METHODOLOGY
IMPACT AI™: adopting AI with evidence, not demos
From curiosity about AI to measurable, governed and sustainable adoption.
IMPACT AI™ is how we support an organisation from the first conversation about AI until it uses AI in its processes, measures the results and governs it without depending on us. Six phases, each closed with a gate: a control question, minimum evidence and a documented decision. A meeting, a workshop or a prototype are activities; passing a gate requires evidence.
We don't start with which AI tool to use. We start with the business, the problem, the process, the people, the data and the expected result. Technology comes afterwards.
Six phases, six decision gates
- IdentifyGate G1: Do we understand the problem and the opportunities?
- MeasureGate G2: Is the company ready?
- PrioritizeGate G3: Which case deserves to go first?
- ActivateGate G4: Does the solution prove its value?
- ConsolidateGate G5: Is it used and does it deliver results?
- TransferGate G6: Can the company carry on?
Phase 1 · IDENTIFY
Gate G1: Do we understand the problem and the opportunities?
Understanding the business as the people who run it experience it
We talk to management, but above all to the people who do the work every day. That is where the real friction and the opportunities that never appear on an organisation chart come to light.
- Interviews on strategy, customers, operations, information, decisions and exceptions: what should happen and what really happens.
- Light mapping of each process: users, inputs, activities, data, outputs, indicators and pain points.
- Inventory of the AI already in use, including AI built into applications.
- Each opportunity as a testable hypothesis: 'if we apply this capability to this process, we could improve this result for this user'.
- Still without choosing a tool or a vendor.
Deliverable: AI Opportunity Canvas™ and Process & Pain Map™
Phase 2 · MEASURE
Gate G2: Is the company ready?
Measuring readiness with evidence, not with an opinion
We establish the starting point with the AI Readiness Index™, which assesses six dimensions: strategy, processes, data, technology, people, and governance and risk. Every score is backed by evidence; anything that cannot be checked is recorded as 'not verified'. We measure the data dimension with the UNE 0077 (governance), 0078 (management) and 0079 (quality) specifications, assessed with UNE 0080, and the DAMA-DMBOK framework.
- AI Readiness Index™: six dimensions assessed with evidence.
- Data maturity according to UNE 0080 and the DAMA-DMBOK knowledge areas: quality, metadata, master data, security, architecture…
- Critical gaps and risks that would block a pilot.
- Enabling actions, and a sponsor who understands what needs fixing.
Deliverable: AI Readiness Index™ and gaps to close
MEASURE IN DETAIL
AI Readiness Index™: six dimensions
We assess each dimension with evidence. Where there is no evidence, it is recorded as 'not verified' rather than assumed.
Strategy
Objectives, sponsor, and decision-making and investment capacity.
Processes
Digitalisation, stability, indicators and existing automation.
Data
Availability, quality, access and ownership, measured with UNE 0080.
Technology
Applications, integration, cloud and AI tools in use.
People
Skills, current use of AI, resistance and capacity for change.
Governance and risk
Privacy, security, control and oversight.
Five readiness levels
- ExplorerStart small and train.
- StarterQuick wins with support.
- ReadyActivate the priority use cases.
- AdopterConsolidate and expand.
- ScalableUse case portfolio and scaling.
THE DATA DIMENSION
Data maturity levels
According to UNE 0080, each data governance, management and quality process reaches a maturity level. This is what each level looks like and what it takes to move up to the next.
Performed
Data processes are carried out, but reactively and dependent on specific individuals.
To move up a level: Identify critical data, appoint owners and apply the first rules.
Managed
Processes are planned and tracked in specific areas or projects, with identified owners.
To move up a level: Define a corporate model: policies, roles and common processes for the whole organisation.
Established
Processes are standardised and documented across the whole organisation.
To move up a level: Measure them with indicators, targets and thresholds, and analyse the causes of deviations.
Predictable
Processes are managed with quantitative indicators and their outcome is predictable.
To move up a level: Establish continuous improvement and innovation in data processes.
Innovating
Continuous improvement: data is managed to obtain maximum value and adapts to new objectives.
To move up a level: Sustain the level and extend it to new domains, data and AI initiatives.
Phase 3 · PRIORITIZE
Gate G3: Which case deserves to go first?
Choosing the right place to start
We score each opportunity on value, feasibility, adoption, effort and risk. The score does not replace professional judgement: it forces a structured conversation and puts the reasons in writing. We prioritise short quick wins that deliver visible business value and frame each use case as an MVP: the minimum scope needed to validate the hypothesis with real data before investing further.
- Scoring from 1 to 5 on value, feasibility, adoption, effort and risk, with the reasons documented.
- Light business case: current volume and cost (measured or estimated) and expected improvement as a hypothesis, not a promise.
- Quick wins lasting a few weeks, framed as MVPs and with defined pilot KPIs.
- Meeting with 3–5 finalists and a decision by the sponsor: GO, conditional GO, PARK or NO GO.
Deliverable: AI Opportunity Portfolio™ and GO / NO GO decision
Phase 4 · ACTIVATE
Gate G4: Does the solution prove its value?
Piloting with control, not with faith
The prioritised use case becomes learning and real evidence. Training is tied to the use case, using tasks from the process itself, not delivered as a disconnected course. The pilot has a controlled scope, is governed from day one and does not 'work' because a demo impresses: it works if it meets the defined KPIs.
- Use case design in writing: problem, user, current workflow, solution, data, scope, KPIs, risks and owner.
- Applied training: essential concepts, the limits of AI and human review, validated with a real task.
- Pilot Charter before starting: hypothesis, users, duration, metrics, success criterion and stop condition.
- Pilot log: usage, issues, user feedback, KPIs and adjustments.
- Evidence-based decision: adopt, iterate or stop.
Deliverable: Use Case Canvas™, Pilot Charter and measured result
ACTIVATE IN DETAIL
How we know whether the project is succeeding
In addition to the committees, each pilot has its Pilot Charter and its own control dashboard, with measures defined before starting.
Steering committees
- Steering committee: decisions, priorities and risks.
- Operational committee: progress, issues and next steps.
- Decisions recorded at each gate.
Pilot Charter
- Hypothesis, scope, users and agreed duration.
- Metrics and success criterion set before starting.
- Dependencies and stop condition.
Quick win control
- Baseline measured before starting.
- Business KPIs: hours, errors, cost per transaction, sales.
- Success criterion reviewed at every committee.
Project control
- Timeline, cost and scope against the plan.
- Quality: accuracy of responses or predictions.
- Log of usage, issues and user feedback.
Phase 5 · CONSOLIDATE
Gate G5: Is it used and does it deliver results?
Making sure it is used and keeps working
A solution that works but nobody uses adds nothing. In this phase we build the change into the real way of working, assign owners, control the risks and measure the results. If the use case is not being used, we first look for the barrier; then we talk about scaling.
- Adoption Plan: people and champions, future process, training, support, communication and follow-up.
- Human oversight: what the AI produces, what a person reviews and which thresholds require escalation.
- AI Governance Lite™: owner, users, data, risks, controls and review, without unnecessary red tape.
- Success Dashboard with usage, efficiency, business, adoption and risk indicators.
Deliverable: Adoption Plan, AI Governance Lite™ and Success Dashboard
CONSOLIDATE IN DETAIL
From pilot to way of working
Adoption requires real use and observed results. These are the pieces we leave up and running.
Adoption Plan
- Affected users, champions and owners.
- Which process step changes and what the exceptions are.
- Training, support, communication and follow-up.
Human oversight
- What the AI produces and what a person reviews.
- Which decisions can be automated and which cannot.
- Escalation thresholds and the owner of the final result.
AI Governance Lite™
- Purpose, business owner and users.
- Data, risks and controls: review, permissions and security.
- When each review takes place and who carries it out.
Success Dashboard
- Usage: active users, frequency and tasks.
- Efficiency and business: time, cost, errors, quality, conversion.
- Adoption and risk: sustained use, issues and human reviews.
Phase 6 · TRANSFER
Gate G6: Can the company carry on?
Our work ends when you are in control
Transfer is not a final meeting with a document. We apply shadow and reverse shadow: first your team works alongside us while we operate; then your team operates while we support and supervise, until it can carry on alone.
- Shadow: your team accompanies ours in operations and in every decision.
- Reverse shadow: your team operates and we supervise and answer questions.
- Capability by role: management, use case owner, users and technical profile.
- AI Action Plan™: prioritised backlog, data capabilities, training, controls and next milestones.
- A new readiness measurement, compared with the starting point, and closure with the sponsor.
Deliverable: AI Action Plan™ and roadmap
TRANSFER IN DETAIL
Shadow and reverse shadow
We transfer knowledge by working together, not with a manual at the end of the project.
We operate
We build and operate the solution, with your team kept informed from the outset.
Shadow
Your team works alongside us: it sees how we operate, make decisions and resolve issues.
Reverse shadow
Your team operates and we supervise, correct and answer questions.
Full control
Your team operates on its own, with documentation, backlog and roadmap. We remain available if you need us.
Six principles we don't compromise on
Business before technology
We don't propose tools before understanding the goal and the process.
Evidence before opinion
We separate observed facts, perceptions and hypotheses.
Value and adoption
A technically feasible case nobody will use is not a priority.
Proportionate risk
Controls suited to the risk and the size of the company, without red tape.
Experiment before scaling
We test hypotheses with a narrow scope and KPIs.
Transfer
Every engagement builds internal capability. No dependency.
FAQ
Frequently asked questions
Straight answers to what we're asked most.
How long does an IMPACT AI™ project take?
It depends on the scope. The first phases are short, and quick wins are designed to deliver results in a few weeks. The schedule is set when the project starts.
Do we have to go through every phase?
We start wherever each organisation is. If you already have your use cases prioritised, for example, we start with Activate, applying the checks from the earlier phases.
How do you measure data maturity?
With the UNE 0077, 0078 and 0079 specifications, assessed according to UNE 0080, and the DAMA-DMBOK framework, complemented by technological readiness, people and AI governance.
What is a quick win in IMPACT AI™?
A short use case, framed as an MVP, that delivers visible, measurable business value and validates the hypothesis before investing further.
What is a gate?
The close of each phase: a control question, minimum evidence and a documented decision by the sponsor. You do not move on to the next phase for having held meetings or workshops, but for having evidence.
What happens if the pilot does not deliver the expected result?
The decision is made with data: adopt, iterate or stop. A NO GO is also a useful result: it avoids investing in what does not work and leaves learning and capability in the company.
What is the training like?
Learning by doing. Training runs through every phase and is built on the company's own use cases and tasks, including the limits of AI and human review. By the end, each participant should be able to explain the use case, carry it out and know when not to trust the result.
What do shadow and reverse shadow mean?
Two stages of transfer: in shadow, your team accompanies ours while we operate; in reverse shadow, your team operates and we supervise, until it is in control.
Shall we start with the assessment?
A first 30-minute conversation, no strings attached.