The AIDA Manifesto

The AIDA Case Study

Most “AI success stories” are marketing.
AIDA case studies are evidence. Use this template to document impact a CEO can trust, a CIO can replicate, and a team can learn from.

CEO / Board Executive Card
CIO / CDO Full Case Study
Product Teams Sprint Narrative
Ground Rules

Case Study Non-Negotiables

If it doesn’t meet these five standards, it’s not an AIDA case study—it’s a press release.

№
Numbers, not adjectives.
“3.2% reduction in churn”—not “significant improvement.”
◎
Baseline first.
If you don’t know “before,” you don’t know “impact.”
⏱
Time to value.
Days and weeks, not quarters and years.
🔒
Ownability.
If the vendor leaves, can you sustain it?
⚑
Honesty.
Include what you killed, what failed, and what you would not repeat.
The One-Pager

Executive Card

Copy/paste into a board deck. If you can’t fit it on one page, do more work on it until it fits.

Use Case Summary
Keep this to one page. Every field is mandatory.
Use Case Title
[Short, specific]
Primary User
[Role / function]
Business Problem
[What decision or outcome was broken?] — one sentence
Baseline
[Metric + current value + time period]
Hypothesis
We believe that [intervention] will [improve metric] because [mechanism]
Discovery Sprint (4 Weeks)
Week 1
[Problem deep-dive]
Week 2
[Data reality check]
Week 3
[Off-the-shelf experiments]
Week 4
[Proof of Value]
Evidence Ladder Reached
0Opinion
1Quantified
2Experiment
3Production Value
4Scaled Impact
Solution (Minimum Viable AI/Data)
[What you actually shipped]
Time to First Measurable Value
[X days / weeks]
Cost to Prove Value
[Internal effort + vendor spend (if any)]
Impact
Metric #1
[before → after]
Metric #2
[before → after]
Financial Proxy
[CHF / € / $]
Sustainability / Owner
Product owner: [Name] • Team: [Roles] • Run model: [Who monitors, retrains, improves]
What We Did NOT Build
[Bespoke platform, custom integration, etc.]
Next Scaling Step
[What happens next, and what must be true]
The Full Picture

Full Case Study

Recommended length: 2–4 pages. Nine sections. Every one earns its place.

1

Context — The “Why Now”

  • What changed? Market, regulation, customer behavior, cost pressure, competition
  • Why now and not later? What made inaction unacceptable?
2

Problem Definition

  • Broken decision / workflow: What specifically was failing?
  • Who felt the pain? Name the user group
  • Cost of inaction: What happened every month this went unsolved?
Anti-Pattern

“We needed AI.” — AIDA starts with: “We needed a better outcome.”

3

Baseline & Success Metrics

  • Baseline metric(s): What was measured, how, and by whom?
  • Success threshold: What would count as “proved value”?
  • Measurement cadence: Weekly / monthly — and who owned it?
4

Discovery Sprint Narrative

Week 1 — Problem Deep-Dive

  • Workflow map — where decisions happen
  • Root cause assumptions — what you believed before seeing data
  • Stakeholder alignment — what was explicitly out of scope

Week 2 — Data Reality Check

  • Data sources considered (system + owner)
  • Availability: what existed vs. what was imagined
  • Quality findings (top 3 issues)
Continue Kill Reframe

Week 3 — Off-the-Shelf Experiments

  • Tools used (off-the-shelf only)
  • Experiments run (top 3): hypothesis → result
  • Non-AI comparator: rules, heuristics, process change
Rule: If you can’t beat a simple baseline, stop.

Week 4 — Proof of Value

  • Test design: what was evaluated, on what sample, with what controls
  • User feedback: what changed in behavior, not just opinions
  • Measured outcome vs. baseline
  • Go / no-go decision and rationale
5

What Was Shipped (and What Was Not)

  • MVP description: The minimum that keeps the value alive
  • Integration points: Only what was required for adoption
  • Monitoring & ops: Drift, data quality, failure modes
  • What you avoided building: And why that was the right call
  • What you deferred: Features reserved for post-validation
6

Ownership & Capability

  • Product roadmap owner: Who decides what comes next?
  • Data quality owner: Who ensures the inputs stay reliable?
  • Drift / performance monitor: Who watches, who acts?
  • Internal skills required: What the team had to learn
  • Vendor role: Coach, specialist, temporary capacity — not owner
7

Impact — Show Your Work

  • Outcome metrics: Before → after (with time period)
  • Financial translation: Revenue impact, cost reduction, hours saved, risk reduced
  • Leading vs. lagging: What predicts future value vs. what confirms past value?
  • Confidence level: High / Medium / Low — and why
8

Lessons — Make It Reusable

3 Decisions That Mattered
  • [Decision 1]
  • [Decision 2]
  • [Decision 3]
3 Things Never Again
  • [Mistake 1]
  • [Mistake 2]
  • [Mistake 3]
1 Capability Built
  • [The one thing that makes the next use case faster]
9

Scaling Plan — How Impact Compounds

  • Prerequisites to scale: Data, adoption, governance, platform
  • What is reusable? Data assets, features, prompts, pipelines, dashboards, policy
  • Next adjacent problem: Where does this value chain lead?
Appendix

Supporting Artifacts

Optional but valuable.

AIDA case studies are not trophies.

They are operating instructions. If your case study cannot be used to reproduce the outcome in another domain, it is not an AIDA case study yet.