The Week in Data, Analytics & AI #1
Again a great week in data! Just to name a few things...
I experience every week many Data, Analytics and AI aspects. While I wrote about before on LinkedIn articles or the last two weeks in Substack Notes, I make now a try to make a real Substack article out of this weeks experience.
Data Catalog failure mode 1 - Not the first time this week that I heared something like “McKinsey recommended a Data Catalog for Data Governance. But the customer now don’t know why they really need it…”. It is a repeating pattern. As an external consultancy our job sometimes is to recommend. But if we do strategy without execution, possibly nothing happens. Sometimes our job is just to understand better what a Data Catalog can do for them. Believe me. This is the hard job.
Data Catalog failure mode 2 - I just had several discussion about “we already have a ‘catalog’ in AWS (-> Glue), Databricks (-> Unity), SAP Datasphere (-> Catalog), Snowflake (-> Horizon) … - while those ‘catalogs’ expand their capabilities they are still not and maybe will never be an Enterprise Data Catalog. If you need something for business-oriented End-to-End Data Governance those kinds will probably never be the solution you need. Talking to my colleague about we ended with “You know that, I know that. Now we have to bring this understanding to the people.
Data, AI and AI news literacy - 5 cent for every news someone telling that my job will be obsolete in 2 years because AI will take over… We talk about Data Literacy (understanding and being able to work with data), AI literacy (understanding how AI works to use it the right way) but no one talks about how to check latest studies or news about AI job apocalypse. Maybe AI will have impact maybe not. I believe my job is like an octopus. Adaptive as hell.
Fig. 2: Anthropics Report about Observed Exposure (Instagram) Data Products are not for everyone - I just worked with a customer where especially a group building the data governance practice was so in love with the data product concept that they throw it everywhere. Everything was now a data product. I asked them, what data products means for thema and no one could answer. Now the term is forbidden until they get a better understanding.
Fig. 3: Data-as-a-Product as part of Data Mesh (Instagram) Organization Model for Data - Is a federated model better than a centralized model and decentralized models are the enemy? Hell, no! It always depends. And there is never a black or white state. Typically I work with a 5 option model or even rather with a continuous model to don’t get stuck on one mode. Never oversimplify just because it is easier to explain to decision makers. Always explain what is important to set up your data organization.
Fig. 4: Data Organisation Models Learning in Data, Analytics and AI - is like boiling the ocean. There are three rules of thumb from my perspective.
First: Learn what you have to learn for your daily job. There is no alternative.
Second: Follow your passion and learn things not necessary tomorrow but you are interested in.
Third: Learning is a marathon not a sprint. Find ways to do something regularly is better than shifting everything to a learning block at the weekend.
Fig. 5: From my LinkedIn post
AI - focus on the next step - Many customers think today in renewing their business model with AI. Making everything better and different with artificial intelligence. Reallity in 2026 still is learning to understand how to work right with AI, understand the costs when you see “AI inflation”, pick a process to optimize. It is still more realistic for most to think iterativ than revolutionary.
Fig 6. : From my Instagram daily data insights So far some insights from my week. How was your week in Data, Analytics and AI?
Hei,
I would like to reuse Fig. 4. How do I cite you?