The required Coca-Cola case in full — five applications, the Digital Academy, the $463bn headline — then a hard look at the evidence underneath it.
July 2026 · Case 2 of 6
As you read — hold this question
If a case study’s proof point predates the technology it’s meant to prove, is it still evidence — and how would you know?
$463bn
the McKinsey number the case leans on — and it doesn’t say what the case implies it says.
The case presents Coca-Cola as a tidy, uniformly positive AI success. Taught straight, it is five applications, a 500-manager Digital Academy, three named challenges, and a $463bn McKinsey headline. Audited, its central evidence traces to a July 2022 article about manufacturing — four months before ChatGPT — and its flagship has never published a result.
The set case, taught straight
The textbook versionQuiz: Set case study
Coca-Cola’s AI-powered marketing transformation, in five moves
Application
What the case claims
1. Personalisation
Freestyle machines and weather-based vending in Japan
2. Generative content
“Create Real Magic” (2023) — OpenAI + Bain; fans make artwork with DALL·E & ChatGPT
3. Predictive analytics
Forecasting which campaigns will land
4. Social listening
Real-time sentiment during events like the World Cup
5. Product innovation
Coca-Cola Y3000 — a flavour it says was co-created with AI
Plus a Digital Academy (500+ managers trained, 20+ initiatives), the McKinsey headline that gen-AI could be worth ~$463bn a year, and three named challenges: data privacy (GDPR), bias & fairness, and brand authenticity.
Audit the evidence
The other side
The single sharpest thing you can say about this case
The impressive “500 managers, 20+ initiatives” proof traces to one McKinsey article from July 2022 — four months before ChatGPT launched — and it is about manufacturing (“ten-plus sites in the manufacturing network”), not marketing. The word “AI,” in the generative sense, is not even in the source. The figures are real — quote them — but the leap from a factory-floor programme to proof of an AI-marketing transformation is the weak link.
Read the $463bn carefully. McKinsey does not say productivity rises 5–15%; it says gen-AI could add value worth 5–15% of total marketing spend — modelled potential, not a measured result.
A measurement vacuum. “Create Real Magic” made headlines in 2023, but no participation, sales or brand-lift figures have ever been published.
Where it actually went. By 2026 Coca-Cola’s durable AI win was in finance — an in-house tool (“Fuel Light 360”) that cut a marketing-spend decision cycle “from two weeks to one hour” (per Bain / CFO Dive). The value landed in decision speed, not AI-generated ads.
Answering the 7.5 guiding questions
Two are traps. Q2 (AR / metaverse) has no basis in the case at all — you must draw on the required reading’s Theme 5 (immersive marketing). Q4 (brand-equity measurement) is the case’s weakest ground: its Results section claims benefits with no figures, baselines or time period, so the honest answer is that the case does not actually measure it.
Take this away
A tidy case can still rest on undated evidence and an unproven flagship. Learn the five applications and the challenges for recall — then read the footnotes for the discussion.
Quick recall — without looking back
Test yourself on this case
Question 1 of 3
Name the five AI applications in the Coca-Cola case and the three challenges it lists.
Applications: (1) personalisation (Freestyle, Japan weather vending); (2) generative content (‘Create Real Magic’, 2023, OpenAI+Bain); (3) predictive analytics; (4) real-time social listening; (5) product innovation (Y3000). Challenges: data privacy (GDPR), bias & fairness, and brand authenticity.
Question 2 of 3
What is wrong with the case’s ‘500 managers’ evidence and its $463bn figure?
The ‘500 managers / 20+ initiatives’ proof traces to a July 2022 McKinsey article — before ChatGPT — about manufacturing, not marketing. And McKinsey’s $463bn is value worth 5–15% of marketing spend as modelled potential, not a measured productivity rise.
Question 3 of 3
Which two of the 7.5 guiding questions can’t be answered from the case, and why?
Q2 (AR/metaverse) — the case says nothing about it, so you answer from the reading’s Theme 5. Q4 (brand-equity measurement) — the case claims benefits with no figures, baselines or time period, so it does not actually measure brand equity.
BUSN9049 Module 7 — Coca-Cola’s AI-powered marketing transformation (case PDF) and the 7.5 activity page. Flinders University, 2026.
Evidence audit
McKinsey, “Coca-Cola: the people-first story of a digital transformation” (25 Jul 2022); McKinsey gen-AI value estimate (Dec 2023); CFO Dive coverage of Coca-Cola’s “Fuel Light 360” (2026).