MBUS 853 — Session 4: Digital Organization

Block 1 — Session Theme: The Organization Is the Bottleneck

From "Can We Build It" to "Will Anyone Use It"

Sessions 1–3 established what digital strategy is, what leadership execution requires, and how fast an AI-driven advantage can erode. Session 4 asks the question underneath all three: even with a sound strategy, capable leadership, and a genuinely differentiated capability, what organizational structure and culture actually determine whether it gets adopted? Pernod Ricard's KDPs (D-STAR, Matrix) are technically sound, in-house built, and proven in pilots — the entire remaining question in the case is organizational: how to get 70+ largely autonomous affiliates across 160+ countries to actually use them.

Only 8% of firms engage in the core practices that support widespread AI adoption.

Fountaine, McCarthy & Saleh's headline statistic — technology and talent are rarely the constraint. Culture, structure, and incentives are.

Why Pernod Ricard Is a Sharper Test Than DBS or GE

DBS (Session 1) and GE (Session 2) were both relatively centralized organizations executing a top-down digital vision. Pernod Ricard is structurally different: a deliberately decentralized group, historically built that way to respect local market autonomy and legal variation across alcohol regulation in 160+ countries. That decentralization is a genuine strategic asset for a spirits company — and it is precisely what makes company-wide adoption of centrally-built digital tools structurally harder than at DBS or even GE.

Block 2 — Article: Building the AI-Powered Organization (Fountaine, McCarthy & Saleh, HBR 2019)

Technology Isn't the Biggest Challenge. Culture Is.

Based on surveys of thousands of executives, the authors find most companies' AI efforts stall not from a lack of pilots but from a failure to scale past them — because scaling requires three organizational shifts most companies never make.

Shift 1

Siloed work → interdisciplinary collaboration

AI has the biggest impact when cross-functional teams (business + operational + analytics expertise) work side by side, so initiatives address real organizational priorities rather than isolated technical problems.

Shift 2

Experience-based → data-driven decisions at the front line

Employees at every level must trust algorithmic recommendations enough to act without escalating to a superior first — which requires abandoning the traditional top-down approval chain.

Shift 3

Rigid and risk-averse → agile, experimental, adaptable

AI applications rarely launch fully baked. A test-and-learn mindset reframes early mistakes as discoveries rather than failures, letting small teams ship minimum viable products in weeks, not months.

Org Model

Hub, spoke, and gray area

A hub (central group under a C-level analytics leader) owns talent strategy, standards, and partnerships. Spokes (business units) own end-user adoption, workflow redesign, and incentives. A negotiated "gray area" in between owns project direction, data architecture, and change management.

Hub

Talent, standards, partnerships

Recruitment and training strategy, performance management, AI standards and policies — a central group headed by a C-level analytics executive.

Gray Area

Direction, delivery, architecture

Project direction, change management, data strategy and architecture — owned by hub, spokes, or shared with IT depending on firm maturity.

Spoke

Execution, adoption, incentives

End-user training, workflow redesign, incentive programs, performance tracking — the business unit closest to the people actually using the tool.

"Nearly 90% of companies with successful scaling practices spent more than half their analytics budgets on adoption activities."

— Fountaine, McCarthy & Saleh — technology spend is the smaller half of a successful AI program, not the larger one

10 Ways to Derail an AI Program — the Shortlist Most Relevant to Pernod Ricard

#7 — squandering time on enterprise-wide data cleaning instead of aligning data consolidation with the most valuable use cases first (echoes D-STAR's per-market data-availability constraints).
#6 — isolating analytics from the business rather than letting analytics and business experts work closely together (the exact problem BCG's early involvement, then the GDA's structure, was designed to solve).
#9 — neglecting to quantify bottom-line impact with a clear performance framework (Matrix's year-long TLO period before ROI was observable is a direct illustration of how long this can take even when done right).

Reinforcing the change: the article's four levers — walk the talk, make businesses accountable, provide incentives for change, track and facilitate adoption — map almost one-to-one onto specific tactics Pernod Ricard actually used: Platform Pioneer sessions with top management, product owners embedded in market teams, bonus incentives tied to D-STAR recommendation adoption, and the "control tower" that filtered out algorithmically-invalid recommendations before they reached sales reps.

Block 3 — Case: Pernod Ricard: Uncorking Digital Transformation

From Absinthe to AI

Founded in 1805 by Henri-Louis Pernod, Pernod Ricard grew via a 1975 merger (called "the equivalent of a merger between General Motors and Ford" by the press) and an aggressive acquisition strategy — Seagram's Chivas/Glenlivet/Martell (2001), Allied Domecq's Ballantine's/Malibu/Mumm (2005), and Vin & Sprit's Absolut (2008) took the group from #7 to #2 globally in spirits. By 2023: 90+ production sites, 70+ largely autonomous affiliates, operations in 160+ countries, ~19,500 employees, €10.7B in FY2022 net sales, and €55B market cap.

Why Digital, Why Now

Three converging pressures: (1) competitive catch-up — rival Diageo had already launched Catalyst, a marketing analytics tool, in 2017, and investors began directly questioning Pernod Ricard's pace of digitalization; (2) portfolio complexity — 13 brands acquired 2015–2018 alone made traditional, human-heuristic brand management increasingly unmanageable ("if you have six blockbuster brands, it's relatively easy for the human brain to handle... but if you have 13 blockbusters, 15 craft brands, plus additional strategic local brands, it becomes increasingly complex"); (3) channel underperformance — mature markets like France held leadership in off-trade (retail) but lagged in on-trade (bars/restaurants), pushing senior executives toward digitalization as a lever.

The Four Key Digital Programs (KDPs)

KDP	What It Does	Pilot Markets	Adoption at Scale
D-STAR	ML-driven sales visit recommendations — which outlets to visit, which products/promotions to push, based on 30–50 data sources per store	France, Germany, US (Florida), India (Nov 2020)	85% adoption rate at scale
Matrix	AI-optimized marketing budget allocation across brands and channels (TV, digital, social, in-store)	Germany, Japan (Oct 2020)	60–70% average adoption
Vista Rev-up	Simulates future scenarios to determine optimal promotions	Not detailed further in case	Not detailed further — case notes only that D-STAR and Matrix "had matured the fastest" of the four
Maestria	AI and data to predict consumer choices	Not detailed further in case	Not detailed further — case notes only that D-STAR and Matrix "had matured the fastest" of the four

Case-precision note: in April 2020 the executive committee prioritized six projects total — these four became the "core business KDPs"; two other, unnamed projects became separate "new business ventures" the case never revisits. Vista Rev-up and Maestria are core KDPs, not those ventures.

All four were built in-house rather than bought off-the-shelf — CDO Calloc'h's reasoning: off-the-shelf tools would need heavy recoding to match granularity requirements that vary market to market (a three-tier US market differs enormously from the UK), each KDP embeds ~50 sensitive business parameters the company didn't want to hand a third party, and qualified external vendors for this specific use case were scarce.

Why D-STAR Scaled Faster Than Matrix

D-STAR — Why It Worked

Sales reps are the "defined user" who executes and amplifies a predetermined strategy without needing to understand the algorithm's mechanics. Framed to reps as a competitive advantage (US market). Bonus incentives tied to following ≥60% of recommendations. Sales teams have more organizational stability than marketing.

Matrix — Why It Lagged

Marketing's self-image was "to create emotion" — Matrix demanded marketers also become analytical and structured, a bigger mindset shift. Recommendations sometimes meant cutting historically-favored brand investments, triggering emotional resistance from managers attached to specific brands.

Germany — A Recovery Story

Initially one of the most skeptical D-STAR markets (sales reps feared losing "commercial acumen"), Germany's adoption rose to 85% once the tool was properly embedded into existing sales software (it initially required working in a separate system — literally double the effort) and reluctant early adopters were deliberately given a voice in shaping it.

France — A Cautionary Note

Despite having richer retail data than Germany, some French managers resisted Matrix due to emotional attachment to specific brands and misalignment between the algorithm's recommendations and their own convictions — a reminder that better data alone doesn't guarantee faster adoption.

Block 4 — Scaling the KDPs: From BCG to an In-House Data Organization

Building the Global Digital Acceleration Team (GDA)

BCG got Pernod Ricard's digital transformation off the ground technically, but Jean-William Cousin's framing was blunt: "If you are following a partner, however strategic that partner is, you are not creating competitive advantage." Starting March 2021, Calloc'h and global data/analytics director David Lepicier internalized the capability — recruiting through "tech recruitment nights" (8–10 new hires in a few hours) and growing to a 130-person GDA team by 2022, with roughly half on freelance contracts to access highly specific, project-based skill profiles without long-term headcount commitments.

The Organizational Design Choice

The GDA adopted a network structure with a flatter management scheme — deliberately different from the traditional pyramid-shaped structure of Pernod Ricard's marketing and sales teams — with staff assigned to cross-functional squads for day-to-day work while remaining under the GDA umbrella. Lepicier's summary: "The reporting is not as interesting to me as how teams deliver. Lead data scientists are shaping programs and staff managers are shaping teams." This is close to a textbook version of Fountaine, McCarthy & Saleh's hub model — HQ held "the what" (business vision and product ownership), while market companies were "on the receiving end," dealing mainly with "the how."

Incentivizing Real Engagement, Not Just Compliance

As KDPs scaled, HQ shifted from a purely supportive posture to embedding KDP targets directly into affiliates' budgets and three-year strategic plans — e.g., if a project was expected to generate $1M in additional profit, HQ asked markets to commit at least 60% of that to their own targets, so overperformance became visible and rewarding rather than merely expected. This mirrors the article's "make businesses accountable" lever directly: ownership sits with the market, not the central digital team.

The still-open scaling tension: As of March 2023, Porta and Calloc'h face exactly the trade-off the article frames as a maturity question — should responsibility for a mature KDP (like D-STAR, which North America's transformation office expected to fully migrate into "day-to-day business" by FY25) move from the hub/gray-area to the spokes, and does that migration free up central capacity to launch new KDPs in new markets, or does premature decentralization risk losing the consistency the GDA was built to protect?

Block 5 — Memo Protocol: Team Case Study Memo #2

Format Reminder Before the Team Writes

Due Thursday 11:59pm before Session 4, based only on the Pernod Ricard case, two pages, 11-point font, written wholly by the team. This block is a structure check, not a draft.

To One decision maker — Pierre-Yves Calloc'h (chief digital officer) is the strongest single choice, since he owns KDP adoption and scalability directly; Christian Porta is a defensible alternate if the team's issues lean more toward overall digital-acceleration strategy than execution.

Issues Exactly 5, each a 3–5 word sub-heading + 20–30 words, each tied to a specific case fact (e.g., the 85% vs. 60–70% adoption gap, the France/Germany contrast, the Japan data-digitization delay).

Problem/Decision 40–60 words on the underlying cause behind most issues, ending in a decision framed as a question — the case's own closing tension (new markets vs. deepen existing ones) is a strong candidate framing, but the team should verify it's genuinely the root cause and not just the case's last line.

Alternatives Exactly 3, mutually exclusive, feasible, not simultaneous, not status quo.

Criteria Exactly 3 standards for judging the alternatives.

Evaluation/Recommendation 120–140 words, pros/cons per alternative per criterion, no table, ending in a justified pick.

Actions Exactly 3 steps not already taken in the case.

Case-only constraint: the Pernod Ricard case is unusually rich in named quotes from many executives (Porta, Calloc'h, Cousin, Lepicier, Coulon, Morogan, Genot, Nicolas, and more) — it's tempting to over-index on the most quotable lines rather than the underlying data (adoption percentages, timeline, sales-uplift figures). Ground every issue in a number or fact, not just a quote.

Academic integrity — GenAI is banned in submitted work for this course. The diagnostic analysis below is discussion prep, not memo text — the team's actual submission must be written independently, in the team's own words.

Block 6 — Case Diagnostic: Issues, Decision, Position (Discussion Prep)

Applying the Case Prep Protocol

Step 1 — Who and What

Decision maker: Pierre-Yves Calloc'h, chief digital officer. Core challenge: two of four KDPs (D-STAR, Matrix) are technically proven and generating measurable uplift (1.5–4.5% sales growth, up to 15% marketing efficiency) — the remaining problem is entirely about organizational adoption speed and consistency across a deliberately decentralized 70-affiliate structure, not further technology development.

Step 2 — Candidate Issues Grounded in Case Facts

Adoption gap between tools. D-STAR reached 85% adoption at scale versus Matrix's 60–70% — the same organization, same GDA support, materially different outcomes depending on which function (sales vs. marketing) owns the tool.
Emotional resistance in mature markets. France showed reluctance toward Matrix specifically due to managers' emotional attachment to brands they'd long managed, despite having richer underlying data than comparable markets like Germany.
Integration friction slows even willing markets. Germany's D-STAR skepticism was as much about the tool requiring double-effort (a separate system from existing sales software) as about substantive disagreement with its recommendations.
Data availability varies unpredictably by market. Japan's Matrix rollout was delayed weeks when weekly sales data assumed to not exist turned out to simply be undigitized — an avoidable delay that consumed pilot-phase confidence-building time.
Talent and continuity risk inside the GDA itself. With ~50% of the 130-person GDA on freelance contracts and marketing teams experiencing more turnover than sales teams, sustaining KDP momentum depends on a workforce structure the company deliberately built for flexibility rather than permanence.

Step 3 — A Position

Underlying problem, one sentence: Pernod Ricard solved the technology problem (both KDPs work, and work well) but the two tools sit on opposite sides of Fountaine, McCarthy & Saleh's adoption divide — D-STAR asks users to execute a predetermined strategy (low mindset change), while Matrix asks users to internalize new analytical judgment that can override felt expertise (high mindset change) — so uniform scaling tactics produce non-uniform results.

Counterargument to weigh: One could argue the deeper problem isn't the mindset-change gap between tools, but that HQ under-tailored its change-management approach to each market's specific culture and structure (management-led buy-in worked in France for D-STAR; individual persuasion of resistant staff worked in Germany) — meaning the fix is a more market-specific playbook, not a function-specific one. The strongest response has to decide whether the France/Germany contrast or the D-STAR/Matrix contrast is the more load-bearing pattern in the case's own evidence.

Second counterargument — the false binary: the case's own closing question doesn't just ask "new markets or existing ones" — it explicitly asks whether Pernod Ricard "could do this in parallel to reinforcing adoption where Matrix had already been launched." Treating expand-vs-deepen as an either/or may be answering a question the case itself frames as open-ended. The real constraint may not be strategic focus at all but GDA capacity — a 130-person team, roughly half on freelance contracts, already stretched across four KDPs and multiple new-market rollouts (UK, China, South Africa, Canada for D-STAR alone). On this reading, the sharper diagnostic question isn't "which priority" but "does the GDA have the bandwidth to do both, and if not, which one does its talent-continuity risk (Issue 5) force it to sacrifice."

Step 4 — 30-Second Cold-Call Answer

Pernod Ricard didn't fail to build good AI tools — D-STAR and Matrix both work, and both were built in-house on purpose to create real competitive advantage rather than rent it from a vendor. What's still unsolved is that the two tools sit on opposite sides of the adoption divide: D-STAR asks sales reps to execute a strategy someone else designed, so it hit 85% adoption, while Matrix asks marketers — whose whole professional identity is "we create emotion" — to subordinate brand intuition to a response curve, so it's stuck at 60-70% even in markets like France with richer data than Germany. Scaling the same change-management playbook uniformly across both functions was always going to produce uneven results, and that's the choice Calloc'h and Porta actually have to make now: fix Matrix's adoption gap before expanding into new markets, or risk repeating that same gap at greater scale.

Block 7 — Discussion Questions & Sharp Answers

Likely Professor Questions

Framing to expect: (1) Why did D-STAR scale faster than Matrix? (2) Was building the KDPs in-house the right call? (3) Should Pernod Ricard prioritize new markets or deeper scaling in existing ones?

Q1: Was building all four KDPs in-house, rather than buying off-the-shelf tools, the right organizational choice?

Yes, on the case's own terms — Cousin's line is the clearest evidence: "if you are following a partner, however strategic, you are not creating competitive advantage." An off-the-shelf tool could have launched faster, but Pernod Ricard's core problem (60–160 country-specific regulatory and market variation, ~50 sensitive business parameters per KDP) meant heavy customization was inevitable either way — building in-house converted that customization cost into owned, compounding capability (the 130-person GDA) rather than recurring vendor dependency.

Redamo Labs made the same build-vs-buy call for identity verification — vendor tools existed, but owning the capability in-house was what let the team iterate the flow that cut verification time from 8 to 2 minutes.

Q2: Why did the same GDA team, using similar change-management tactics, get such different adoption results for D-STAR versus Matrix?

The tools ask fundamentally different things of their users. D-STAR's sales reps stay in an execution role — the algorithm decides, they act, and the case explicitly notes they don't need to understand the mechanics. Matrix asks marketers to change how they think, not just what they do — trusting a "response curve" over years of brand intuition, in a function whose stated purpose ("to create emotion") is philosophically at odds with the tool's premise. Fountaine, McCarthy, and Saleh's Shift 2 (experience-based to data-driven decision-making) is much harder to complete when the profession's self-concept is built around the very intuition being replaced.

At Prodigy Education, A/B testing tools were easier to get engineering teams to adopt (execution-role fit) than to get some brand/marketing stakeholders to fully trust over creative instinct — a smaller-scale version of the same D-STAR/Matrix divide.

Q3: Should Pernod Ricard expand the KDPs into new markets, or focus on deepening adoption where they've already launched?

The case's own data favors depth over breadth in the near term: Matrix adoption is still only 60–70% in markets where it has already launched, well below D-STAR's 85% — meaning there's more unrealized value sitting in existing markets than in unlaunched ones. Expanding into new markets before closing that gap risks repeating Matrix's slower adoption pattern at greater scale, rather than fixing the underlying mindset-change problem the case has already surfaced.

This is the strongest question to lead class discussion with — it's directly answerable from case data (the adoption percentages) rather than opinion, which is exactly the kind of evidence-grounded position this course rewards.

Block 8 — Participation Hooks & Taju's Edge

How to Contribute Distinctively

Open Strong

Don't open with "Pernod Ricard's tools work well." Open with the divide: the exact same organization, using the same team and largely the same change-management playbook, produced an 85% adoption tool and a 60–70% adoption tool — the gap is the whole case.

Push the Consensus

Class will likely say "marketing resisted more than sales." Push further: it's not resistance to data per se — France had richer data than Germany and still resisted more — it's resistance to a tool that overrides professional identity, not just habit.

Bridge to the Article

Map the GDA's hub/gray-area/spoke evolution directly onto Fountaine, McCarthy & Saleh's model — Pernod Ricard is a rare case where a company visibly moved from an external-partner hub (BCG) to an internal one (GDA), which the article frames as the harder and more durable path.

Taju's Edge — Redamo Labs

Leading a 12-person cross-functional team through an IAM overhaul required exactly the kind of incentive-and-framing work D-STAR used successfully — positioning the change as giving people a competitive edge, not just a new process to follow.

Taju's Edge — Prodigy Education

Cross-functional growth teams across three time zones show how uneven adoption speed across functions is normal and predictable — the fix is diagnosing why a specific function resists, not assuming a uniform rollout plan will work everywhere.

Taju's Edge — Stutern

Scaling Stutern's B2B SaaS platform to 2,500+ business partnerships required constant negotiation between what HQ wanted standardized and what individual partners needed customized — the same hub-vs-spoke tension Pernod Ricard is navigating at far larger scale.

Block 9 — Reflections Journal Prep (Fill In After Class)

Concept Must Come From Live Slides

As with Sessions 2–3, the journal's concept half needs to be captured live from Prof. Mufti's Session 4 slide deck, not guessed at here. The example half is ready to pair with whatever concept the session covers.

Concept — Fill In From Session 4 Slides

[3–7 word identification — write after class]

150–200 words, from class slides only. Organization-focused candidates to listen for: hub-and-spoke operating models, change management frameworks, organizational ambidexterity, or talent/incentive design for digital adoption — write down whichever the professor actually presents.

Candidate Example — Ready to Pair With Whatever Concept Fits

Redamo Labs — Different Teams, Different Adoption Curves for the Same Change

Rolling out the same new analytics dashboard to two teams inside the IAM initiative produced strikingly different adoption speeds. The operations team, whose role was largely to execute against clear service-level targets, adopted the dashboard within days — it made an existing job easier without asking them to reconsider how they made decisions. The risk-review team, whose work depended heavily on individual judgment built over years, was slower and more skeptical, since the dashboard's recommendations sometimes contradicted long-held intuitions about which accounts warranted scrutiny. The lesson wasn't that one team was more resistant to change in general — it was that tools asking people to execute differently adopt faster than tools asking people to think differently, and change management has to be designed differently for each.

Block 10 — Key Takeaways

What to Walk Away Knowing

Technology maturity and adoption maturity are different clocks. Both D-STAR and Matrix were technically ready at the same time; only one reached high adoption on the same timeline.

Tools that ask people to execute differently spread faster than tools that ask people to think differently. The D-STAR/Matrix divide is a close-to-perfect natural experiment on this exact distinction.

Organizational structure has to evolve as capability matures. Pernod Ricard's shift from BCG-led to internally-owned (the GDA), and its planned shift of mature KDPs from the transformation office into day-to-day business, both show structure following maturity rather than being fixed upfront.

Incentive design determines whether adoption becomes genuine or superficial. Tying bonuses to D-STAR recommendation follow-through, and later tying KDP targets into affiliates' own budgets, converted passive compliance into active ownership.

Looking Ahead — Session 5: Digital Architecture

→ Session 5 (Architecture)

Session 4 asks "what organizational structure enables adoption?" Session 5 (Stop Tinkering with AI; Harley-Davidson case) asks "what technical and data architecture decisions are foundational versus tactical, once the organization is ready to scale?"

↔ Recurring Thread: Build vs. Buy

Pernod Ricard's in-house build decision (vs. off-the-shelf) recurs as a live architecture question in Session 5 — how much of the technical stack a company should own versus source externally as it scales.

Memo #3 Due Before Session 5

The team's third Case Study Memo (Harley-Davidson) is due the Thursday before Session 5 at 11:59pm.