Methodology & Evidence
What exactly does “measurement-first” measure — and how?
We openly explain the method behind our reports — not our question bank or scoring formula, but the approach itself. That is our transparency promise.
What is the MindDX methodology?
The MindDX methodology is the working framework that measures a company's digital maturity and ties that measurement to a delivery chain: an 8-dimension maturity model, a question set adapted to the company type, evidence types whose source is always named, a confidence score that rates the answers themselves, and a written human/AI division of labour per product. This page explains how the measuring is done and where it stops — why we work this way is on the approach page, what comes out of it is on the sample outputs page.
What we measure
The 8-dimension maturity model
The full Checkup measures these 8 dimensions with 100+ questions adapted to your company profile. Technical dimensions answer “can we do it”, the leadership/employee dimensions answer “will we adopt it” — both are scored together:
Profile adaptation: same model, a different question set for each of three company types
“Adapted to your company profile” means something concrete: the question bank splits by company type. Manufacturers see production line, bill-of-materials, planning and supply questions; companies that don't manufacture are never asked them. Trade and distribution firms get demand and inventory planning, reorder points, stock accuracy, goods-in/dispatch and channel profitability questions; service firms get service delivery, resource planning and delivery quality questions. You state your company type at the start of the assessment.
We do this not out of courtesy but because of measurement error: an unanswerable question produces a dip in that dimension that isn't real. The dimensions, the layers and the scoring logic are identical across all three profiles, and the number of questions applied to each profile is kept equal. That second point is a deliberate design decision — it is the only way a manufacturer's score stays on the same scale as a distributor's or a service firm's.
Answer quality: one guided question at a time, with level examples
The biggest risk in a long questionnaire is that the company clicks without reading — and a score built on corrupted input is worse than no score at all. So in the full assessment questions arrive one at a time, the dimensions are tracked on a rail at the top, and next to each level option sits a concrete example of what that level looks like in practice. The aim isn't to make answering easier, but to make the right answer distinguishable.
Two answers: where you are today, and where you want to go
On every question we take two separate marks: your current level and your target level. A firm at 2 aiming for 3 and a firm at 2 aiming for 5 have the same score but not the same priority. The gap between target and today shows which dimension to tackle first with a clarity the score alone cannot provide.
The declared target does NOT enter the score. Your maturity score is computed solely from the answers about your current state; the target is used in the report's priority ranking and roadmap. We keep that separation deliberately — otherwise the score would stop being a measurement, become a statement of intent, and lose its comparability across companies.
What it rests on
Scientific basis: why we measure interaction alongside technology
Our measurement model rests on the socio-technical systems approach: lasting results come from designing technology and organisation together — through joint optimisation. Decades of research show that even the best technology isn't adopted if the organisation and its interactions aren't ready for it. So we measure processes, roles and interaction first, and technology second.
That is why the leadership and employee dimensions are an integral part of the report. We score “can we do it” (technical) and “will we adopt it” (human) together — because transformation is carried not by technology but by the interaction that turns it into results.
Current consulting evidence points the same way: in McKinsey's State of AI research, the three factors separating high performers — workflow redesign, defined human validation processes and executive ownership — map one-to-one to MindDX's three core principles: process/role/interaction first, supervised AI, and a leadership layer measured on its own.
Interaction is assessed at three levels:
- With yourself — The manager owning the decision, self-awareness, learning agility and giving direction under uncertainty.
- With your team — A working culture based on trust, openness and participation that carries change together.
- With your environment — The relationship with suppliers, customers, technology providers and the business ecosystem in a digital context.
Where AI moved the effort — and why we measure first
BCG's 2025 analysis shows that GenAI shifts the effort in ERP transformation away from coding and configuration toward design, data readiness and process decisions: as technical implementation gets cheaper, what determines value is getting the right scope and data foundation in place up front. That is exactly why MindDX runs measurement-first — it front-loads planning, maturity measurement and configuration documentation. BCG, 2025.
Evidence discipline
Evidence types — what do we mean by “evidence-based”?
Every finding has a visible source. A finding rests on one or more of these evidence types; its weight in the report depends on the source:
| Evidence type | What it means |
|---|---|
| Self-assessment | Your questionnaire answers — the starting point, declarative. |
| Stakeholder interview | Answers gathered from different roles in the discovery chat. |
| Uploaded document | Files such as process docs, org charts, existing reports. |
| System data | Structured information from your current setup (where available). |
| Process evidence | Findings derived from real transaction flows (partner-delivered, on request). |
| Consultant observation | Notes added by the expert consultant during review. |
Confidence score — how much can we trust the answers?
Every report carries a confidence score: answer consistency, disagreement between stakeholders and unanswered areas are assessed. Low confidence is never hidden — it is flagged openly, with the extra evidence that would raise it.
Numbers and their sources
Every research statistic used on this site lives in a registry entry: source link, publication context and last verification date. A number that is not in the registry does not go on a page; every registered number is re-verified every 90 days, and an entry past its date turns our automated check red — so a stale figure cannot quietly survive. Today 7 statistics are registered, and all of them come from these primary sources:
Last verified: 2026-07-22
Humans and AI
Humans and AI: who does what, on which product
- AI produces the report DRAFT from your answers and evidence.
- The draft passes schema and process rules (missing/inconsistent areas are caught).
- On the measurement reports (Checkup and Audit) an expert consultant reviews, corrects and approves the findings.
- On self-service deliverables (Discover, Design, Train) the entitled customer generates the document; where a consultant approval exists it is stamped on it.
- Critical actions (e.g. pushing settings to your system) are never applied without preview + human approval.
We do not say “every deliverable is expert-approved”, because it isn't. What the AI produces and what a human checks is written per product in the table below — where a product has no human in the loop, the table says exactly that.
Assurance: AI produces, humans approve
We don't sell raw AI output. For every product, what the AI produces and what a human reviews and approves is explicit:
| Product | AI produces | Human review & approval |
|---|---|---|
| MindDX Checkup | Produces the report draft (8 dimensions, confidence score) | An expert consultant reviews and approves — final after review |
| MindDX Audit | Produces a department-level health assessment + action list from a structured questionnaire | A consultant reviews and approves |
| Discover · Scope Discovery | Produces the scope-discovery draft | Self-service; consultant-guided on older Odoo versions |
| Discover · Deep-Dive + Blueprint | Produces the fit/gap analysis + conceptual design draft | Self-service; consultant-guided on older Odoo versions |
| Design (config + UAT) | Produces configuration instructions + UAT/test scenario drafts | The entitled customer generates it; consultant approval is stamped on the document |
| MindDX Train | Produces department- and role-based training document + quiz drafts | The entitled customer generates it; quizzes are scored deterministically and results are visible to consultant and manager — no consultant approval |
| Process (Process X-ray) | Feeds the process analysis (partner technology) | Partner (ProcessMind) delivery, consultant-guided |
| Support Copilot | Produces answers from the knowledge base of the product you bought | None — these answers are not reviewed by a consultant; when the knowledge base cannot answer, it points you to a support agreement |
At checkout you choose «self-service» or «consultant support». That choice decides who runs the SETUP: in self-service your team proceeds with guides and pushes to your own system with preview + approval; with consultant support a MindDX consultant runs the process. Report assurance follows the delivery model above and is independent of this choice.
The UAT and test scenarios Design produces are compared against real flows in the Process (process verification) layer: before go-live we rehearse the configuration against the scenarios, catching surprises in testing rather than at the customer.
Limits
Our deliberate limits — and the reasons
We deliberately don't do some things. Each one is there because it makes the measurement more reliable or more honest:
| We don't connect to your live system | Measurement runs on a structured questionnaire, documents you upload and a consultant interview. So it carries no risk into your production system and your data security stays with you — a connection is set up only when you explicitly ask, in a controlled scope. |
| Scoring runs on a fixed formula | The same answers produce the same score; the result is reproducible and comparable. A free-generating language model wouldn't hand out a different “score” each time — trust and fairness require this. |
| Custom code development is out of package scope | For gaps that standard configuration can't cover we produce a technical document; where needed, AI-assisted development proceeds under consultant supervision, in a separate scope. The aim isn't a “we do everything” promise, but clarity on what is delivered and how. |
One principle we publish openly: the leadership layer is never blended into your overall maturity score — it is measured as a separate score and read together with Organizational Maturity; because research shows executive ownership to be one of the strongest differentiators, and a self-assessment should not carry the headline number. The content of our question bank, the dimension weights and the scoring formula remain trade secrets — what we publish is the framework and the evidence discipline.
Frequently Asked Questions