On 21st September, 2022, we had Phillip Ade-Akanbi as the PM guest on our monthly #ProductDiveAMA.
Philip is a Data Science Product Manager with 6 years of professional experience. He started his career in Sales & Marketing before pivoting to Product Management during his time at the Independent National Electoral Commission (INEC) where he delivered digital products for the 2019 general elections in Nigeria.
He is passionate about democratizing data and manages the developments of Data-as-a-Product (DaaP) and the strategy for Data-as-a-Service (DaaS) solutions that continues to save thousands of dollars in operating cost, promotes work efficiency and boosts customer acquisition & retention.
Together with using data and advanced analytics to build products that help businesses and customers, he is passionate about learning and is currently undergoing an advanced study in Data Science at the University of East London.
He finds Open Banking fascinating, loves traveling, is an aspiring travel blogger and a fan of Liverpool FC.
ProductDiver: My first question will be everyone’s first question, haha
What is the role(s) of a Data Science Product Manager?
Phil: The role of a data science product manager (mostly called “data product manager”) to vaguely put, is a product manager with an advanced knowledge of data capabilities.
There are many definitions and even the role varies from company to company.
We will all agree that Product Managers ask questions such as “Who are the users”, “What is the market size/fit”, “What is needed to deliver this product” etc. However, a data scientist asks “What data is required (to achieve a business use case”, “How clean is the data”, “How should the data be processed and presented”
A data product manager sits at an intersection of who a product manager is and who a data scientist is.
For me, I would say that a data product manager is someone with product instincts and knowledge and can ideate how data can help to achieve a business goal, unlock the maximum value and make customers happy.
It is widely known that a product manager wears many hats, a data product manager wears even more hats!
Some might say data product management is harder than the regular PM because of the deep understanding of both worlds needed, especially the technicalities surrounding data generally, however, the solutions that the capabilities of data can afford makes it worth it immensely.
ProductDiver: What will you say has been your experience as a data science product manager? Would you prefer it to being a traditional product manager?
Phil: My experience as a data science product manager has been very interesting. Data product management takes the ordinary and adds some little ‘extra’ to it and that alone makes it interesting for me to keep doing what I do.
I very much prefer it to the traditional product management because the capabilities of data are just endless and you have to continue to learn something new and on-the-go.
ProductDiver: What’s the career progression to becoming a Data Science Product Manager?
Phil: It basically entails having expertise and sound knowledge in both fields.
For example, Data scientists build machine learning models and product managers build products. A data science product manager (DS PM) will be someone who takes a business problems or use case and creates a product with the output of the machine learning model.
Just as we have the traditional product management where the product managers have some form of domain knowledge in the area they are building in, a data science product manager’s main domain knowledge is understanding both worlds.
A product manager can become a DS PM and a data scientist can become a DS PM. It is just going that extra mile to understand both worlds.
ProductDiver: What does it entail to build a data product?
Phil: To build a data product, there must first be a business understanding/domain knowledge/business objective. Once these are understood, it will be easier to know the who, what, and why of the data product is being built.
Then, you prepare your data. Let me say at this point that the data affair is a messy affair. Every company in the world deals with dirty data and in the data world, it is Garbage In Garbage Out (GIGO). So if your data is nonsense before running an advanced analytics model, you will have nonsense as the result. In very clear terms.
Next, is the modelling proper where the company’s data is taken through a series of advanced analytics/high level statistics to get a type of output that speaks to the business problems.
After this is the evaluation of the results to ensure its accuracy and finally the delivery of the output (which must be in very clear terms to the nontechnical stakeholders).
So in all these, the DS PM is the point of contact for the entire (analytical) development cycle just as we have in the traditional product management.
ProductDiver: How is Data Management different from the regular product Management?
Phil: Data product management shares major semblances of the regular or traditional product management including user research, product/design thinking, user stories, backlog management, using Agile methodologies, creating requirement documents etc.
The major difference is the heavy reliance on data and advanced analytics to build the data products.
Same ProductDiver: Data-as-Product vs Data-as-a-Service?
Phil: Very interesting. To set the tone, let me ask this.
Why is it that you will google something, let’s say a work station and then you start seeing Ads of work station every where you turn to on your phone? That is the capability data at its finest refinement.
Another question, how does Boomplay know which music you might like or Netflix know which movie to recommend when you use their product the first time?
Data-as-a-product as the name implies, is basically creating and treating data as we would treat our digital products. Millions of data are produced everywhere per second and everything we interact with electronically produces data.
The phrase “data is the new oil” might have been overused in industry terminologies but many do not understand why that term exists. As it is with crude oil where it is drilled out from the ground and made to undergo numerous forms of processing to produce petroleum products, same it is for data.
DJ Patil, a popular data scientist coined the term “data products” and describes it as a product that facilitates an end goal through the use of data. Going by this definition, any product that relies on majorly on data can be termed a data product.
Data products are created with a specific purpose in mind and usually has a business case. Data products usually serve a specific purpose in an organization and utilizes mostly the same frameworks of the popular products we create as product managers.
A data product can be in many forms. For example, a marketing unit in an organisation can constantly request for data to track their customer’s behavior on a product from the data team and a data product can be created that helps them to access the needed data by themselves without the direct need of the data team. (Self-service)
Let’s not talk plenty grammar, a data product can be as simple as a visualisation dashboard or report, and as complex as a fraud detection engine.
As an example, how does a company like Visa or MasterCard know that a transaction that has not happened or is currently ongoing is fraudulent?
They have built a data (advanced analytics) product that analyses the transaction based on so many features to identify fraudulent transactions. The features can include the time of purchase, amount of item being purchased, type of store, transaction history of the customer etc. So they have aggregated this data and used it to build data product that does the analysis for them.
Still the same ProductDiver: How is Data Management different from the regular Product Management?
Phil: Data-as-a-Service (DaaS) is just like any “as-a-service” cloud-based software tool.
DaaS basically provides information or model outputs in which data are made available to customers over a network.
Companies that collect all forms of data from their customers are now looking at how they can get greater business value from the data they have and usually resort to DaaS to leverage data as a business asset for greater business performance.
A company like Facebook generates 4 petabytes of data per day. Now what do they do with these data in their care? They create a strategy where they can share this data with their partners who can then use it to derive further insights or create targeted Ads for its users across all its platforms.
With the advent of Open Banking in the world today, I found it very interesting when CBN created a framework for regulation of Open Banking in Nigeria.
What does this mean? It means every bank in Nigeria can share it’s heavily protected data with multiple registered partners who can then derive further insights or create a wider solution for the masses. This is a form of DaaS where the data one company has provides value to a wider pool of companies.
The same ProductDiver: Wooooow!
Phil: Companies like Okra and Mono are major players in this. They get customers financial data from their bank partners (with permission from the customers and regulations from CBN) and build advanced analytical models with it, usually a Credit Risk Assessment of the customers and then plug it to loan companies through an API. The loan company then uses that to determine if a customer will pay back their loans or the loan will become non-performing.
Connect with our guest https://www.linkedin.com/in/philipadeakanbi/