Why the success of your data strategy starts with central and decentralised data teams

Data silos can be a challenge. Even organisations that have a data platform find that teams still work on their own islands. Finance works with Excel sheets, IT builds its own dashboards and operations designs its own extractions. Everyone wants to be fast and agile, but only together will you get further. So why is it that working together with data remains so difficult? And more importantly, how do you solve it?

The cause is not in your tech stack

On paper, it’s simple. With the right tools, you can unlock data centrally, process it and make it available to the whole organisation. In practice, the barrier is rarely technical. Instead, it is organisational structures, ownership and culture that stand in the way of cooperation.

When IT cannot respond quickly enough to requests from the various departments, initiatives arise outside the central data teams. Departments hire their own data engineers, platform engineers and data architects, build their own solutions and try to gain speed. In the short term, this works. In the long term, however, it mainly results in duplication of work and thus inefficiency.

What does a good alternative look like?

The solution is not in complete centralisation or radical decentralisation. It is the balance that makes the difference. In a well-functioning data model, the technical infrastructure is central, while the knowledge and execution is decentralised as much as possible. That sounds abstract, but it simply means:

• Central teams take care of infrastructure, security and data source access.
• Decentralised teams focus on building data products and insights close to the business, leveraging their specialist knowledge.

This hybrid setup resembles the idea of a data mesh, in which domain teams own their own data but are supported by central frameworks. And the great thing is: it works.

Step by step towards a working ecosystem

If you want to break through data silos, don’t start with technology, but with overview and agreements. These steps help:

1. Map where you stand
   Which teams have what data? What has already been built? Where is there overlap?
2. Determine what you organise centrally and decentrally
   Opening up core systems such as your ERP, CRM and HR systems is best organised centrally. Domain-specific data analyses, on the other hand, belong in the teams themselves.

Frameworks for data ingestion and publication are also ideally organised centrally. Think of standard pipelines for easy connection of data sources (ingestion), or streamlined publication channels for dashboards or reports. This allows decentralised teams to focus on value creation, without reinventing the technical wheel each time.

1. Get the right people in the right place
   Decentralised teams need people who understand the business and can work with data/IT. Then give them the mandate to adjust processes. If you don’t, you run the risk of ending up in a spreadsheet culture.
2. Make it attractive to participate
   New solutions should not only be good, they should be better than what people are using now. A standard disclosure that every team can get started with quickly is more powerful than a loose integration in a silo

.

When do you know you are ready for change?

There are some clear signs that your data strategy could use an upgrade:

• IT can’t move fast enough
  Your business departments are waiting too long for reports or new dashboards.
• Data engineers are in the business
  If platform or data engineers work in decentralised departments such as finance or sales, it is a sign that their work should actually be centrally organised.
• Everyone accesses the same data source in their own way
  Multiple teams separately build integrations with the same systems. Result: duplication of work, inconsistency, unnecessary cloud and software costs

.

What does it deliver?

If the puzzle is right and you have implemented the plan, you will quickly see results. Your organisation becomes more agile. Data products are developed faster, insights are better shared and your IT landscape becomes clearer. Moreover:

• Cloud costs go down, because you don’t have to maintain infrastructure and processes twice.
• Operational costs go down, thanks to less manual work and less error-prone data flows.
• Time-to-market speeds up, because domain teams can build directly on what has been put in place centrally.

Collaboration as key

At Blenddata, we believe in blend, build, deliver and the interplay between people, technology, organisation. We guide organisations through the transition to a central-decentralised data model and from analysis & advice to implementation. Not with thick reports, but with working solutions that fit your organisation’s agility.
A clear trend we see: central teams making ingestion and publication frameworks available to decentralised teams. This lowers the technical threshold and accelerates adoption of the data platform ovwer the entire organisation.
Wondering how to set this up smartly and scalably within your organisation too? Blenddata helps with both analysis and implementation. Feel free to contact us. We are happy to think along with you.