The BioSync Wellness Patch is a flexible, non-invasive patch that tracks wellness indicators like stress, sleep, hydration, and activity recovery. A smartphone app integrates real-time data. Price sensitivity, wearable forms, and wellness feature preferences are all examined in this study. The results reveal that people are not very sensitive to price changes in the $50–$200 range, value long battery life, and choose patches and wristbands somewhat above rings. In order to match consumer needs in the wellness tech industry with the BioSync Wellness Patch, these insights will guide design and marketing initiatives.
Introduction
As wearable wellness technology advances quickly, the BioSync Wellness Patch seeks to differentiate itself by providing a simple, non-invasive method that combines integrated health tracking into a single, user-friendly patch. With its discrete and adaptable design, the patch offers ongoing tracking of important health indicators, such as stress, hydration, sleep quality, and exercise recovery, all of which are connected to a smartphone app to deliver individualized insights. Sweat analysis and sophisticated skin sensors are two features that distinguish the BioSync Patch from other wearables. These technologies give customers a more individualized and comprehensive awareness of their health by enabling the patch to measure stress signals like cortisol, hydration levels, and even nutritional balance in real-time.This innovative method gives users access to information that goes beyond simple activity tracking, allowing them to control their wellbeing at a more profound physiological level.In our pilot survey, we also chose to look at other significant product aspects, such as real-time nutrient monitoring, stress and mood tracking, and sleep monitoring. Stress monitoring to ascertain whether users would be interested in or currently monitor their sleep cycles. Monitoring users’ stress levels and ensuring that monitoring is necessary to identify any changes in heart rate, etc., are the goals of stress mood monitoring. Real-time tracking of nutrition to monitor vitamin consumption, blood sugar, and hydration levels. We added three options for battery life: exceptional (lasting more than 14 days), average (lasting up to a week), and poor (lasting fewer than 36 hours) to help people answer conjoint questions and decide which option they would choose based on pricing and battery life. Our main goal is to assist users in prioritizing their health and fitness goals, and we hope that our survey will enable us to achieve this.
Survey Design
We decided that only respondents who were 18 years of age or older should be included in our survey because we assumed that they would be more interested in enhancing their health and would have greater experience and knowledge regarding fitness and health. Our expectation was that they would be more interested in the subject matter than those under the age of 18. We collected the generation to which they belonged and analyzed the responses of each generation to the question to search for any trends. Knowing how each generation reacts to each issue helps us anticipate what to expect. Our questions focused on their race, gender, and annual income. The aim of these analyses is to detect any patterns. People’s willingness to pay for a health tracker device was gauged by asking them about their annual income. In order to obtain a fresh perspective on the similarities and contrasts among the responses of every population and income level, these questions were posed.Our welcome page provides a brief overview of fitness trackers, including their description and the date of their creation. The attributes of the trackers, including type, cost, battery life, location, and activity monitoring, are also included. We recently included location and battery life because we can see how each user chooses to monitor with the various features (low, medium, and premium). With Bluetooth, they can only track steps, which is regarded as low. With wifi, which is regarded as medium, they can track their steps and heart rate; with satellite, which is regarded as premium, they can track their steps, heart rate, and calorie count.This material was supplied so that respondents would have a basic idea of the topics that will be covered when answering the survey’s questions. Six questions make up the choice-based conjoint section, from which participants can select their preferred wellness tracker. There are three alternatives available to respondents. Type, price range (from $50 to $350), battery life (bad, average, or exceptional), activity monitoring (steps, heart rate, and both with calorie count), and location (bluetooth, wifi, satellite) are commonly asked questions. Option 1 will cost $50, and it will feature a patch, bracelet, or horrible and a short battery life. It will also track your position and activities using steps and Bluetooth.The second option will range in price from $150 to $350, have a type, an average battery life, and wifi, steps, and heart rate in addition to activity monitoring and location. The third option will cost between $100 and $350, come in a variety, have a long battery life, and track location and activity using satellite, heart rate, steps, and calories. Since the questions are randomly produced, it sometimes displays two of the same types with different average qualities. There wasn’t much changes we made from the pilot survey to the final survey. The only change we made from adding two more attributes in our conjoint survey questions which is location and activity tracking.
This is our table summarizing the attributes & levels below:
Low
Medium
High
Price
$50
$150
$350
Activity Tracking
Only steps
Steps and heart rate
Heart rate, step and calorie count
Location
Bluetooth
Wifi
Satellite
Battery
Low
Medium
High
Please see the screenshot below for a summary of the meaning of each feature or attribute:
This is a sample conjoint question from the survey.
Code
# Define the option vectorwellness_tracker_buttons_option <-c("option_1", "option_2", "option_3")# Change the names of each element to display markdown-formatted# text and an embedded image using htmlnames(wellness_tracker_buttons_option)[1] <-" **Option 1**<br> <img src='images/patch.jpg' width=150><br> **Type**: Patch<br> **Price**: $ 50 / Unit<br> **Battery life**: Average<br> **Activity Tracking**: Low<br> **Location**: Bluetooth"names(wellness_tracker_buttons_option)[2] <-" **Option 2**<br> <img src='images/wristband.jpg' width=150><br> **Type**: Wristband<br> **Price**: $ 400 / Unit<br> **Battery life**: Excellent<br> **Activity Tracking**: Medium<br> **Location**: Wifi"names(wellness_tracker_buttons_option)[3] <-" **Option 3**<br> <img src='images/ring.jpg' width=150><br> **Type**: Ring<br> **Price**: $ 100 / Unit<br> **Battery life**: Poor<br> **Activity Tracking**: Premium<br> **Location**: Satellite"sd_question(type ='mc_buttons',id ='cbc_practice',label ="Choice question sample",option = wellness_tracker_buttons_option)
*
Code
sd_next()
Data Analysis
Data Cleaning
The raw data had 251 entries. The survey screened out individuals that did not consent to the survey conditions or did not meet age requirements. It also screened out individuals that chose the wrong device that corresponded to the image shown, indicating that the survey taker was not paying attention to their responses. After screening these respondents out, the data was reduced to 202 entries. Since the focus of the data analysis was in the conjoint questions, the filtering process also filtered out individuals that had not responded to all the conjoint questions or that had given the same answers to all the conjoint questions. After this filter, the data was reduced to 195 entries. The final filter dropped the fastest 5% of survey takers, indicating that these individuals had not given proper thought to their answers and responded to the survey in a hurry. After all filtering, the data was reduced to 185 entries.
Sample Description
Looking at our respondent sample data, the total sample size has increased significantly, reflecting a more comprehensive set of responses. However, as with the previous dataset, a small subset of respondents skipped some demographic questions, likely due to privacy concerns or survey fatigue. Responses to race indicated diverse racial representation, with the majority identifying as white, followed by Black, Hispanic, and Asian. Smaller numbers identified as Native or Islander,, and 5 participants preferred not to specify their race. Income data revealed a broad distribution. The largest group earned between $30k-$49k, followed by those earning less than $30k. Smaller groups represented higher income brackets, such as $100k-$149k and $150k or more, with 7 respondents preferring not to disclose their income. Gender representation was mostly female (91), with male respondents (80) and a smaller group identifying as other (4). Millennials dominated the sample, with Gen X and Gen Z also significantly represented. Baby Boomers accounted for a few responses, and 2 respondents preferred not to state their generation.
Below you can find a demographics summary of our cleaned sample:
A sample population’s demographics are broken down by gender, income, generation, and race in the set of bar graphs. The distribution of males and females in the “gender” plot is almost equal. The “income” plot displays a wide variety of income brackets, with the largest percentages falling into the $50k–$74.9k and $74k–$99.9k ranges. The “generation” plot indicates that the sample is dominated by millennials, with tiny distributions from baby boomers coming in second and third, respectively, to Gen Z and Gen X. Other racial groups, including Asians, have smaller proportions than white people, according to the “race” plots. These bar plots aid in comprehending the makeup of the sample and possible data errors.
Power Analysis
Larger samples result in smaller standard errors, particularly for the “price” variable, as this graph shows how sample size affects the accuracy of each coefficient’s estimate. For instance, the coefficients stabilize as the same size rises, and subsequent increases in the same size only slightly lower the standard error.
The link between the sample size and the standard error for different coefficients is displayed in this power analysis figure. The standard error falls with increasing sample size, suggesting more accuracy in predicting the coefficients. The graphic also shows that a standard error of 0.05 may be achieved with a sample size of 600. This guarantees that the research is sufficiently powered to derive a trustworthy result on the impact of various product features.
Modeling
For our final analysis, we have built the models which are listed below.
Simple Logit Model
The model below considers attributes such as the price of the health tracking device, while the four remaining attributes are dummy coefficients. Dummy coefficients have been used for the attributes below:
There are multiple products in the market that have fitness tracking capabilities. Some of our key competitors are
-Apple: With a line of smart watches (wristband) with different tier of qualities for fitness tracking. -Oura: Smart Ring product with different tier of qualities for fitness tracking.
We modeled a market study comparing our products to Apple and Oura. Down below are the simulations coefficients.
When comparing individuals who are categorized as high income ($80k to $150k+) versus those who are categorized as low income($0 to $79k), we found some peculiar results. Using a grouped model we were able to find that Group A (high-income) shows low price sensitivity, as their negative ‘scalePar’ suggests they prioritize premium features over cost. In contrast, Group B (low-income) has a surprising positive ‘scalePar’, indicating they might perceive higher prices as a signal of quality or prestige, rather than a negative. This is unusual for lower-income groups and could reflect unique perceptions or potential issues in model estimation.
For attributes, we were able to see that high-income respondents favor premium features, placing strong emphasis on excellent battery life and premium activity tracking, highlighting their willingness to invest in advanced, high-quality devices. They dislike patch-type devices and have mild negative preferences for ring-type designs. Group A also shows clear liking to Wi-Fi and satellite location tracking, possibly due to privacy concerns or perceived lack of utility.
Low-income respondents (Group B) prioritize practicality and affordability. They strongly prefer patch-type devices and value medium activity tracking, indicating a focus on essential, cost-effective features. Conversely, they have a strong aversion to premium activity tracking and show little interest in excellent battery life, reflecting a preference for simpler, budget-friendly options. While they dislike Wi-Fi tracking, they show mild preference for satellite tracking due to its perceived practicality.
In summary, Group A prioritizes premium, high-performance features, while Group B values affordability and essential functionality, underlining the need for segmented product strategies catering to these distinct preferences.
Results
We demonstrated our results creating different analyses in this section. You can observe how respondents are able to select their preferred alternative in our single market scenario. You may observe how respondents are impacted by different factors to choose which choice they prefer in our multiple market scenario. Price usually has a significant impact based on the utility model; however, we can observe that it doesn’t have a significant impact on our results. In our utility model we were able to see how people don’t necessarily care for extra batteries compared to locations where people might have an interest in. In our willingness to pay plot you are able to see how people are not interested in getting the patch type product compared to a wristband or ring. People tend to want more of a premium activity,due to the added capabilities.These graphs all demonstrate a variety of responses to the questions.When comparing with mainstream brands our product didn’t perform as well hinting maybe that brand is factor that people look into.
This graph’s objective is to model consumer choices in a single market scenario where three competing options( a patch, ring, and wristband) are present. The procedures we followed are:
Uses the loaded model to forecast the market share probability for each option based on the specified parameters (price, kind, and battery).
Determines confidence intervals to illustrate forecast uncertainty.
Shows the anticipated probabilities (market shares) for the three options in a bar chart with error bars.
Our result was a clear visual depiction of how buyers could select between the Ring, Wristband, and Patch in a single market. Based on our results above users don’t have a type preference.
Utility
With the y-axis representing a customer preference analysis, this utility plot shows the links between various attributes (price, battery, etc.) and their corresponding utility values. A relationship of 0 in the “price” plot indicates a low sensitivity to price changes. The “type” figure illustrates clear preference differences by contrasting the utility with types such as patches, rings, and wristbands. The bars indicating errors show the degree of uncertainty in the predictions for the “battery” and “activity” tracking sections, which represent increasing effects on utility. The three location preferences are displayed in the “location” plot, which also offers information on the qualities that respondents value most.
Willingness to pay
The amount that respondents are willing to pay for particular product features is displayed in this plot.The WTP($) is displayed on the y-axis in each of the graphs below that correspond to the attributes (type, battery, activity, and location). As can be seen from the “type” plot, wristbands have the highest WTP. The desire for superior battery life over poor options rises in the “battery” plot. The “activity” plot indicates that premium features have a greater WTP than low and medium features. Bluetooth has a lower WTP than WiFi in the “location” plot. These insights reveal respondents’ preferences, determining what they value most and how much they are willing to spend.
The link between the sample size and the standard error for different coefficients is displayed in this power analysis figure. The standard error falls with increasing sample size, suggesting more accuracy in predicting the coefficients. The graphic also shows that a standard error of 0.05 may be achieved with a sample size of 600. This guarantees that the research is sufficiently powered to derive a trustworthy result on the impact of various product features.
Sensitivity Analysis
The graph predicts how market share will change with different product prices by displaying the Market Share Sensitivity to Price based on data from choiceData.csv. The product’s price ($50–$400 observed range) is represented by the X-axis; and solid lines indicate accurate forecasts within this range. Using a shaded confidence interval representing prediction uncertainty that rises at extreme price ranges, the Y-axis shows market share (0 to 1). Market share is close to 100% at low prices ($50–$100), drops sharply in the price-sensitive mid-range ($150–$250), and reaches a plateau of 0% at high prices ($300–$400), according to the S-shaped curve. The model’s use of logistic decay reveals that the mid-range has the sharpest market share decrease, indicating substantial price sensitivity there, while the extremes’ broader confidence intervals show a lack of data. This graph provides useful information for pricing strategies by highlighting places where changes might be less successful and highlighting competitive price points.
Market Simulation
Our ideal product was built through a process of gathering the attributes that performed best in our model, and combining them to create an ideal product. This product is a $50 patch that has low battery life, medium activity monitoring, and satellite location services. This product went against the well known Apple Watch and Oura Ring, two well established products in the health monitoring device industry.
The market shares of the alternative brands (Apple Watch, Our Ideal Product, and Oura Ring) are contrasted in this market simulation plot. Our Ideal Product has the most market share, indicating that it suits the preferences of the respondents. Although they are similar, the Apple Watch and Oura Ring show smaller market shares. Which suggests that while the two brands are competing, their products are not as closely aligned with our ideal one. All things considered, this simulation demonstrates the brands’ competitive edge and indicates that there would be competition if we introduced our product.
Final Recommendations and Conclusions
It is hard to say for sure whether our product can successfully compete with its present rivals based on the information available. According to our market simulation, buyers perceive our product as competitive in the battery segment, but the willingness-to-pay research indicates that higher pricing decreases market share. To make a definitive conclusion, we need further information, such as a willingness-to-pay plot for particular price ranges. Clarifying the competitive potential of our product would require more survey respondents and in-depth analysis.
Users don’t demonstrate a lot of price sensitivity, probably due to the small difference in pricing.
Users are willing to pay ~$75 more for a product medium battery but not necessarily more for an excellent battery quality.
Users are willing to pay almost $100 more for premium activity tracking.
People are willing to pay ~$25 more for WiFi location capabilities, while they would not be willing to pay for satellite location capabilities.
People are willing to pay more for a ring product type than for a wristband. The WTP significantly lowers when referring to a patch, and that is probably attributed to its disposable nature.
Based on the analysis above, this report recommends to focus on the development of a product that has average battery quality, and premium activity tracking. For different tier options we would advise to have a ring format with WiFi connectivity as the premium tier and wristband and Bluetooth as medium tier. The top opportunities we identified for increasing/demand is to have an excellent battery life and increase for premium activity attributes. Per WTP image, the ring has the highest type while the patch has a low consumer preference.
Limitations
In order to have a well rounded market study, this analysis should dive into the service aspect of this product, that is, its software capabilities and subscription prices; which is a limitation that the presented study does not address. Most products in the market, although they describe their hardware capabilities, they highlight their software and data insights, which our product lightly covers. It would be valuable to dive more into the software counterpart to have a better understanding of the market behavior. Other unknowns that limit this study is an accurate breakdown of the production cost of the product. This plays a vital role in pricing and revenue calculation, which was not easily accessible without diving into supply chain and manufacturing factors, which were outside the scope of this study.
Appendix
Survey
Source Code
---title: "Final Analysis"subtitle: "Health Tracker Device"author: Abdul, Ben, Jade and Katherinedate: "`r format(Sys.time(), '%d %B, %Y')`"format: html: toc: true theme: flatly code-fold: true code-tools: true self-contained: true---```{r}#| label: setup#| include: falseknitr::opts_chunk$set(warning =FALSE,message =FALSE,fig.path ="figs/",fig.width =7.252,fig.height =4,comment ="#>",fig.retina =3)# Load libraries herelibrary(tidyverse)library(here)library(spelling)library(surveydown)#loading data from other filesload(here("models", "mnl_model.RData"))load(here("models", "mnl_wtp.RData"))load(here("models", "market_model.RData"))load("cleaningdata.RData")load(here("data", "demographics.RData"))load(here('models', 'mnl_group_wtp.RData'))#Run on console to check:#spelling::spell_check_files("report.qmd") #SPELLING```# AbstractThe BioSync Wellness Patch is a flexible, non-invasive patch that tracks wellness indicators like stress, sleep, hydration, and activity recovery. A smartphone app integrates real-time data. Price sensitivity, wearable forms, and wellness feature preferences are all examined in this study. The results reveal that people are not very sensitive to price changes in the $50–$200 range, value long battery life, and choose patches and wristbands somewhat above rings. In order to match consumer needs in the wellness tech industry with the BioSync Wellness Patch, these insights will guide design and marketing initiatives.# IntroductionAs wearable wellness technology advances quickly, the BioSync Wellness Patch seeks to differentiate itself by providing a simple, non-invasive method that combines integrated health tracking into a single, user-friendly patch. With its discrete and adaptable design, the patch offers ongoing tracking of important health indicators, such as stress, hydration, sleep quality, and exercise recovery, all of which are connected to a smartphone app to deliver individualized insights. Sweat analysis and sophisticated skin sensors are two features that distinguish the BioSync Patch from other wearables. These technologies give customers a more individualized and comprehensive awareness of their health by enabling the patch to measure stress signals like cortisol, hydration levels, and even nutritional balance in real-time.This innovative method gives users access to information that goes beyond simple activity tracking, allowing them to control their wellbeing at a more profound physiological level.In our pilot survey, we also chose to look at other significant product aspects, such as real-time nutrient monitoring, stress and mood tracking, and sleep monitoring. Stress monitoring to ascertain whether users would be interested in or currently monitor their sleep cycles. Monitoring users' stress levels and ensuring that monitoring is necessary to identify any changes in heart rate, etc., are the goals of stress mood monitoring. Real-time tracking of nutrition to monitor vitamin consumption, blood sugar, and hydration levels. We added three options for battery life: exceptional (lasting more than 14 days), average (lasting up to a week), and poor (lasting fewer than 36 hours) to help people answer conjoint questions and decide which option they would choose based on pricing and battery life. Our main goal is to assist users in prioritizing their health and fitness goals, and we hope that our survey will enable us to achieve this.# Survey DesignWe decided that only respondents who were 18 years of age or older should be included in our survey because we assumed that they would be more interested in enhancing their health and would have greater experience and knowledge regarding fitness and health. Our expectation was that they would be more interested in the subject matter than those under the age of 18. We collected the generation to which they belonged and analyzed the responses of each generation to the question to search for any trends. Knowing how each generation reacts to each issue helps us anticipate what to expect. Our questions focused on their race, gender, and annual income. The aim of these analyses is to detect any patterns. People's willingness to pay for a health tracker device was gauged by asking them about their annual income. In order to obtain a fresh perspective on the similarities and contrasts among the responses of every population and income level, these questions were posed.Our welcome page provides a brief overview of fitness trackers, including their description and the date of their creation. The attributes of the trackers, including type, cost, battery life, location, and activity monitoring, are also included. We recently included location and battery life because we can see how each user chooses to monitor with the various features (low, medium, and premium). With Bluetooth, they can only track steps, which is regarded as low. With wifi, which is regarded as medium, they can track their steps and heart rate; with satellite, which is regarded as premium, they can track their steps, heart rate, and calorie count.This material was supplied so that respondents would have a basic idea of the topics that will be covered when answering the survey's questions. Six questions make up the choice-based conjoint section, from which participants can select their preferred wellness tracker. There are three alternatives available to respondents. Type, price range (from $50 to $350), battery life (bad, average, or exceptional), activity monitoring (steps, heart rate, and both with calorie count), and location (bluetooth, wifi, satellite) are commonly asked questions. Option 1 will cost $50, and it will feature a patch, bracelet, or horrible and a short battery life. It will also track your position and activities using steps and Bluetooth.The second option will range in price from $150 to $350, have a type, an average battery life, and wifi, steps, and heart rate in addition to activity monitoring and location. The third option will cost between $100 and $350, come in a variety, have a long battery life, and track location and activity using satellite, heart rate, steps, and calories. Since the questions are randomly produced, it sometimes displays two of the same types with different average qualities. There wasn’t much changes we made from the pilot survey to the final survey. The only change we made from adding two more attributes in our conjoint survey questions which is location and activity tracking.This is our table summarizing the attributes & levels below: <table class="table table-hover"> <thead> <tr class="table-light"> <th scope="col"> </th> <th scope="col">Low</th> <th scope="col">Medium</th> <th scope="col">High</th> </tr> </thead> <tbody> <tr class="table-light"> <th scope="row">Price</th> <td>$50 </td> <td>$150</td> <td>$350</td> </tr> <tr> <tr class="table-light"> <th scope="row">Activity Tracking</th> <td>Only steps</td> <td>Steps and heart rate</td> <td>Heart rate, step and calorie count</td> </tr> <tr class="table-light"> <th scope="row">Location</th> <td>Bluetooth</td> <td>Wifi</td> <td>Satellite</td> </tr> <tr class="table-light"> <th scope="row">Battery</th> <td>Low</td> <td>Medium</td> <td>High</td> </tr> </tbody></table>Please see the screenshot below for a summary of the meaning of each feature or attribute:<center><img src="images/survey4.jpg" width="600"/></center><center><img src="images/survey5.jpg" width="600"/></center>This is a sample conjoint question from the survey. ```{r}# Define the option vectorwellness_tracker_buttons_option <-c("option_1", "option_2", "option_3")# Change the names of each element to display markdown-formatted# text and an embedded image using htmlnames(wellness_tracker_buttons_option)[1] <-" **Option 1**<br> <img src='images/patch.jpg' width=150><br> **Type**: Patch<br> **Price**: $ 50 / Unit<br> **Battery life**: Average<br> **Activity Tracking**: Low<br> **Location**: Bluetooth"names(wellness_tracker_buttons_option)[2] <-" **Option 2**<br> <img src='images/wristband.jpg' width=150><br> **Type**: Wristband<br> **Price**: $ 400 / Unit<br> **Battery life**: Excellent<br> **Activity Tracking**: Medium<br> **Location**: Wifi"names(wellness_tracker_buttons_option)[3] <-" **Option 3**<br> <img src='images/ring.jpg' width=150><br> **Type**: Ring<br> **Price**: $ 100 / Unit<br> **Battery life**: Poor<br> **Activity Tracking**: Premium<br> **Location**: Satellite"sd_question(type ='mc_buttons',id ='cbc_practice',label ="Choice question sample",option = wellness_tracker_buttons_option)sd_next()```# Data Analysis## Data CleaningThe raw data had `r nraw_data` entries. The survey screened out individuals that did not consent to the survey conditions or did not meet age requirements. It also screened out individuals that chose the wrong device that corresponded to the image shown, indicating that the survey taker was not paying attention to their responses. After screening these respondents out, the data was reduced to `r num_afterscreenout` entries. Since the focus of the data analysis was in the conjoint questions, the filtering process also filtered out individuals that had not responded to all the conjoint questions or that had given the same answers to all the conjoint questions. After this filter, the data was reduced to `r num_afterCQ` entries. The final filter dropped the fastest 5% of survey takers, indicating that these individuals had not given proper thought to their answers and responded to the survey in a hurry. After all filtering, the data was reduced to `r nRespondents` entries. ## Sample DescriptionLooking at our respondent sample data, the total sample size has increased significantly, reflecting a more comprehensive set of responses. However, as with the previous dataset, a small subset of respondents skipped some demographic questions, likely due to privacy concerns or survey fatigue. Responses to race indicated diverse racial representation, with the majority identifying as white, followed by Black, Hispanic, and Asian. Smaller numbers identified as Native or Islander,, and 5 participants preferred not to specify their race. Income data revealed a broad distribution. The largest group earned between $30k-$49k, followed by those earning less than $30k. Smaller groups represented higher income brackets, such as $100k-$149k and $150k or more, with 7 respondents preferring not to disclose their income. Gender representation was mostly female (91), with male respondents (80) and a smaller group identifying as other (4). Millennials dominated the sample, with Gen X and Gen Z also significantly represented. Baby Boomers accounted for a few responses, and 2 respondents preferred not to state their generation.Below you can find a demographics summary of our cleaned sample:<center><img src="figs/plot_demographics.png" width="600"/></center>A sample population's demographics are broken down by gender, income, generation, and race in the set of bar graphs. The distribution of males and females in the "gender" plot is almost equal. The "income" plot displays a wide variety of income brackets, with the largest percentages falling into the $50k–$74.9k and $74k–$99.9k ranges. The "generation" plot indicates that the sample is dominated by millennials, with tiny distributions from baby boomers coming in second and third, respectively, to Gen Z and Gen X. Other racial groups, including Asians, have smaller proportions than white people, according to the "race" plots. These bar plots aid in comprehending the makeup of the sample and possible data errors. ## Power AnalysisLarger samples result in smaller standard errors, particularly for the "price" variable, as this graph shows how sample size affects the accuracy of each coefficient's estimate. For instance, the coefficients stabilize as the same size rises, and subsequent increases in the same size only slightly lower the standard error.<center><img src="figs/power_plot.png" width="650"/></center>The link between the sample size and the standard error for different coefficients is displayed in this power analysis figure. The standard error falls with increasing sample size, suggesting more accuracy in predicting the coefficients. The graphic also shows that a standard error of 0.05 may be achieved with a sample size of 600. This guarantees that the research is sufficiently powered to derive a trustworthy result on the impact of various product features. # ModelingFor our final analysis, we have built the models which are listed below.## Simple Logit ModelThe model below considers attributes such as the price of the health tracking device, while the four remaining attributes are dummy coefficients.Dummy coefficients have been used for the attributes below:_Reference Levels:_- **Battery:** poor- **Type:** wristband- **Activity:** Low- **Location:** Bluetooth$$u_j = \beta_1 x_j^{\mathrm{price}} + \beta_2 \delta_j^{\mathrm{ring}}+ \beta_3 \delta_j^{\mathrm{patch}} + \beta_4 \delta_j^{\mathrm{average}} + \beta_5 \delta_j^{\mathrm{excellent}} +\beta_6 \delta_j^{\mathrm{activMedium}} $$$$+\beta_7 \delta_j^{\mathrm{activPremium}} + \beta_8 \delta_j^{\mathrm{Wifi}} + \beta_9 \delta_j^{\mathrm{satellite}} +\varepsilon_j$$### Model coefficients#### Utility Coefficients```{r}print(coefs) ```##### Standard Errors```{r}print(covariance)```## Mixed Logit ModelThere are multiple products in the market that have fitness tracking capabilities. Some of our key competitors are -Apple: With a line of smart watches (wristband) with different tier of qualities for fitness tracking.-Oura: Smart Ring product with different tier of qualities for fitness tracking.We modeled a market study comparing our products to Apple and Oura. Down below are the simulations coefficients. ```{r}print(sim_mnl_linear)```## Subgroup Logit Model### Utility CoefficientsWe performed an analysis among survey takers dividing them into High Income Earners and Low Income Earners. Below are the results: **Group A: High Income Earners**```{r}print(coefs_A)```**Group B: Low Income Earners**```{r}print(coefs_B)```When comparing individuals who are categorized as high income ($80k to $150k+) versus those who are categorized as low income($0 to $79k), we found some peculiar results. Using a grouped model we were able to find that Group A (high-income) shows low price sensitivity, as their negative ‘scalePar’ suggests they prioritize premium features over cost. In contrast, Group B (low-income) has a surprising positive ‘scalePar’, indicating they might perceive higher prices as a signal of quality or prestige, rather than a negative. This is unusual for lower-income groups and could reflect unique perceptions or potential issues in model estimation.For attributes, we were able to see that high-income respondents favor premium features, placing strong emphasis on excellent battery life and premium activity tracking, highlighting their willingness to invest in advanced, high-quality devices. They dislike patch-type devices and have mild negative preferences for ring-type designs. Group A also shows clear liking to Wi-Fi and satellite location tracking, possibly due to privacy concerns or perceived lack of utility.Low-income respondents (Group B) prioritize practicality and affordability. They strongly prefer patch-type devices and value medium activity tracking, indicating a focus on essential, cost-effective features. Conversely, they have a strong aversion to premium activity tracking and show little interest in excellent battery life, reflecting a preference for simpler, budget-friendly options. While they dislike Wi-Fi tracking, they show mild preference for satellite tracking due to its perceived practicality.In summary, Group A prioritizes premium, high-performance features, while Group B values affordability and essential functionality, underlining the need for segmented product strategies catering to these distinct preferences. # ResultsWe demonstrated our results creating different analyses in this section. You can observe how respondents are able to select their preferred alternative in our single market scenario. You may observe how respondents are impacted by different factors to choose which choice they prefer in our multiple market scenario. Price usually has a significant impact based on the utility model; however, we can observe that it doesn’t have a significant impact on our results. In our utility model we were able to see how people don’t necessarily care for extra batteries compared to locations where people might have an interest in. In our willingness to pay plot you are able to see how people are not interested in getting the patch type product compared to a wristband or ring. People tend to want more of a premium activity,due to the added capabilities.These graphs all demonstrate a variety of responses to the questions.When comparing with mainstream brands our product didn’t perform as well hinting maybe that brand is factor that people look into. <center><img src="figs/market_share.jpg" width="600"/></center> This graph's objective is to model consumer choices in a single market scenario where three competing options( a patch, ring, and wristband) are present. The procedures we followed are:1. Uses the loaded model to forecast the market share probability for each option based on the specified parameters (price, kind, and battery).2. Determines confidence intervals to illustrate forecast uncertainty.3. Shows the anticipated probabilities (market shares) for the three options in a bar chart with error bars.Our result was a clear visual depiction of how buyers could select between the Ring, Wristband, and Patch in a single market. Based on our results above users don’t have a type preference. ## Utility<center><img src="figs/mnl_linear_unc.png" width="600"/></center>With the y-axis representing a customer preference analysis, this utility plot shows the links between various attributes (price, battery, etc.) and their corresponding utility values. A relationship of 0 in the "price" plot indicates a low sensitivity to price changes. The "type" figure illustrates clear preference differences by contrasting the utility with types such as patches, rings, and wristbands. The bars indicating errors show the degree of uncertainty in the predictions for the "battery" and "activity" tracking sections, which represent increasing effects on utility. The three location preferences are displayed in the "location" plot, which also offers information on the qualities that respondents value most. ## Willingness to payThe amount that respondents are willing to pay for particular product features is displayed in this plot.The WTP($) is displayed on the y-axis in each of the graphs below that correspond to the attributes (type, battery, activity, and location). As can be seen from the "type" plot, wristbands have the highest WTP. The desire for superior battery life over poor options rises in the "battery" plot. The "activity" plot indicates that premium features have a greater WTP than low and medium features. Bluetooth has a lower WTP than WiFi in the "location" plot. These insights reveal respondents' preferences, determining what they value most and how much they are willing to spend.<center><img src="figs/mnl_wtp.png" width="600"/></center>The link between the sample size and the standard error for different coefficients is displayed in this power analysis figure. The standard error falls with increasing sample size, suggesting more accuracy in predicting the coefficients. The graphic also shows that a standard error of 0.05 may be achieved with a sample size of 600. This guarantees that the research is sufficiently powered to derive a trustworthy result on the impact of various product features.## Sensitivity Analysis<center><img src="figs/share_price_plot_sensitivity.png" width="500"/></center>The graph predicts how market share will change with different product prices by displaying the Market Share Sensitivity to Price based on data from `choiceData.csv`. The product's price ($50–$400 observed range) is represented by the X-axis; and solid lines indicate accurate forecasts within this range. Using a shaded confidence interval representing prediction uncertainty that rises at extreme price ranges, the Y-axis shows market share (0 to 1). Market share is close to 100% at low prices ($50–$100), drops sharply in the price-sensitive mid-range ($150–$250), and reaches a plateau of 0% at high prices ($300–$400), according to the S-shaped curve. The model's use of logistic decay reveals that the mid-range has the sharpest market share decrease, indicating substantial price sensitivity there, while the extremes' broader confidence intervals show a lack of data. This graph provides useful information for pricing strategies by highlighting places where changes might be less successful and highlighting competitive price points.## Market Simulation <center><img src="figs/marketsim.png" width="500"/></center>Our ideal product was built through a process of gathering the attributes that performed best in our model, and combining them to create an ideal product. This product is a $50 patch that has low battery life, medium activity monitoring, and satellite location services. This product went against the well known Apple Watch and Oura Ring, two well established products in the health monitoring device industry. The market shares of the alternative brands (Apple Watch, Our Ideal Product, and Oura Ring) are contrasted in this market simulation plot. Our Ideal Product has the most market share, indicating that it suits the preferences of the respondents. Although they are similar, the Apple Watch and Oura Ring show smaller market shares. Which suggests that while the two brands are competing, their products are not as closely aligned with our ideal one. All things considered, this simulation demonstrates the brands’ competitive edge and indicates that there would be competition if we introduced our product. # Final Recommendations and ConclusionsIt is hard to say for sure whether our product can successfully compete with its present rivals based on the information available. According to our market simulation, buyers perceive our product as competitive in the battery segment, but the willingness-to-pay research indicates that higher pricing decreases market share. To make a definitive conclusion, we need further information, such as a willingness-to-pay plot for particular price ranges. Clarifying the competitive potential of our product would require more survey respondents and in-depth analysis.- Users don’t demonstrate a lot of price sensitivity, probably due to the small difference in pricing. - Users are willing to pay ~$75 more for a product medium battery but not necessarily more for an excellent battery quality. - Users are willing to pay almost $100 more for premium activity tracking. - People are willing to pay ~$25 more for WiFi location capabilities, while they would not be willing to pay for satellite location capabilities. - People are willing to pay more for a ring product type than for a wristband. The WTP significantly lowers when referring to a patch, and that is probably attributed to its disposable nature. Based on the analysis above, this report recommends to focus on the development of a product that has average battery quality, and premium activity tracking. For different tier options we would advise to have a ring format with WiFi connectivity as the premium tier and wristband and Bluetooth as medium tier. The top opportunities we identified for increasing/demand is to have an excellent battery life and increase for premium activity attributes. Per WTP image, the ring has the highest type while the patch has a low consumer preference. # LimitationsIn order to have a well rounded market study, this analysis should dive into the service aspect of this product, that is, its software capabilities and subscription prices; which is a limitation that the presented study does not address. Most products in the market, although they describe their hardware capabilities, they highlight their software and data insights, which our product lightly covers. It would be valuable to dive more into the software counterpart to have a better understanding of the market behavior. Other unknowns that limit this study is an accurate breakdown of the production cost of the product. This plays a vital role in pricing and revenue calculation, which was not easily accessible without diving into supply chain and manufacturing factors, which were outside the scope of this study. # Appendix## Survey<center><img src="images/survey1.jpg" width="600"/></center><center><img src="images/survey2.jpg" width="600"/></center><center><img src="images/survey3.jpg" width="600"/></center><center><img src="images/survey4.jpg" width="600"/></center><center><img src="images/survey5.jpg" width="600"/></center><center><img src="images/survey6.jpg" width="600"/></center><center><img src="images/survey7.jpg" width="600"/></center><center><img src="images/survey8.jpg" width="600"/></center><center><img src="images/survey9.jpg" width="600"/></center><center><img src="images/survey10.jpg" width="600"/></center><center><img src="images/survey11.jpg" width="600"/></center><center><img src="images/survey12.jpg" width="600"/></center>
