Week 1 · 14 April 2026
DTNext × Digital Bricks

AI Strategy
Week 1 — Playing to Win

A 3-hour Strategic Choice Cascade for DTNext as an AI-era Microsoft partner. We leave with a filled Cascade, a signed one-pager, and an experiment backlog for Weeks 2–6.

Facilitator
Michiel Heynen
Co-host
Max Dinser
Duration
3 hours
Lens
External / competitive
Opening reframe

This is not a
Copilot kick-off.

It is a strategy session. Before we adopt AI, we decide what winning looks like for DTNext.

"Strategy is an integrated set of choices that uniquely positions the firm in its industry so as to create sustainable advantage and superior value relative to the competition." — A.G. Lafley & Roger Martin, Playing to Win
AI is a capability. It shows up at Question 4 of the Cascade — not Question 1.
The 3 hours

How we spend today — and what you leave with.

  • 0:00–0:10Opening reframe · the game being played
  • 0:10–0:25Strategy Traps · diagnose the biggest risk
  • 0:25–0:55Q1 Winning Aspiration · benchmark & metric
  • 0:55–1:25Q2 Where to Play · including where not to
  • 1:25–1:35Break
  • 1:35–1:55Q3 How to Win · 15 words, one sentence
  • 1:55–2:15Q4 Capabilities · required by Q3 only
  • 2:15–2:30Q5 Management Systems · metric · cadence · owner
  • 2:30–2:45"What would have to be true?" · experiment backlog
  • 2:45–3:00One-pager & close · verbal sign-off

What you leave with

  1. A filled Strategic Choice Cascade — five questions, one page.
  2. A DTNext AI Strategy one-pager, signed off verbally in the room.
  3. An assumption backlog that becomes the experiment list for Weeks 2–6.
Before we start choosing

Five ways strategies fail.

Playing to Win identifies five recurring traps. Let's name the one DTNext is closest to — anonymously — before we pick up a sticky.

1

Do-It-All

"We serve every segment." Refusing to choose. Result: no advantage anywhere.

2

Don Quixote

Attacking the strongest competitor in their strongest segment. Heroic and fatal.

3

Waterloo

Multiple wars on multiple fronts simultaneously. Capabilities spread too thin.

4

Something-for-Everyone

One service line for all customers. Commoditization follows.

5

Dreams-That-Never-Come-True

Ambition with zero capability to execute. "We'll be the AI leader in…"

Q1 · Winning Aspiration

What does winning look like for DTNext in 24 months?

Lafley rule: Too modest aspirations are more dangerous than too bold ones. Modesty guarantees mediocrity.
Worked example · P&G Olay
"Beat Lancôme and L'Oréal in the masstige skincare category." — Specific opponents. Specific category. Testable every quarter.
Q2 · Where to Play

Which segments? Which AI services?
And — critically — where not?

Mandatory test: Write down 3 places DTNext will NOT play. If you can't, you haven't made the choice. — Lafley
Worked example · P&G Olay
Explicitly excluded the mass market (Oil of Olay's old home) and prestige department stores (L'Oréal's home). Only masstige. That single exclusion drove every choice that followed.
Q3 · How to Win

Cost leadership or differentiation.
One sentence. 15 words max.

Martin rule: This is where most strategies collapse into vagueness. Force the answer into 15 words or send it back.
Worked example · Bounty paper towel
"Trap and lock away more liquid per sheet than any competitor, and prove it on-pack." — 15 words. Testable. Defensible for 20 years.
Q4 · Capabilities

What 3–6 capabilities are required by our How-to-Win?

This is where Copilot adoption, agent engineering, AI literacy and governance land — as consequences of the How-to-Win, not as the topic of the session. The rest of the 6-week programme is the build-out of Q4.
Q5 · Management Systems

Metric · Cadence · Owner.
Where most strategies die.

Lafley rhythm: Every business unit reviewed its one-page Cascade with him twice a year. The review was the management system.
The reverse test

"What would have to be true
for this strategy to work?"

Rather than asking "what should we do?", we list every assumption the Cascade depends on — and split them into two piles.

We know

Assumptions we have evidence for today. Move on.

We'd need to test

Assumptions we're betting on without proof. This pile becomes the experiment backlog for Weeks 2–6. Each assumption gets an owner, a test, and a date.

Why this matters: it turns a strategy session into an operating plan — and gives the rest of the programme a strategic spine.
The deliverable

One page. Five boxes. One signature.

At the close of the session, the Cascade gets written up live on screen. Constant reads it aloud. Verbal sign-off in the room.

Q1
Winning Aspiration
Q2
Where to Play
+ NOT play
Q3
How to Win
(15 words)
Q4
Capabilities
HAVE / GAP / MISSING
Q5
Management
Systems
⚑ Assumption backlog — "what would have to be true?" — feeds Weeks 2–6
Lafley: If it doesn't fit on one page, it isn't a strategy.
What happens next

From choice to experiment in 48 hours.

In the room

Signed Cascade

Constant reads the one-pager aloud. Verbal sign-off from the executive team. Photo of the wall.

Within 48 hours

Clean one-pager

Digital Bricks circulates the typed one-pager to all attendees. Any dissent surfaces here — before Week 2.

Week 2 and beyond

Experiment backlog

The "we'd need to test" pile opens every session. Each Copilot, agent and governance workstream proves or disproves an assumption.

The Cascade keeps working: it becomes the parent cascade for every delivery team. A mediocre strategy reviewed monthly beats a brilliant one reviewed annually.
01 / 12