Episode summary

In Episode 12 of '100 Days of Data,' Jonas and Amy dive into the world of diagnostic analytics — the step beyond simply knowing what happened to understanding why it happened. They explore how businesses use tools like root cause analysis, variance analysis, and data mining to uncover underlying drivers behind data shifts. Through real-world examples in finance, healthcare, manufacturing, and retail, the hosts explain how diagnostic analytics helps identify problems at their source, rather than just treating symptoms. The episode also highlights the importance of context, data quality, and domain expertise in making accurate diagnoses. Listeners will gain an understanding of when and how to apply diagnostic techniques to make more insightful, effective decisions based on their data.

Episode video

Episode transcript

JONAS: Welcome to Episode 12 of 100 Days of Data. I'm Jonas, an AI professor here to explore the foundations of data in AI with you.
AMY: And I, Amy, an AI consultant, excited to bring these concepts to life with stories and practical insights. Glad you're joining us.
JONAS: Knowing why something happened matters even more than just knowing what happened.
AMY: That’s what diagnostic analytics is all about — drilling down from “What’s going on?” to “Why is this happening?” So today, we’re unpacking diagnostic analytics, why it’s crucial, and how it plays out in the real world.
JONAS: Let’s start with the basics. Diagnostic analytics tries to explain the root causes behind observed events or changes in data. It’s the step after descriptive analytics, which tells us what happened. Diagnostic analytics tells us why it happened.
AMY: Right, if descriptive analytics is like glancing at your car’s dashboard to see a warning light, diagnostic analytics is opening the hood and figuring out why that light came on.
JONAS: Great analogy! At its heart, this involves investigating drivers or factors that influence an outcome. Think of drivers as variables that push or pull the behavior of whatever we measure. For example, sales going down — what factors caused that decline?
AMY: In practice, companies often use diagnostic analytics to avoid costly guesswork. Instead of blindly throwing marketing dollars at a problem, they ask, “What changed in the last quarter that impacted sales?”
JONAS: A common tool here is variance analysis, where you measure changes between two periods and break down the difference into components. It’s like peeling an onion — layer by layer — to uncover contributors to change.
AMY: I saw this first-hand working with a retail chain. Their sales dropped, and the initial thought was that customers were just shopping less. But through variance analysis, we found that certain high-margin products didn’t sell well because of supply disruptions. That wasn’t obvious without digging into the numbers.
JONAS: Exactly. That’s the power of diagnostic analytics: moving from surface-level observations to causal insights. However, it’s important to remember that diagnostic analytics often relies on correlation and historical data. It doesn’t always prove causation beyond doubt.
AMY: That’s a key point. I’ve seen teams jump to conclusions because they found correlations and assumed cause-and-effect. In one finance client case, a spike in loan defaults correlated with an economic event, but it ignored internal process changes that were equally responsible.
JONAS: To better assess causes, diagnostic analytics typically involves techniques like drill-downs, data mining, and even using data visualization to spot patterns. For example, breaking data into segments by region, customer type, or time can reveal hidden insights.
AMY: Visualization is often the gateway for business leaders. I remember a healthcare provider client who used dashboards showing patient wait times increasing. By drilling down, they identified that a new scheduling process in one clinic was the driver, leading to targeted fixes.
JONAS: Besides identifying drivers, diagnostic analytics also helps prioritize issues. Not every factor impacts the outcome equally, so understanding which drivers have the biggest variance contribution is critical.
AMY: In automotive manufacturing, for instance, a company noticed a spike in warranty claims. Using diagnostic analytics, they segmented the data by car model, assembly line, and supplier. They discovered one supplier’s components caused a disproportionate share of defects, which they addressed head-on.
JONAS: That’s a classic example of root cause analysis, a formalized approach within diagnostic analytics. Root cause analysis seeks the fundamental reason behind a problem, not just symptoms.
AMY: And in business, focusing on root causes saves resources. Fixing symptoms is like putting a band-aid on a bullet wound — it might help briefly but won’t solve the problem.
JONAS: Historically, diagnostic analytics emerged alongside improvements in computing power and data storage. Early business intelligence focused mostly on descriptive analytics because data was scarce and tools were limited.
AMY: Fast forward to today, companies have mountains of data from transactions, sensors, social media — that’s why diagnostic analytics now can uncover patterns and drivers with remarkable granularity.
JONAS: Another important feature is the integration of domain knowledge with data analysis. Pure statistical patterns may mislead if you don’t understand the context.
AMY: Agreed. In finance, for example, regulatory changes might affect data trends temporarily. Without knowing that context, diagnostic analytics might misattribute causes.
JONAS: In terms of frameworks, diagnostic analytics often fits into a broader analytics maturity model: descriptive, diagnostic, predictive, and prescriptive analytics.
AMY: We’re covering predictive analytics in the next episode, but for now, understanding diagnostic analytics is like getting your data detective skills sharpened — it helps you ask the right “Why?” questions before jumping to what might happen next.
JONAS: Let’s look at some common techniques used in diagnostic analytics: root cause analysis, variance analysis, drill-down, and data mining. Root cause analysis is systematic and often uses tools such as the ‘5 Whys’ or fishbone diagrams.
AMY: The ‘5 Whys’ is a favorite in consulting. You keep asking “Why?” five times or more to get to the core reason. I used it once with a logistics company to find why deliveries were late; it turned out to be a faulty scheduling system, not driver issues as originally assumed.
JONAS: Variance analysis, in a nutshell, measures the difference between actual data and expected or previous data and allocates that variance to different causes.
AMY: And drill-down lets you slice the data by timeframes, demographics, geographies, whatever dimensions give you insight. For example, sales may be down overall, but only in a specific region — revealing local market challenges.
JONAS: Data mining, meanwhile, uses algorithms to detect patterns, clusters, or associations that aren’t obvious with simple observation. Often, it’s iterative and exploratory.
AMY: In retail, we’ve used data mining to discover customer segments driving changes in buying behavior, which led to tailored marketing campaigns — a nice blend of diagnostic and prescriptive action.
JONAS: Now, while diagnostic analytics is powerful, it has limitations. It depends heavily on data quality and assumes past data trends explain present issues.
AMY: Spot on — “garbage in, garbage out.” If your data is incomplete or delayed, your diagnostics may be off, leading to bad decisions.
JONAS: Another limitation is complexity. With so many potential drivers, it can be challenging to isolate the exact root cause, especially with interconnected systems.
AMY: That’s when expert input becomes crucial. Data tells us patterns, people tell us what makes sense operationally.
JONAS: To summarize, diagnostic analytics moves beyond what happened to uncover why. It leverages tools like variance analysis and root cause analysis, helping businesses make informed decisions.
AMY: And in my experience, combining data insights with practical business knowledge is what turns diagnostics into real-world impact — whether reducing defect rates, improving customer experience, or cutting costs.
JONAS: Key takeaway from me: Diagnostic analytics is a critical bridge between observing an outcome and understanding its causes. It allows businesses to focus on what truly matters.
AMY: Mine’s simple: Don’t just stop at “What happened?” Keep digging with diagnostic analytics — that’s where you find your best opportunities to improve.
JONAS: Next episode, we’ll look at predictive analytics — how data can not only explain the past but forecast what’s coming next.
AMY: If you're enjoying this, please like or rate us five stars in your podcast app. We’d love to hear your questions or comments — you might even be featured in an upcoming episode!
AMY: Until tomorrow — stay curious, stay data-driven.

Next up

Next episode, discover how predictive analytics can help you forecast future outcomes and trends from your data.