Skip to content

Module 2. Understanding Data

Executive Learning · Free learning

Understanding Data

The raw material of every model, and how to reason about its structure and value.

Data is the evidence base for everything that follows. This module builds a precise vocabulary for describing it and judging whether it can be trusted.

In this module

0 of 6 activities complete

0%

Reading

0% read

Structured, semi-structured and unstructured

Structured data lives in rows and columns with a fixed schema, such as a transactions table. Semi-structured data carries its own labels, such as JSON event logs. Unstructured data has no schema at all: documents, calls, images.

Roughly eighty percent of enterprise data is unstructured, which is precisely why generative models have unlocked so much stranded value.

Data as an asset with a balance sheet

Treat every dataset as an asset with an owner, a cost of maintenance, a decay rate and a value in use. Datasets nobody owns decay fastest and cause the most expensive failures.

Real-world case studies

Case study · Financial services

A retail bank fixes a definition before it fixes a model

  1. Situation

    A retail bank ran four dashboards reporting four different numbers for active customers, and executive meetings routinely stalled on which figure was correct.

  2. Challenge

    Each business line had defined activity differently, from any login in ninety days to any transaction in twelve months. No model built on the underlying data could be trusted while the definition itself was contested.

  3. How data was used

    Account, transaction and channel logs were mapped to a single documented definition with an owner, and historical figures were restated on that basis.

  4. How AI was applied

    Only once the definition was stable was a churn model trained, using consistent activity features across all lines.

  5. Business outcome

    Reporting disputes ended, and the churn model reached usable accuracy at the first attempt because the target variable finally meant one thing.

Lesson for the learner

Definition precedes analysis. A single agreed definition is often worth more than a sophisticated model.

Understanding Data, executive briefing

16:05 · Video briefing

Key takeaways

  • Structure determines which tools and models are possible
  • Most enterprise data is unstructured and historically underused
  • Every dataset needs a named owner and a defined refresh cycle

Inventory your five most important datasets

Build the first page of a data asset register.

  1. Name the five datasets your decisions depend on most
  2. Record the owner, refresh frequency and structure type for each
  3. Flag any dataset with no clear owner

Module 2 quiz

2 questions · 70% to pass

  1. 01Customer support call recordings are an example of which data type?

  2. 02A critical report breaks every quarter and nobody can say why. What is the most likely root cause?Applied

Answer every question to submit.

Your notes

Lesson discussion

Ask questions, share how you are applying this in your organisation, and learn from other executives.

Sign in to join the discussion. Reading is open to everyone.

Loading discussion…

Previous