Is a monetised dataset a "Data Product"?

Yes, but there are probably more valuable products you can craft from that data

I was chatting the other day to a very experienced product coach on whether selling a dataset for money made it a data product.

It was a great discussion where I brought my data domain experience and they brought their product experience to the conversation. Even more interesting given I have some experience in the product domain and they have some experience in the data domain. We both brought a gun and a knife to the fight so to speak.

A planned 1 hour chat ended after 2 and half hours, and only because we ran out of time.

I of course started off by describing how people in the data domain spend endless hours and days arguing the semantic definition of what a data product is and what it isn’t.

So to short cut that process, I always provide the definition I use for an Information Product:

a product
encapsulating data, code, analytics and visualisation
which delivers information within a defined boundary
enabling actions and outcomes that drive business value

You might agree or disagree with this definition, but that’s not the important thing, the important thing is I am telling you the definition I use when framing my part of the conversation. I'm anchoring you to understanding my semantic context and my belief system.

If you have a definition of a data product, then slap that puppy on the table, and again I may agree with your definition or I may not, but I now have your context of which the rest of the conversation will be based.

So next time you are entering into a conservation about Data Products (or any other data domain semantics) try getting both parties to start off with anchoring statements.

Back to the original conversation.

The obvious part of the discussion was if the dataset could be sold for money then that made it a product.

I also used an argument that I was taught in a previous life, if a company was sold, but it was sold without the data about its customers and products being included in the sale, then that companies value would be much less and therefore data intrinsically always had value.

We then moved on and played around with the common Product lens of a product being Valuable, Viable and Feasible.

We felt that if the dataset was unique (which it was) and somebody would pay for it in its raw state (which they have) then it was valuable.

If that dataset could be safely provided to the customer in a way that met the relevant data governance, data privacy, data management principles and policies then it would be feasible.

Viability was an interesting one, it would be based on a couple of things.

Of course there is the pattern of making sure you sell it for more than it costs to produce it, if it has a positive margin then of course it is viable right?

But the other part of the viability discussion was how does that margin fit into the overall business strategy of the organisation that captures that data.

If the value of that data is constrained to a specific use case, or a specific industry, or a specific market, then there are only so many customers that will potential buy it (Total Addressable Market - TAM) and then what is the realistic number of customers in that market that will actually buy it (Serviceable Obtainable Market - SOM).

If we compare the margin from the SOM to the strategy of the organisation, how does it align?

If we are looking at massive revenue or margin growth as a core part of our business strategy, we might have a problem. If we are looking for this being just one of our go to market offerings we might be ok.

And that brought the conversation onto Tomatoes.

I often use the analogy of a restaurant, self service buffets, silver service, a kitchen, a meal, chefs and ingredients to describe the data domain.

So in this conversation our thoughts went something along the lines of:

If we have tomatoes that nobody else has, then we have something that has value, chefs who need those specific tomatoes will buy them from us.

And given we are the only ones with that type of tomato, we can charge a premium for them.

But as some stage we will price those tomatoes to high and chefs will either make do with another form of tomatoes or remove the need for tomatoes all together by removing the meal or adapting the recipe.

And so wouldn’t it be more valuable to make the meal itself.

In the dataset monetisation use case wouldn’t it be more valuable to wrap code, analytics and visualisations around that data to sell an Information Product. A product that solves a specific business problem, provides answers to specific business questions, answers that allow the desired actions to be taken, actions that result in a positive business outcome and therefore delivers business value.

Wouldn’t we be able to extract more value for that Information Product vs selling the raw dataset as a Data Product where the customer still has to do the majority of the work to extract the value needed from it.

Or put another way when we sell tomatoes they still need to cook the meal, so why don’t we just sell the entire meal.

We then bounced around discussing whether it would be delivered as a self service buffet, or a silver service delivered to the table, or a cook at home delivery meal.

And could we design and build multiple Information Products from the one dataset, using the Define Once, Reuse Often (DORO) principle we love in the data domain.

And would this allow us to cross sell and upsell within a single customer, to extract more value compared to the one and done process of selling the dataset.

And would this allow us to iterate and reuse the data factory patterns we would create to build the Information Product. Enabling us to build Information Products that served other use cases, other industries and other markets, potentially with a different dataset powering it.

Where we landed was to understand if the best strategy was to sell tomatoes or meals, more discovery work was needed to understand the Value, Viability and Feasibility of building and selling the Information Product vs selling the dataset.

And then we would need to compare that answer to the Business Strategy to see if monetising the dataset or the selling an Information Product had the best chance of success in achieving that strategy.

Two final thoughts I had after walking away from the conversation.

First, I often distinguish between a Data Asset vs an Information Product, when I am providing my anchoring statements in these conversations. I think I need to anchor the monetisation of a (semi) raw dataset as being a third thing, a Data Product.

Data Asset > Data Product > Information Product.

And then I need to work out if the raw datatset is used internally in the organisation, where money doesn’t trade hands, but value is still swapped, does that make it a Data Asset or a Data Product (got to sweat those semantic definitions!)

Second, when you talk to somebody who is very experienced in the Product Domain you spend a lot more time talking about the value of the thing that needs to be built, vs talking to somebody who is very experienced in the Data Domain where you spend a lot more time talking about how it will be built.