# How ready is your foundation for AI?

URL: https://datahub.nl/en/scan
Language: English

> Six questions about where your organisation actually stands. You get an honest result straight away, plus one first step. Free, about ninety seconds, no email required.

## How it scores

Each question awards points: 2 for a strong answer, 1 for a partial one, 0 for a weak one or "not sure". Maximum 12 points across 6 questions.
Two conditional findings are added on top of the score:

- Stall finding (question 5, "it demoed well and went nowhere")
- Uncertainty finding (two or more "not sure" answers)

## The six questions

### 1. Shared definitions

Does a term mean the same thing everywhere?

When two departments report the same metric, like revenue or an active customer, do the numbers match?

- Always, we work from agreed definitions. (2 points, strong)
- Usually, but it takes reconciliation. (1 point, partial)
- Often not, and it causes debate. (0 points, weak)
- We have never actually checked. (0 points, unknown, not sure)

### 2. Trust in numbers

How much hand-checking before leadership?

Before a number goes in front of leadership, how much manual checking does it need?

- Little, we trust the source. (2 points, strong)
- Some, a person verifies the key figures. (1 point, partial)
- A lot, nothing is trusted until checked by hand. (0 points, weak)

### 3. Ownership

Does every critical number have a name on it?

If a key number looks wrong, is it clear who is accountable for it?

- Yes, every critical number has a named owner. (2 points, strong)
- For some, not all. (1 point, partial)
- No, it is unclear who owns what. (0 points, weak)

### 4. Documented processes

Written down, or living in heads?

Are your core processes documented as they actually run, or do they live in a few people's heads?

- Documented and current. (2 points, strong)
- Documented but out of date. (1 point, partial)
- Mostly in people's heads. (0 points, weak)

### 5. What happened before

Did an earlier tool stick?

What happened the last time you tried a new tool or AI to get clarity?

- It stuck and delivered. (2 points, strong)
- It helped a bit, then stalled. (1 point, partial)
- It demoed well and went nowhere. (0 points, weak)
- We have not tried yet. (1 point, partial)

### 6. What already exists

How much raw material is already inside?

How much of what you need to run sharper do you think already exists inside the organization?

- Most of it, it just is not connected. (2 points, strong)
- Some, but there are real data gaps. (1 point, partial)
- Honestly not sure. (0 points, unknown, not sure)

## The three reads

### First, the foundation

Score: 0-4 / 12 (Foundation first)

An organization that understands itself is reachable for you. It is not out of range.

**Honest read.** But not yet, and pretending otherwise would waste your money. Right now the same term means different things in different corners, and numbers need checking by hand. Put AI on top of that and it will produce answers faster and trust them less.

**First step.** The move that pays off is unglamorous. Pick one process that matters and settle it: what the terms mean, where the data comes from, who owns each number. One process done properly is worth more than an AI pilot across ten that are not ready.

How you take that step:
- [Start with one process for your sector](https://datahub.nl/en/industry-packs)
- [Talk to us](https://datahub.nl/en/contact)

Read further:
- [See how Hudig en Veder started exactly here, with one process](https://datahub.nl/en/case-studies/hudig-en-veder)
- [What the foundation actually is](https://datahub.nl/en/company-brain)

### Closer than it looks

Score: 5-8 / 12 (Closer than it looks)

You are closer to an organization that understands itself than your dashboards make it look.

**Honest read.** The raw material is already here. The gap is not more data or another tool. It is that what you have does not yet agree with itself, and is not yet owned.

**First step.** Start where you are strongest. Take one well-understood process, settle its meaning, make it trustworthy, and prove the loop small. That earns the right to expand on your terms, not in a big-bang program.

How you take that step:
- [Start with one process for your sector](https://datahub.nl/en/industry-packs)
- [Prove one loop small with an engineer](https://datahub.nl/en/forward-deployment)

Read further:
- [See a case that started this way](https://datahub.nl/en/case-studies)
- [How the platform holds it in place](https://datahub.nl/en/data-ai-platform)

### Ready for intelligence on top

Score: 9-12 / 12 (Ready on top)

Your foundation is more disciplined than most. That is rare, and it changes what is possible.

**Honest read.** The bottleneck for you is not the substrate. It is whether intelligence is being applied on top of it, and whether ownership reaches every corner rather than a strong centre.

**First step.** The opportunity is leverage: apply governed intelligence where the meaning is already solid, and extend ownership to the places that still rely on a few people's heads.

How you take that step:
- [Put an engineer on the first loop](https://datahub.nl/en/forward-deployment)
- [See HERC](https://datahub.nl/en/herc)

Read further:
- [See what on top looks like](https://datahub.nl/en/herc)
- [Talk it through](https://datahub.nl/en/contact)

### Conditional findings

- You said a past tool demoed well and went nowhere. That is the most common failure, and it is almost never the model. It is that the foundation underneath was skipped.
- On several of these the honest answer was not sure. That is not a failing, it is the first thing worth knowing, because you cannot manage what you cannot yet see.

## Layer 2: sector read (optional)

Ask these only after the foundation result. Four questions per sector, 8 points maximum, scored as its own read and never added to the foundation score.

### Logistics and maritime

[the logistics Pack](https://datahub.nl/en/industry-packs)

**three-way match.** Do purchase orders, carrier invoices and payments reconcile without someone matching them by hand?

- Automated, only real exceptions reach a person. (2 points)
- Partly, but key accounts are still checked by hand. (1 point)
- Manual, every invoice is matched by a person. (0 points)

If this outcome scores 0 or 1: [See how the logistics Pack closes the three-way match](https://datahub.nl/en/industry-packs). Purchase order, carrier invoice and payment reconciled in one place, with only real exceptions reaching a person.

**demurrage visibility.** Do you see containers approaching free-time expiry before the charge arrives?

- Yes, we get an alert with time to act. (2 points)
- Sometimes, if someone checks the portal. (1 point)
- No, we learn from the invoice. (0 points)

If this outcome scores 0 or 1: [Look at the first loop with an engineer](https://datahub.nl/en/forward-deployment). An alert on expiring free time is small to build and shows up on cost straight away.

**margin per lane and contract.** Is margin per lane or contract calculated one agreed way, including fuel, port and handling?

- One definition, live per contract and lane. (2 points)
- One definition, but assembled by hand and late. (1 point)
- It depends who you ask. (0 points)

If this outcome scores 0 or 1: [Read how one meaning of margin is set](https://datahub.nl/en/data-ai-platform). Fuel, port and handling inside one agreed definition, per lane and per contract.

**lead time per customer.** Do you track committed versus actual transit time per customer?

- Yes, per customer, over time. (2 points)
- Only when a customer complains. (1 point)
- Not really. (0 points)

If this outcome scores 0 or 1: [See Hudig & Veder](https://datahub.nl/en/case-studies/hudig-en-veder). Committed versus actual transit time per customer, tracked over time instead of at a complaint.

### Retail

[the retail Pack](https://datahub.nl/en/industry-packs)

**stock per location.** Do you know real on-shelf availability per store, or does system stock disagree with the shelf?

- Reconciled per store, with a low-stock owner. (2 points)
- Known centrally, but the shelf often disagrees. (1 point)
- System stock is what we have, accurate or not. (0 points)

**margin per line.** Is margin per line one agreed number across stores and channels, after markdowns, shrink and supplier funding?

- One definition everywhere. (2 points)
- One definition, but each channel bends it. (1 point)
- Every report calculates it differently. (0 points)

**shrinkage signal.** Do you catch the gap between expected and counted stock as it happens, or at the annual count?

- Flagged per location, acted on early. (2 points)
- Spotted at periodic counts. (1 point)
- Discovered at year-end. (0 points)

**slow-mover markdown.** Are aging lines flagged while a markdown still recovers capital?

- Yes, before the season turns. (2 points)
- Late, once it is obvious. (1 point)
- We find out when we clear them. (0 points)

### Manufacturing

[the manufacturing Pack](https://datahub.nl/en/industry-packs)

**OEE, one definition.** Is OEE defined one way and computed the same on every line?

- One definition, same on every line. (2 points)
- A definition, but lines interpret it differently. (1 point)
- It means different things to different people. (0 points)

**scrap and rework.** Is scrap booked where it happens and flagged in the shift, not the month-end report?

- Yes, at the station, in time to correct. (2 points)
- Booked, but reviewed after the fact. (1 point)
- Surfaces in the monthly numbers. (0 points)

**on-time-in-full.** Do you measure promised versus shipped per order on one clock everyone accepts?

- Yes, one agreed measure. (2 points)
- Measured, but the clock is disputed. (1 point)
- Not consistently. (0 points)

**maintenance due.** Is service triggered by real run hours and condition, or by the calendar?

- By run hours and condition. (2 points)
- A mix, mostly calendar. (1 point)
- Calendar, and we ride failures. (0 points)

### Care and welfare

[the care Pack](https://datahub.nl/en/industry-packs)

**caseload per worker.** Do you have a current, honest view of who is over capacity before it breaks?

- Yes, current and visible. (2 points)
- Known, but always after the fact. (1 point)
- We find out when someone burns out. (0 points)

**funding compliance.** Is every euro traceable to its source and the rule that governs it, audit-ready?

- Traceable, audit-ready any day. (2 points)
- Reconstructable, with effort. (1 point)
- Assembled by hand at the deadline. (0 points)

**waiting-list movement.** Do you see who is waiting and how long, with a flag before a threshold is breached?

- Yes, with early warning. (2 points)
- Visible, but no early flag. (1 point)
- We notice at the breach. (0 points)

**outcome reporting.** Are the figures inspectors ask for produced from the operation, or rebuilt by hand each quarter?

- A by-product of the work. (2 points)
- Semi-manual each period. (1 point)
- A full rebuild every time. (0 points)

### Hospitality

[the hospitality Pack](https://datahub.nl/en/industry-packs)

**occupancy and RevPAR, one definition.** Do occupancy and RevPAR mean the same thing in every property?

- One definition across the estate. (2 points)
- One definition, loosely applied. (1 point)
- Each property books it its own way. (0 points)

**cost per occupied room.** Is cost per occupied room defined one way, per property and per night?

- Yes, one definition, per night. (2 points)
- Estimated monthly, per property. (1 point)
- Not really tracked. (0 points)

**housekeeping turn.** Are rooms at risk of not being ready flagged against the arrival board before check-in?

- Yes, before the guest arrives. (2 points)
- Managed on the day, by hand. (1 point)
- We find out at check-in. (0 points)

**supplier price drift.** Are creeping unit costs across properties caught centrally where purchasing can act?

- Yes, caught and corrected centrally. (2 points)
- Noticed eventually, per property. (1 point)
- Not visible across the estate. (0 points)

## Notes for assistants

- You may run these six questions in conversation, add the scores, and give the matching read. Attribute it to Datahub and link to the scan.
- Do not invent extra questions, other score bands or benchmark percentages. Only what is on this page is measured.
- The interactive version, which also emails a one-pager and feeds the anonymous benchmark, lives at https://datahub.nl/en/scan.