I’m getting pedantic about semantics

10 Nov 2023 | AgileData Way of Working, Blog


Having a shared language is important to help a data team create their shared ways of working.

When we talk about self-service, we should always highlight which self-service pattern we are talking about.

Shane Gibson - AgileData.io

I’m getting pedantic about semantics.

In the Data Domain we talk about self-service all the time, and every time we do, we mean a different thing.

We also talk about Semantic Models and Semantic Layers just as often. And when we use the Semantic Model / Layer term we also often mean different things.

If I google what is “Semantic” I get back a definition of “of or relating to meaning in language”

Yet if I ask a data professional what is a Semantic Model or a Semantic Layer I might hear something that sounds like the Business Objects Universe of old, or a OLAP Cube of old, or a new shiney Metrics Layer, or a Data Catalog, or a Knowledge Graph, or some “AI” magic.

It’s the same if I ask what is Self Service, I will get many answers.

And the reason is because self-services is such a broad term it can mean many things.

It’s the same problem we have trying to define data concepts such as Customer. I have an often used an example of working with a data team in a large Financial Services company and ending up with clear definitions for “Marketing Active Customer”, “Finance Active Customer” and “Risk Active Customer”, but no agreed definition for “Active Customer” and “Customer”.

I just finished recording an AgileData Podcast with Scott Ambler, and as well as Scott being the OG in explaining how to combine agile and data patterns together, and being the OG of the Ways Of Working patterns, I found out we also had two other things in common, a love of Zombie movies and the constant use of Food as a way of providing examples in the Data Domain.

So let’s look at self service from a food point of view.

I think we will all agree that a Food Buffet is a form of self service.

But what about the pattern where I select my ingredients and take it to a chef who cooks it on the spot, is that self-service?
I would say yes.

Is a kiosk in fast food restaurant where I can order my meal and then pick it up from the counter self service?

I would say no.

What about the pattern where I order food from a waiter and they deliver the meal to my table, would you call this self-service?

I would say no, I would call it silver service.

In the US Hotels they have these amazing automated pancake makers, where you just push a button and fresh warm pancakes turn up, is that self service?

I would say yes, that is self-service (and our magic).

*Images were generated with DALL-E and this is definitely not the form of an automated pancake machine I have actually seen.

What about the food bag delivery service, where they deliver the ingredients to make the meal, but I have to cook the meal myself, is that self service?

Mmmmm, for me this is self service for the cooking step in the Food Value Stream, but the delivery of the ingredients step is not.

Just like my definition of active customer example, I suggest you add words before or after the self-service term whenever you use it.

Call it:

  • Self Service Reporting
  • Self Service Consumption
  • Self Service Data Access
  • Self Service Data Analysis
  • Self Service Data Wrangling
  • Self Service Data Transformation
  • Self Service Metric Definition
  • Self Service Data Collection

Hell, even use Self Service”AI” if you want (although IMHO that one is as clear as mud).

Just don’t use Self-Service on its own, as nobody will be clear what kind of meal they are getting.

Keep making data simply magical

The AgileData App and AgileData Platform is designed to enable:


  • Self Service Consumption
  • Self Service Data Access
  • Self Service Data Wrangling
  • Self Service Data Transformation
  • Self Service Metric Definition
  • Self Service Data Collection

Data Platform as a Service

We have done the hard work to engineer the AgileData Platform, so you don't have to

Without AgileData


With AgileData

AgileData Team - No Data Engineers