A key responsibility for any data team is to understand the core metrics driving their business. Starting from the top, these metrics often include figures like gross revenue and expenses. However, these high-level metrics can feel too far removed and abstracted from the actual business.
Many companies, therefore, break down these top-line metrics into more specific, easy to understand(in terms of how they impact the business) ones that collectively build up to key business goals. These metric trees reveal deeper layers of context, connecting individual data points to overarching metrics. The further you dive, the more insight you gain into the underlying factors that shape each metric.
In this article, we’ll explore a real-world example of one of a real company’s key metrics, and how data teams could approach better understanding how to be strategic.
Starbucks’ same-store sales decline
If you’ve been following the latest from America’s favorite coffee and sugar seller, Starbucks, you’ll know they recently reported a decline in comparable store sales:
“Starbucks said global same-store sales fell 7 percent for the fourth-quarter that ended in September, with a decline of 6 percent in North America and a 14 percent drop in China.” - Starbucks Reports a Slide in Sales and Traffic - NYT
For this article, we'll focus on comparable store sales, also known as same-store sales. This metric compares a store's revenue over a specific period to its revenue during a similar period in the past.
As a caveat, the calculation can vary by company, with different organizations including or excluding specific data points.
Now that you’ve got a basic understanding, the next question is: what underlying metrics drive this figure? Luckily, Starbucks provides those as well.
“Comparable store sales 6% decline in U.S. comparable store sales, driven by a 10% decline in comparable transactions, partially offset by a 4% increase in average ticket…Additionally, China comparable store sales declined 14%, driven by an 8% decline in average ticket compounded by a 6% decline in comparable transactions, weighed down by intensified competition and a soft macro environment that impacted consumer spending.” - Starbucks 8-K Oct 30th
These two metrics—comparable transactions and average ticket—are the core components of comparable store sales. This makes sense, if you think about it, as comparable transactions represent the number of transactions, while average ticket reflects the average amount spent per transaction.
Thus, as the 8-K report states, the decline in the U.S. comparable stores was
“driven by a 10% decline in comparable transactions, partially offset by a 4% increase in average ticket.”
This highlights how a drop in one metric—like the number of transactions—can be somewhat balanced by another, such as average ticket size. In other words, even if fewer transactions occur, increasing either the price of the product or the quantity purchased per transaction can help mitigate the loss.
With a baseline understanding of what a company’s key metric is, data teams are better positioned to make a strategic impact. Instead of being distracted by projects and ad hoc requests, they can focus on initiatives to improve these metrics—assuming their team has the ability to influence them.
A data team doesn’t have to be a passive cost center or task-taker. To elevate its role, it must understand the key metrics, the business context, and the technology that can meaningfully impact the top and/or bottom line. Without that context, even the best data insights are limited in their potential.
Let’s dig into some of that context.
Digging Deeper Into The Context
One notable line from the 8-K report mentions that Starbucks’ sales decline in China was “weighed down by intensified competition.”
Having recently spent 10 days in China, specifically in the Shanghai/Hangzhou/Suzhou area, my wife and I both wondered why someone might go to Starbucks given the range of popular, more affordable coffee brands available. During our trip, I found myself at two local chains—Luckin Coffee and Manner Coffee—logging around 20 cups in 10 days(apparently another large competitor is Cotti, but I actually never saw it).
While price isn’t the only factor driving customer choice, as I talk about in some of my consulting content, it’s certainly impactful. Here’s a general comparison of prices(7 RMBs are about 1 USD):
In addition to their competitive pricing, most Luckin and Manner stores I visited emphasized digital orders. In fact, when my wife and I ordered in person at the first Manner location, the staff seemed almost irritated, and at the second, we were flat out told it would be a “30-minute wait” for an in-person order but only about 5 minutes through their app.
These shops were typically smaller, with baristas focused on rapidly fulfilling digital orders. Incentives like discounts and collectible stickers are also common with app-based purchases
Luckin’s focus on efficiency is echoed in their 2023 SEC Form 20-F, stating that, “[w]e primarily operate two types of stores, namely pick-up stores and relax stores, for different purposes, and we strategically focus on pick-up stores, which accounted for 98.5 percent of our total self-operated stores as of December 31, 2023.” - Luckin Form 20-F
This approach may have helped Luckin scale rapidly, with a reported 18,360 stores in 2024, giving it a far larger footprint than other brands in the region.
By contrast, the two Starbucks locations we visited in Shanghai offered a more spacious, slower-paced experience, clearly targeting a different market segment. The baristas were attentive and anticipated in-store orders, and the atmosphere was more relaxed. For example, when my wife browsed for a mug (as she often does when visiting a city, even places she grew up), one of the baristas even came out to assist her.
What To Do As A Data Team
When a key driving metric shows a clear decline, it often overshadows other projects. For data professionals, this presents a valuable career opportunity: you want to be involved in the projects or initiatives that directly address these critical metrics. From a career standpoint, making a measurable impact on a business’s core metrics is invaluable experience.
If, for example, Facebook suddenly lost 5% of its monthly active users, everything else is window dressing, as data teams prioritize initiatives to address that core metric.
Now if you’re the data analytics team, you’d want to start by digging even further into the problem. To be clear, I am sure the Starbucks analytics team and I are in no way telling them how to do their job! But it’s kind of a fun quick exercise.
So here are a few questions I’d ask:
Are the declining transactions coming from specific groups or demographics?
Are promotions or loyalty programs effectively impacting comparable store sales?
Which product categories have seen the largest declines or gains? - Understanding what’s still selling well (or not) could help Starbucks decide where to innovate or reduce offerings.
How does digital ordering impact in-store transactions, and are there significant differences in spending behavior between digital and in-person orders? - Since competitors like Luckin Coffee emphasize digital orders, understanding whether Starbucks’ digital experience impacts transactions could highlight areas for improvement.
How does Starbucks’ brand perception compare to local competitors in terms of quality, convenience, and price value? - Knowing how the brand is viewed relative to local competitors could help refine marketing messages and operational strategies.
Are there specific days that Starbucks does an increased number of sales with more people sitting in the store (my theory here would be the only reason
I don’t know if they have any good examples of stores not next to some competitor, but if there were some, I’d just be curious to see if there was clearly an issue with substitution or not. Like if there is no other option, will a similar customer base pick Starbucks?
Taking it a step further, here are some projects that different data teams might take on during this initiative. I am sure some of these are already being done or are done at Starbucks HQ.
Data Engineering
Integrate Foot Traffic and Geo Data: Pull in location-based data on foot traffic, proximity to competitors, and surrounding amenities like malls for each store. This can help Starbucks understand which stores are more exposed to competitive pressures or where there may be untapped traffic.
Build Loyalty Program Data Pipelines: I assume they have this already, but if they don’t, then I’d create data flows that capture Starbucks Rewards usage, tracking how loyalty program engagement affects comparable store sales. By integrating this with transaction data, the team can see which customer behaviors drive higher sales.
Build an Automated Feedback Pipeline: Develop a robust data pipeline to collect and process customer feedback from various sources, such as social media, Starbucks’ app reviews, and online platforms. This pipeline would clean and prepare the data using natural language processing (NLP) techniques. Then, once processed, the feedback would be categorized by sentiment (positive, neutral, negative) and stored.
Data Analytics
Promotion and Upsell Effectiveness: Examine the impact of specific upsell campaigns (try a Frappuccino with your pastry) on transaction value and customer loyalty. This could help Starbucks refine its product bundling strategies to maximize the average ticket size per order.
Demographic and Behavioral Segmentation Analysis: Analyze customer data to segment users by demographic factors and visit behaviors. This helps Starbucks tailor offers or experiences to retain key groups and drive more frequent visits.
In-Store Traffic Pattern Analysis: Identify trends in how customers use stores in various regions, particularly around peak hours and weekends. For example, is there a particular time of day when people buy larger orders or prefer certain types of products? This could inform staffing or product promotion strategies.
Data Science
Predictive Modeling for Store Sales Decline: Develop a machine learning model that predicts which stores are most likely to experience a drop in transactions based on competitor density, economic indicators, and historical foot traffic. This helps Starbucks prioritize intervention in at-risk locations.
Personalized Promotion Model: Build a model to suggest targeted offers based on a customer’s past purchase behavior, encouraging higher spend or more frequent visits. For instance, suggesting a new seasonal drink to loyal coffee-only customers.
Sentiment Analysis on Product Feedback: Use natural language processing to analyze customer feedback on Starbucks’ products and experiences across regions. Identifying common complaints or positive sentiments around specific drinks or store experiences could guide product development and customer experience improvements. For example, one recent complaint I had when talking to someone was the lack of outlets some Starbucks has.
Of course, if data teams were to take on this level of ownership and were able to actually use their insights to increase this metric, I’d hope they’d be compensated for it.
What Metrics Drive Your Business?
For everyone out there who works in data and is wondering how to understand the business, you can always look at real companies, and try to understand them. What makes them tick, what actually drives sales, etc.
Start by picking a company, especially one that has either recently seen a large rise, like Chilis recently has, or one that has seen their key metrics drop such as Starbucks and start to ponder what is happening.
Think about what metrics are underlying the metrics you’re reading about and:
What questions would you want to answer if you were the C-suite to make better strategic decisions
What leverage company might have to impact this? Reduce price, increase marketing, change new store location strategies?
What could be the implications of those decisions? For example, if Starbucks approaches their Shanghai strategy in more of a turn and burn approach where they only focus on digital sales and reduce costs, will it impact perception of the brand? Is price the right area to compete on and does it actually increase comparable store sales.
This can help you move your data team from just being a task taking organization or cost center into a strategic partner.
Thanks for reading!
Join My Data Engineering And Data Science Discord
If you’re looking to talk more about data engineering, data science, breaking into your first job, and finding other like minded data specialists. Then you should join the Seattle Data Guy discord!
We are now well over 7000 members!
Join My Technical Consultants Community
If you’re a data consultant or considering becoming one then you should join the Technical Freelancer Community! We have over 700 members!
You’ll find plenty of free resources you can access to expedite your journey as a technical consultant as well as be able to talk to other consultants about questions you may have!
Articles Worth Reading
There are 20,000 new articles posted on Medium daily and that’s just Medium! I have spent a lot of time sifting through some of these articles as well as TechCrunch and companies tech blog and wanted to share some of my favorites!
Presto® Express: Speeding up Query Processing with Minimal Resources
Presto® is an open-source, distributed SQL query engine designed for running interactive analytic queries on data sources of any size, from gigabytes to petabytes.
At Uber, Presto is a critical engine for data analytics across various departments. The Operations team relies on it for dashboarding, while Uber Eats and marketing teams use its query results for pricing decisions. Presto is also essential to Uber’s compliance, growth marketing, and ad-hoc data analytics, making it a cornerstone of the company’s data-driven operations.
The Data Leader’s Guide to Hiring and Team Building - Part 1
Hiring the right talent for your data team is crucial to making your data work for the business. It’s not just about finding someone with the right skills—it’s about finding the right fit for your team’s needs. And that’s always easier said than done. There are plenty of challenges when it comes to identifying the right candidates, whether you’re searching for data engineers, analysts, or simply figuring out which type of role should be prioritized.
That’s why I view hiring as a pipeline that starts long before you write up the first job requisition, and certainly doesn't end once you’ve sent the confirmation email. You need to create an environment where new hires not only want to work but can also see a clear path for growth and success.
So, how do you go about hiring a data engineer, analyst, or scientist—and set them up to thrive? Let’s dive in.
End Of Day 150
Thanks for checking out our community. We put out 3-4 Newsletters a week discussing data, tech, and start-ups.