Data strategy and ownership
By Redaktion techport.ai, IT-Beratung · Last updated on
Data only gets discussed in companies when something does not work: when two reports show different numbers, when an automation fails on empty fields, or when an AI initiative cannot start because the basis is missing. Until then data counts as a by-product of systems.
A data strategy in the mid-market is not a programme with committees. It consists of a few decisions: which data really matters, which system is authoritative for which data, who owns it and which rules apply to entry and maintenance.
How you notice it
- Two analyses of the same question produce different results.
- When asked where a figure came from, nobody can trace it to the source.
- There is no binding definition of central terms such as customer, order or active.
- Data maintenance is felt to be an annoying extra task that nobody owns.
Why this happens
Data arises where work happens, not where it is later analysed. The person creating a record has no interest in an analysis six months later but wants to complete their transaction. As long as nobody connects entry with usage, and as long as poor data produces no noticeable feedback, quality is a matter of personal diligence. That is exactly why it varies so widely between departments.
How we go about it
- Name the business critical data. We determine with the departments the few data objects the business depends on, typically customers, articles, prices, orders, suppliers and employees, and describe per object what the data is used for.
- Set ownership and authority. We define per object which system is authoritative and which person owns it professionally. That role decides on mandatory fields, value lists and rules, not IT.
- Define the terms. We fix the central terms in binding form, with definition and calculation rule, so that analyses are read the same way across the company. That collection grows with every report and belongs in one central place.
- Anchor the rules in the systems. We implement the definitions where data arises: mandatory fields, validations, value lists, duplicate checks. Rules that exist only in a document have no effect.
What you gain
- Analyses whose figures are no longer debated.
- A solid basis for automation and AI initiatives.
- Clear responsibility for data questions instead of referrals between areas.
From our projects
Defining an authoritative system per data object is the single measure with the greatest effect and at the same time the one that generates the most discussion. The reason is rarely technical: whoever owns a system gives up part of their freedom. We therefore run that discussion along examples from daily work rather than along principles, for example a specific case where a changed address in one system never arrived in the other. The second recurring finding: definitions are missing exactly where they cause the most friction, particularly around what counts as an active customer and when an order is considered complete.
Häufige Fragen
Do we need a dedicated role for data ownership?
Not a dedicated post, but a named role in the departments. In most mid-sized companies that is an experienced person per data object who takes this on in addition, with a clearly limited time share. What matters is that the role has decision rights, otherwise it becomes a collection point for complaints.
What comes first, the data strategy or the reporting?
In practice both together. A data strategy without a concrete occasion stays abstract and does not get implemented. We therefore usually start with one important analysis, settle definitions and ownership for the objects involved along the way, and transfer the result to the others afterwards.
Let us talk about Data strategy and ownership
In a thirty minute first call we work out where your biggest lever sits and whether we are the right people for it.
Further reading
Back to the field Data and Information