Daten-Readiness-Check – TRAIL1 – KI-Beratung für den Mittelstand

Block 1 of 5

Context

Size, responsibilities and system landscape

A1 How large is your company?

a Up to 50 employees b 50–250 employees c 250–1,000 employees d More than 1,000 employees

A2 Who looks after data and IT today?

a Nobody in particular — everyone handles their own area b One person on the side (the “IT colleague”) c An IT department, but with no explicit data focus d A data team or dedicated data owners

A3 Do confidentiality or customer requirements limit your use of the cloud?

a No, cloud is generally an option b EU cloud / GDPR-compliant hosting is a condition c Certain data must not leave the building (local processing required) d We don’t know for certain

A4 How many different systems does your business data come from (ERP, CRM, industry software, spreadsheet worlds …)?

a Essentially one system b 2–3 systems c 4–6 systems d More than 6 — nobody has the full picture

Block 2 of 5

Operational data

Reporting, KPIs, master data and production data

B1 How does your monthly management reporting come together?

a Largely automatically, straight out of our systems b Partly automated, partly reworked by hand c Mostly manual work in Excel d Every month it’s the same laborious copy-and-paste across several sources

B2 Do departments (sales, finance, production) turn up in meetings with different numbers for the same thing?

a Practically never — there is one agreed set of numbers b Rarely, and when it happens the cause is found quickly c Now and then, and it regularly costs discussion time d Regularly — “whose numbers are right?” is a standard question

B3 Roughly how many hours a month go into manually transferring and copying data across the company?

a Under 5 hours b 5–20 hours c 20–50 hours d Over 50 hours — or: “I’d rather not know”

B4 Do customers, articles or suppliers appear twice, or spelled differently, across your systems?

a No, master data is actively maintained b Occasionally, but it hardly bothers anyone c Yes, regularly — it makes per-customer and per-article analysis unreliable d Constantly — or: we don’t know

B5 Do you capture production, machine or quality data — and do you analyse it?

a We capture and analyse systematically / not relevant for us b We capture it and analyse it occasionally c We capture it but barely analyse it — the data sits idle d We barely capture it, even though it affects our business

Block 3 of 5

Knowledge

Findability, key people, free text and documentation

C1 How long does someone typically search for a specific piece of information (a contract clause, a spec, an instruction, an old case)?

a Minutes — central storage, decent search b Up to an hour, depending on the topic c Often longer — several repositories, a lot of clicking d People ask the colleague who knows — searching isn’t worth it

C2 Will key knowledge holders leave the company within the next 3 years (retirement, career move)?

a No, nothing on the horizon b One or two, but they are replaceable c Yes, individuals with critical knowledge d Yes, several — and their knowledge is barely documented

C3 Do you produce free text that nobody analyses systematically — service reports, complaint notes, enquiry emails, minutes?

a Hardly any relevant free text b Yes, on a small scale c Yes, noticeable volumes — analysing it would be valuable d Yes, large volumes — there is a lot of unused knowledge in there

C4 When documentation is found: how current and reliable is it?

a Current and maintained b Mostly usable, with a few gaps c Partly outdated — people double-check anyway d Nobody really trusts it

Block 4 of 5

Ability to start

Internal capacity and previous AI experience

D1 Is there someone in-house who is “good with Excel and data” and could be given time?

a Yes, we could name them right away b Probably, we’d have to check c No — everyone is tied up in day-to-day work d We would want to outsource this entirely

D2 What has already been tried with AI?

a Nothing yet b Individuals privately (ChatGPT and the like) c We bought licences (e.g. Copilot), the effect is unclear d First productive use, with rules in place

Block 5 of 5

Goal

What should be different in 12 months

E1 What should be different in 12 months? (Pick the most important one)

a Analysis and reporting run automatically instead of by hand b We make decisions based on reliable KPIs and forecasts c Our internal knowledge is searchable by simply asking a question d Text and documents (reports, emails, complaints) are processed automatically e We’re not sure yet

What should you actually build with your data?