9705 6.48 6.48 0 3.19 0.780 2.51
## 7 1503960366 4/18/2016 13019 8.59 8.59 0 3.25 0.640 4.71
## 8 1503960366 4/19/2016 15506 9.88 9.88 0 3.53 1.32 5.03
## 9 1503960366 4/20/2016 10544 6.68 6.68 0 1.96 0.480 4.24
## 10 1503960366 4/21/2016 9819 6.34 6.34 0 1.34 0.350 4.65
## # … with 930 more rows, 6 more variables: sedentary_active_distance <dbl>,
## # very_active_minutes <dbl>, fairly_active_minutes <dbl>,
## # lightly_active_minutes <dbl>, sedentary_minutes <dbl>, calories <dbl>, and
## # abbreviated variable names ¹activity_date, ²total_steps, ³total_distance,
## # ⁴tracker_distance, ⁵logged_activities_distance, ⁶very_active_distance,
## # ⁷moderately_active_distance, ⁸light_active_distancedailyActivity <- rename_with(dailyActivity, tolower)
dailyActivity <- rename(dailyActivity, date = activitydate)
dailyActivity$date <- as_date(dailyActivity$date, format = "%m/%d/%Y")
clean_names(sleepDay1)## # A tibble: 410 × 5
## id sleep_day total_sleep_records total_minutes_…¹ total…²
## <dbl> <chr> <dbl> <dbl> <dbl>
## 1 1503960366 4/12/2016 12:00:00 AM 1 327 346
## 2 1503960366 4/13/2016 12:00:00 AM 2 384 407
## 3 1503960366 4/15/2016 12:00:00 AM 1 412 442
## 4 1503960366 4/16/2016 12:00:00 AM 2 340 367
## 5 1503960366 4/17/2016 12:00:00 AM 1 700 712
## 6 1503960366 4/19/2016 12:00:00 AM 1 304 320
## 7 1503960366 4/20/2016 12:00:00 AM 1 360 377
## 8 1503960366 4/21/2016 12:00:00 AM 1 325 364
## 9 1503960366 4/23/2016 12:00:00 AM 1 361 384
## 10 1503960366 4/24/2016 12:00:00 AM 1 430 449
## # … with 400 more rows, and abbreviated variable names ¹total_minutes_asleep,
## # ²total_time_in_bedsleepDay1 <- rename_with(sleepDay1, tolower)
sleepDay1 <- rename(sleepDay1, date = sleepday)
sleepDay1$date <- as_date(sleepDay1$date, format = "%m/%d/%Y %I:%M:%S %p", tz=Sys.timezone())
Checking our cleaned data tables:
# Checking cleaned data tables
head(dailyActivity, 3)## # A tibble: 3 × 15
## id date totals…¹ total…² track…³ logge…⁴ verya…⁵ moder…⁶ light…⁷
## <dbl> <date> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 1503960366 2016-04-12 13162 8.5 8.5 0 1.88 0.550 6.06
## 2 1503960366 2016-04-13 10735 6.97 6.97 0 1.57 0.690 4.71
## 3 1503960366 2016-04-14 10460 6.74 6.74 0 2.44 0.400 3.91
## # … with 6 more variables: sedentaryactivedistance <dbl>,
## # veryactiveminutes <dbl>, fairlyactiveminutes <dbl>,
## # lightlyactiveminutes <dbl>, sedentaryminutes <dbl>, calories <dbl>, and
## # abbreviated variable names ¹totalsteps, ²totaldistance, ³trackerdistance,
## # ⁴loggedactivitiesdistance, ⁵veryactivedistance, ⁶moderatelyactivedistance,
## # ⁷lightactivedistancehead(sleepDay1, 3)## # A tibble: 3 × 5
## id date totalsleeprecords totalminutesasleep totaltimeinbed
## <dbl> <date> <dbl> <dbl> <dbl>
## 1 1503960366 2016-04-12 1 327 346
## 2 1503960366 2016-04-13 2 384 407
## 3 1503960366 2016-04-15 1 412 442
Merging our data tables:
daily_activity_sleep <- merge(dailyActivity, sleepDay1, by = c("id", "date"))
head(daily_activity_sleep)## id date totalsteps totaldistance trackerdistance
## 1 1503960366 2016-04-12 13162 8.50 8.50
## 2 1503960366 2016-04-13 10735 6.97 6.97
## 3 1503960366 2016-04-15 9762 6.28 6.28
## 4 1503960366 2016-04-16 12669 8.16 8.16
## 5 1503960366 2016-04-17 9705 6.48 6.48
## 6 1503960366 2016-04-19 15506 9.88 9.88
## loggedactivitiesdistance veryactivedistance moderatelyactivedistance
## 1 0 1.88 0.55
## 2 0 1.57 0.69
## 3 0 2.14 1.26
## 4 0 2.71 0.41
## 5 0 3.19 0.78
## 6 0 3.53 1.32
## lightactivedistance sedentaryactivedistance veryactiveminutes
## 1 6.06 0 25
## 2 4.71 0 21
## 3 2.83 0 29
## 4 5.04 0 36
## 5 2.51 0 38
## 6 5.03 0 50
## fairlyactiveminutes lightlyactiveminutes sedentaryminutes calories
## 1 13 328 728 1985
## 2 19 217 776 1797
## 3 34 209 726 1745
## 4 10 221 773 1863
## 5 20 164 539 1728
## 6 31 264 775 2035
## totalsleeprecords totalminutesasleep totaltimeinbed
## 1 1 327 346
## 2 2 384 407
## 3 1 412 442
## 4 2 340 367
## 5 1 700 712
## 6 1 304 320
Now that the tables have been selected, cleaned, and merged, we
will now move forward with the analysis.
5. Analyze
Our analysis will be focusing on understanding the following:
Relationship between calories burned vs. total number of steps
Relationship between of sedentary minutes and total minutes slept
Percentage of the sample size which is above the daily sedentary minutes threshold
Frequency of smart device usage
i - Relationship between calories vs. total no. of steps:
We will first analyze the data between the amount of calories burned vs. total number of steps taken on a daily basis. To do this, we have utilized the code below:
d_steps_calories <- select(daily_activity_sleep, totalsteps, calories)
ggplot(data = d_steps_calories, aes(x = totalsteps, y = calories)) +
geom_point() +
stat_cor(method = "pearson", type = "text", label.x = 0.5, label.y = 0.9) +
stat_smooth(method = "lm", se = FALSE) +
labs(title = "Calories Burned vs. Total Steps")
From an overall view, we can see that as the number of steps increasing, the calories burned also increases. Initially, we can say that there is a positive correlation. However, the calculated R value (0.41) indicates a low positive correlation. This means that although total steps may have an effect on calories burned, other factors such as body size and composition, birth sex, age, etc. also play a role in calories burned. (Source)
ii - Relationship between sedentary minutes vs. minutes slept:
An interesting analysis would be to understand how sedentary minutes may affect the minutes slept during the day. To form a comparison, the following code was used:
d_sedentary_sleep <- select(daily_activity_sleep, sedentaryminutes, totalminutesasleep)
ggplot(data = d_sedentary_sleep, aes(x = sedentaryminutes, y = totalminutesasleep)) +
geom_point() +
stat_cor(method = "pearson", type = "text", label.x = 0.5, label.y = 0.9) +
stat_smooth(method = "lm", se = FALSE) +
labs(title = "Sedentary Minutes vs. Total Minutes Slept")
From the plot, we can see that an indirectly proportional relationship is observed when comparing sedentary minutes to the time spent asleep (daily level). Although, it is important to note that this relationship only falls on the moderately negative correlation due to the -0.6 R value. However, an article regarding the study of sedentary behavior and sleep quality further highlights that sedentary behavior was associated to a higher risk of insomnia, sleep disturbance, and poorer sleep quality (Source). This study further emphasizes that sleep quality, or in this case, “time slept”, is negatively affected by increased sedentary time.
iii - Quantifying sedentary minutes:
According to the Department of Health (Govt. of Australia), the maximum sedentary time for adults is between 7 – 10 hours (420 – 600 minutes) (Source). To further understand and visualize the percentage of the sample following this recommendation, we have utilized the following code to generate the bar graph shown below.
Getting the average sedentary minutes of each user and creating
a table to track each user into a sequential ID (1, 2, 3, etc.) for
easier plotting:
d_sedentary_total <- select(daily_activity_sleep, id, sedentaryminutes) %>%
group_by(id) %>%
summarise(sedentaryminutes_avg= mean(sedentaryminutes))
d_sedentary_total$id_series <- seq(1, 24)ggplot(d_sedentary_total, aes(x = id_series, y = sedentaryminutes_avg)) +
geom_bar(stat="identity", fill = "darkblue", colour = "black") +
geom_hline(yintercept = 600, linetype = "dashed", color = "red") +
labs(title = "Average Sedentary Minutes")
Note: Each bar in the “id_series” axis represents a unique user (24 bars for 24 unique users)
iv - Measuring user smart device usage
Last case of the analysis would be to understand the frequency of
smart device usage between the 24 unique user IDs. This was done by
counting the number of records per user over the 2-month period of the
data collected.
Smart device usage was categorized into 3:
High Usage = 21 to 30 days
Average Usage = 11 to 20 days
Low Usage = 1 to 10 days
First, we will create a new table to count the number of days
each unique user ID had a record.
d_usage <- daily_activity_sleep %>%
group_by(id) %>%
summarise(days_used = sum(n()))
Then we will modify the table to categorize the users between
usage frequency, and the percent each category represents:
d_usage_summary <- d_usage %>%
mutate(usage = ifelse(d_usage$days_used <= 10, "Low Usage",
ifelse(d_usage$days_used <= 20, "Average Usage", "High Usage"))) %>%
group_by(usage) %>%
summarise(count =n()) %>%
mutate(percent_usage = round((count/24)*100))
Plotting a pie chart to visualize the categorization of users on
usage frequency:
d_usage_summary %>%
ggplot(aes(x = "",y = percent_usage, fill = usage)) +
geom_bar(stat = "identity", width = 1)+
coord_polar("y", start=0)+
theme_minimal()+
theme(axis.title.x= element_blank(),
axis.title.y = element_blank(),
panel.border = element_blank(),
panel.grid = element_blank(),
axis.ticks = element_blank(),
axis.text.x = element_blank(),
plot.title = element_text(hjust = 0.5, size = 14, face = "bold")) +
geom_text(aes(label = c("12%", "50%", "38%")), position = position_stack(vjust = 0.5)) +
scale_fill_manual(values = c("#3982b8","#6abce2","#b8e2f0")) +
labs(title="Daily Usage of Smart Device")
It can be seen from the chart that only 50% of the users have a high usage of the smart devices. This indicates that not all users diligently use their smart devices to track their health and fitness conditions.
6. Act
Bellabeat’s dedication to empower women with knowledge about their own health and habits is critical for their business to continue their rapid growth. With the analysis provided in this case study, Bellabeat will be able to make informed, data-driven decisions, which will further cement their presence in the tech-driven wellness space for women.
Data collection: One key aspect in understanding consumer trends is by collection of the data itself. Bellabeat could utilize their Bellabeat app to store and track user data instead of relying on third party data collection sources. This will ensure that data tracked will include demographic information of users which may be a vital factor in making an informed marketing strategy.
Ultimately, the findings from the analysis phase enabled us to
draw specific recommendations to help guide Bellabeat’s marketing
strategy:
| Recommendation | Descripton |
|---|---|
| Enable notifications for sedentary time | We have classified that 22 out of 24
(91%) of users in the sample size exceeded the
maximum recommended daily sedentary minutes. To tackle this, Bellabeat
could send notifications to users notifying them of their sedentary
minutes by utilizing the geolocation services feature (to track distance
moved). Note that there was a negative correlation with sedentary
minutes vs. minutes slept, indicating that lesser sedentary minutes
could lead to a better quality and longer sleep. Bellabeat could utilize the marketing strategy of “longer and quality” sleep by advocating for a more active lifestyle (reduced sedentary minutes). |
| Enable notifications for inactivity & give daily incentives for smart device usage (daily log-in tracking). | We have quantified that 50% of the sample size
(12% - Average Usage, 38% - Low Usage) were not using
their smart device trackers as often as possible. This insight could
help Bellabeat campaign for the importance of using their smart devices
(as often as possible) to accurately track and record each individual’s
wellness level. Bellabeat could include a daily “log-in” incentive that increases exponentially (Day 1 – Earn 0.25 coins, Day 2 – Earn 0.5 coins, etc.), which caps and resets on a weekly basis (Day 7 – Earn 2 coins). Assuming that Bellabeat has partner organizations, these coins could then be used as a discount voucher for purchases made within those partner businesses. |
| Trophy system for users | The analysis shows that more steps taken = more
calories burned, and less sedentary time = longer and
better quality sleep. This implies that a healthier lifestyle
involves more “movement” and activity. Bellabeat could market the importance of having a healthier lifefestyle by including a trophy system for users who earn specific badges for specific milestones. A sample idea would be: * Roadrunner Badge – achieved when a user completes 50,000 steps in a week * Speed Demon Badge – achieved when a user completes 100,000 steps in a week. These badges could be in categorized in tiers (to highlight degree of difficulty) and should also be publicly available, so that users may display it in the profile section in the Bellabeat app, and share it amongst people within their community. This encourages a healthy sense of competition amongst users. |
All in all, the recommendations listed in the table above
can be further emphasized by reinstating the general (must
have) in Bellabeat’s smart devices:
Include various designs – so users can carry the devices to any occasion
Promote durable and long-lasting batteries – so users will not worry about battery life
Weather-proof build - so the feature-packed devices are complimented by the robust build
Warranty – so the trust and loyalty of consumers remain