1 of 37

Brewed Awakening

12/1/24

HOYALYTICS

2 of 37

Laksh Patel (he/him)

Data Consultant

Finance & OPAN | ‘27

Jessica Chen (she/her)

Data Consultant

Finance| ‘28

Serafin Burgulla (he/him)

Data Scientist

Computer Science | ‘28

Ellie Seo (she/her)

Data Consultant

Economics | ‘28

Meet our team

Adrian Ng (he/him)

Data Scientist

Computer Science | ‘26

HOYALYTICS

3 of 37

Client Deliverable

Presented by Hoyalytics

HOYALYTICS

4 of 37

Client Overview

Tableau: Data Analyzation

Python: Advanced Modelling

Recommendations

AGENDA

HOYALYTICS

5 of 37

Current State of Operations

HOYALYTICS

3

5

LOCATIONS

CHANNELS

PRODUCTS

6 of 37

Sales Data

HOYALYTICS

7 of 37

Even Split on Revenue Generation

The products, coffee and sandwich, generate the most revenue, 20.6% & 20.2% respectively

Tea sales lag behind in terms of revenue generation, 19.3%

Overall, an even split suggests that all products are valued and demanded

Revenue by Product

HOYALYTICS

8 of 37

Revenue Generated Dependent on Month

High points in Mar, May, Aug
Low points in Feb, Apr

Sum of Revenue Per Month

Spike in August
Possible cannibalization

Revenue of Food Per Month

High sales in May
Low sales in February

Revenue of Drinks Per Month

HOYALYTICS

9 of 37

Products sell and generate more revenue closer to borders

Revenue & Quantity Sold Map

Cause: travelers or commuters who find these locations to be convenient and accessible

Enhance drive thru in border locations

Cross-border Traffic

HOYALYTICS

10 of 37

Mobile Orders on the Rise

Purchase Method vs Product Type Heatmap

33.9% of products sold through mobile orders -> increase in online orders

Coffee and pastries taking the lead in mobile cart orders

Focus on improving customer experiences using mobile carts.

HOYALYTICS

11 of 37

Time of Day Impacting Sales

Revenue vs Weekday & Time of Day

Majority of the revenue generated is during opening (33.4%) -> convenient before work, freshness of products

The evening generates the least amount of revenue (16.6%) -> dinner times

Promotions during morning and night could attract more customers and increase sales

Morning Demands

HOYALYTICS

12 of 37

Location Data

HOYALYTICS

13 of 37

Demographic Age Data

Lack of Sales from 50+
Majority from 36-50

Age v Monthly Sales

Similar CCI
Highest from 50+

Age v CCI

Lack of traffic from 50+
Majority again 36-50

Age v Foot Traffic

HOYALYTICS

14 of 37

Indefinite Relationship between Median Income & CCI

Highest income segment doesn’t have a high CCI.
Variable CCI between the income segments.

Target the lower income segment.

Median Income v CCI

HOYALYTICS

15 of 37

Foot Traffic ≠ Sales

Foot Traffic v Avg Monthly Sales Map

Possible cannibalization in Seattle area
No strong correlation between Location & Sales/Foot Traffic

Yakima has the most sales

HOYALYTICS

16 of 37

K-Means Clustering of Location Data Reveals Groups of Locations that Share Similar Characteristics

This allows us to understand the distinct types of markets we operate in.

Although clusters may not reveal dramatic differences, this process still helps us identify locations with similar needs and challenges.

E.g. High-income cluster with average sales suggest opportunities for targeted offers such as premium product lines.

3D K Means Clustering of Locations

Summary of Cluster Metrics

HOYALYTICS

Adrian

Cluster 1:

Foot Traffic: 6,051.75 (Moderate to High)

Avg Monthly Sales: $15,243.25 (Moderate to High)

Median Income: $92,558.25 (High)

Locations in this cluster have high median income and moderately high foot traffic, resulting in a moderately high average monthly sales. The high median income indicates that customers might be more willing to spend, but sales could be improved with targeted offers. This could be a high-value opportunity if you can further boost engagement.

Cluster 2:

Foot Traffic: 1,841.00 (Low)

Avg Monthly Sales: $14,969.67 (Moderate)

Median Income: $49,206.67 (Low)

These locations have low foot traffic and low median income but moderate average monthly sales. Despite low foot traffic and lower-income customers, these locations achieve moderate sales, suggesting there might be a loyal customer base or that they are located in areas where customers prioritize spending on your products. Consider whether these locations could benefit from increased marketing efforts to boost foot traffic.

Cluster 3:

Foot Traffic: 7,905.33 (High)

Avg Monthly Sales: $17,521.67 (High)

Median Income: $49,220.00 (Low)

Locations in this cluster have high foot traffic but a low median income, with high average monthly sales. Despite lower income levels, customers here are highly engaged, likely due to affordable pricing or product preferences that align well with the community. These locations might benefit from loyalty programs or additional product offerings that cater to high-volume customers. This cluster represents high-performing locations in terms of sales, which indicates that you should maintain or increase resource allocation to maximize these high-traffic, high-sales areas.

Cluster 4:

Foot Traffic: 4,336.90 (Moderate)

Avg Monthly Sales: $7,660.80 (Low)

Median Income: $60,693.30 (Moderate)

This cluster has moderate foot traffic and moderate income, but low average monthly sales. This suggests there may be an untapped opportunity to improve sales through targeted strategies, such as promotional campaigns, special events, or expanding product variety. The moderate median income indicates that customers have the potential spending power, but they may need incentives to purchase more. Improvement Opportunities: Consider analyzing customer preferences in these locations to understand why sales are lower and then adjust pricing, product mix, or marketing to better fit their needs.

These initial results suggest that there may be more similarities across our locations than we expected. For the next iteration, we could:

Add new metrics to further differentiate locations.
Focus on customer behavior metrics—such as repeat visits or loyalty engagement—to see if we can derive more actionable clusters.

Explain How Clustering Still Informs Strategy:

Even without dramatic insights, the results are still valuable. Having a baseline understanding of our clusters, even if they’re not drastically different, helps us make data-driven decisions about which locations require similar management strategies, such as pricing, staffing, or marketing campaigns.

17 of 37

Loyalty Data

HOYALYTICS

18 of 37

Effects of the Loyalty Program

Average Total Revenue

Ineffective loyalty program

Average Lifetime Value

Average # of Transactions

HOYALYTICS

19 of 37

Effects of the Loyalty Program

Total Revenue

Infectivity is worsened compared to the average value
Evidence that the Loyalty Program does not attract customers

Total Lifetime Value

Total # of Transactions

Transformation in the Loyalty Program is needed.

HOYALYTICS

20 of 37

Customer Segmentation Insights

Helps prioritize our high-frequency, high-spending loyal customer base and tailor marketing and operational strategies to increase profit margins.
Obtain customer IDs every time we sell a product to cross-reference with sales data, location, and time.
We can then create location-specific targeted promotions and exclusive offers to high-value, loyal customers.

Driving Targeted Strategies with K-Means Clustering

HOYALYTICS

21 of 37

Feedback Data

HOYALYTICS

22 of 37

Review Keywords with High Correlation

Location ID vs Review Keywords

Poor quality is less mentioned on shops with a clean environment and friendly staff
The good quality of coffee does not lead to a better quality rating

HOYALYTICS

23 of 37

Frequency of Review Keywords

A clean environment is important for good ratings.

HOYALYTICS

24 of 37

Keyword TF/IDF Analysis by Location

With feedback data, we analyze keyword prominence by location with a TF/IDF model. �The heatmap shows locations and keywords measuring relevance.

This helps Brewed Awakening identify strengths and areas for improvement.

25 of 37

Actionable Recommendations based on TF-IDF Analysis

Tailor location-specific improvement plans based on their feedback profile.

Mitigate critical issues such as "long wait" in underperforming locations like Vancouver and Tacoma.
E.G. Invest in staff efficiency training.

Align marketing strategies with strengths

Optimize location- specific marketing campaigns and resource allocation.
Facilitate data-driven recommendations, leveraging strengths as key brand differentiators.

HOYALYTICS

26 of 37

Sentiment Analysis of Customer Feedback by Location

Assigned individual sentiment scores for every customer feedback using a pre-existing dictionary, then computed average sentiment scores for each location

Focus on improving customer experience in cities with negative scores while leveraging successful strategies from top-performing locations.

HOYALYTICS

27 of 37

Continuous Improvement using Machine Learning

Real-time updates on store likeability

Builds historical data for trend analysis & targeted improvements

Identifies strengths & areas of improvement for each location

Future / Continuous Uses of Sentiment Analysis

More Subjective Feedback Options

Present - Feedback relies on rigid options or flawed keyword isolation

Future - Open-ended customer feedback

Focus on most critical negative reviews for actionable improvements

Flexibility in Feedback Prioritization

Custom dictionaries can be created to match priorities (e.g. good coffee > low price)

Enables Brewed Awakening to align sentiment scoring with brand goals.

HOYALYTICS

28 of 37

THANK YOU!

Connect with Hoyalytics!

@hoyalytics

29 of 37

HOYALYTICS STYLE GUIDE

COLOR PALETTE

Dark Blue

#072f5f

Dodger Blue

#1261a0

True Blue

#3895d3

Turquoise

#58cced

Sky Blue

#cbeaff

Faded Blue

#ecf7ff

FONT

Lexend - titles

Avenir - main text

Dark Grey

#595959

Light Grey

#eeeeee

30 of 37

A sentence that provides the audience the main takeaway from the graphs.

X% of our respondents have [finding].
Another bullet point that summarizes the insights offered by the chart.

An example recommendation based on the charts on the slide.

Current State of Operations / Profit Analysis / Recommendations / Machine Learning Applications

HOYALYTICS

31 of 37

Sentence about main takeaway from chart

Only 23-25% of [findings].
Another bullet point on your findings.

Finding

Current State of Operations / Profit Analysis / Recommendations / Machine Learning Applications

HOYALYTICS

32 of 37

Main Statement

Point #1
Point #2
Point #3

HOYALYTICS

33 of 37

Two Different Approaches

HOYALYTICS

34 of 37

Template Table Slide

35 of 37

Template Subtitle Slide

Detail 1
Detail 2
Detail 3

Subtitle 1

Detail 1
Detail 2
Detail 3

Subtitle 2

Detail 1
Detail 2
Detail 3

Subtitle 3

HOYALYTICS

36 of 37

Biggest takeaway from the data given, provides analysis that connects the two graphs together.

Main problem statement.

Evidence that backs up this statement.

Even more evidence that backs up the statement.

Description of first graph.

Description of second graph.

Footnote 1.
Footnote 2.

HOYALYTICS

37 of 37

3 Different Approaches

HOYALYTICS