EBEDWIN BLANCHARD / CMO
Marketing AI, done properly

From AI strategy to execution.

Building rigorous approaches to deploying AI, from the strategic rationale and practical use cases to scaled change and adoption across the enterprise, and building a narrative for change.

The Methodology

AI marketing transformation: eight steps from vision to adoption

I follow this stepwise approach to embedding AI within organisations.

Tap or swipe through the steps to find out more.

Annual planning and adoption reviews restart at stage one

Step 01

Marketing vision

What should marketing become, and where does AI create value for customers?

What happens

  • Build clarity on the value and role of marketing
  • Clarify the marketing archetypes and the customer promise for the future
  • Answer three linked questions: the business problems, the marketing problems and how AI sharpens the customer promise
  • Turn the vision into a destiny story the team can repeat

We build

  • A marketing vision the leadership team agrees
  • A clear view of the business challenges marketing will overcome
  • A short internal story about where marketing is heading

We build a destiny story

  1. The world is changing
  2. The old way is breaking
  3. AI opens a new possibility
  4. People stay central
  5. This is our destiny
Step 02

Readiness

Can the organisation and its data carry the plan?

What happens

  • Run a baseline pulse survey on how people feel about AI
  • Assess skills gaps by role and team
  • Assess the data: sources, quality and access
  • Test whether technology is the right fix. Process and data often come first

We build

  • A readiness view by team and role
  • A data gap list with owners
  • A clear list of what to fix before buying anything
76% vs 31%

A BCG survey found 76% of executives believed employees were excited about AI. Only 31% of employees said they were. Measure the real number before you design anything.

Step 03

Workflows

Which workflows should change first, and what do they look like afterwards?

What happens

  • Find the friction in today’s workflows with the people who run them
  • Co-design the future state end to end
  • Map the gaps in ways of working, cadence and handovers
  • Rank the use cases by value and feasibility

We build

  • Target workflows with current and future maps
  • A prioritised, staged list of use cases
  • Process objectives the team signs up to

Automating one step only moves the bottleneck. Redraw the whole journey.

Step 04

Governance

What can people do with AI today, and how does oversight keep pace as use grows?

What happens

  • Set the policy: an AI vision tied to brand values, an acceptable use policy and a risk-tagged list of use cases
  • Govern data and models: a lawful basis, consent and clear documentation
  • Build the controls into the work: sign-off before launch, brand guardrails and a QA step before anything publishes
  • Monitor, audit and feed what you learn back into policy every quarter

We build

  • Guardrails by marketing activity
  • A green, amber or red route for any new use case
  • A committee sized to the maturity, with a calendar and a named owner

Four domains and a loop

  1. Strategy and policyThe why and the what
  2. Data and modelsThe with what
  3. Operations and controlsDay-to-day guardrails
  4. Monitor and improveClosing the loop

Oversight runs across the whole lifecycle. It is not a one-off approval.

Step 05

Technology choices

Which tools and agents fit the workflows, and what must be true underneath?

What happens

  • Map the architecture and the data sources
  • Assess vendors and the current stack
  • Run RFPs and proofs of concept
  • Specify bespoke agents where off-the-shelf tools fall short

We build

  • A shortlist of tools and partners
  • High-level technology choices
  • Agent specifications tied to named workflows

Four tests for any tool

  1. Does it solve a named workflow problem?
  2. Does the data underneath support it?
  3. Does it help people do new things, not only automate old tasks?
  4. Can we measure the result against the plan?

A stack is not a plan.

Step 06

Organisation and role design

How do roles, skills and structure change when AI does more of the execution?

What happens

  • Draft the future marketing model
  • Write skills and job profiles
  • Plan for the people whose roles change most, including the identity concerns that come with it
  • Pilot new ways of working with a small team before scaling

We build

  • A blueprint of the future marketing organisation
  • Role and skills profiles
  • A plan for the people whose work changes most

With AI taking on execution, fewer people are needed in the lower layers, so the pyramid becomes a diamond. Hiring and skill-building follow.

Roles
Builders, orchestrators and standard bearers
Unit of change
The workflow, not the department
Step 07

Execute

Which use cases go live first, and how do we know they work?

What happens

  • Score every use case on revenue impact, effort, data readiness and risk
  • Pilot the top two or three with one team and a named owner
  • Set a baseline and one metric per use case before launch
  • Review at 30 days, then scale, fix or stop

We build

  • A ranked list of use cases
  • Live pilots, each with an owner and a metric
  • A 30-day results review that decides what scales

Execute where use cases pay back

Revenue
Deep CRM personalisation, product advisor, competitor switcher
Insight
Churn signal, expansion scoring
Productivity
Content production, lead management
Step 08

Change management

How does the team build skills and habits that persist?

What happens

  • Train by role and tie it to real work, not generic courses
  • Ask leaders to share one personal AI use case each week
  • Share small wins and failures openly
  • Track adoption and review it on a regular rhythm

We build

  • A role-specific training plan
  • Adoption measures the team can read
  • A review rhythm that feeds the next plan

What drives adoption

Capability
Learning that lasts, not one-off courses
Opportunity
Less friction, and no “AI does not work here” myth
Motivation
Leaders who model it and small wins that get broadcast

Make it easy, attractive, social and timely.

“Edwin brought clarity and structure to ambiguity, always asking the right questions and helping the team focus on what truly mattered for long-term impact.”
CCClaudia CassanHead of Manpower Europe
AI practice

Practice behind the AI transformation method

ManpowerGroup

Nov 2021 to Jul 2025 · enterprise scale
  • Built the AI strategy, reporting to the US Deputy CEO
  • Prioritised three key use cases and agreed the implementation workflow
  • Defined a new value-add capability for the business
  • Took the AI roadmap through the executive committee, which approved it

Boldstream.ai

2026 · Transformational Marketing AI
  • Set up the AI transformation consultancy practice, working directly with the CEO
  • Built the methodology, playbook, pricing and sales approach
  • Supported clients including Wickes and The Outnet, from AI implementation to an automated marketing calendar
  • Supported the website rollout
Questions people ask

AI in marketing: common questions answered

What are the steps to implement AI in a marketing team?

There are eight: marketing vision, readiness, workflows, governance, technology choices, organisation and role design, execution and change management. Annual planning and adoption reviews send the team back to the first step.

How do you prioritise (prioritize) AI use cases in marketing?

Score each use case on revenue impact, effort, data readiness and risk. Pilot the top two or three with one team and a named owner, set a baseline and one metric each, then review at 30 days and scale, fix or stop.

How should marketing govern AI?

Cover four domains: strategy and policy, data and models, operations and controls, and monitoring and improvement. Treat oversight as a loop across the whole lifecycle, not a one-off approval, and review it every quarter.