Finding, Adding and Maximising Value in Your Data Platform
Finding, Adding and Maximising Value in Your Data Platform
How to find value, how to bring value early as possible and where to start designing and building your Data Platform.
Note, this is part of an in-progress guide on “How to Build a Data Platform”, subscribe for future updates:
Introduction
What is the right way to build a Data Platform? It may not be what you originally designed or even what you wanted to build, but it is the thing that brings you most value to the organisation.
For the most part, everything else can be considered a footnote or technicality compared to the features of your data platform that bring value to your organisation. Without bringing value, you just built set of data pipelines that burns money.
So what brings value?
For most organisations “value” means money, in money earned or saved. This can be difficult to calculate for data solutions because it is often seen as support for value generation, rather than generate value directly. But working out the organisation’s value in it’s data solutions and services has the following benefits:
It helps build a business case for now and the future (you may have the board backing now, but will you have in the future?).
It can show the rest of the organisation that data can help grow the business directly rather act as support.
It’s can be fine to only get approximate value especially if the value gained is far more than cost of building a Data Platform.
Another way to measure value is ask the data consumers what value they gain from data and aggregate that up; when collecting requirements from a customer, think about asking the customer what monetary value it bring to the company. This helps in the following ways:
This will allow to prioritize requests by value added
Also will give you a rough idea of your data platform value - which should make building business cases in the future easier.
Lastly, you can decentralise your Data Team into each of the business domains (marketing, finance, etc.) and have budgets for each of those domains. This will make it very clear how much each business is spending on data. If more investment in the business domain’s data is required, then the business can ask for it directly.
This is the Data Products and/or Data Mesh approach, which we mentioned in the introduction and will discuss in more detail later, though many organisations are moving this way as data teams can generating value directly rather than supporting it.
This does mean you need to completely reorganise your data team(s) if they are not already in this structure, Data Meshes can also give you a less strong central data governance and may not make a lot of sense for small Data Teams (less than 5 people), so it does require some thought into how implement to this approach.
Will a New Data Platform Even Bring Value to Your Organization?
Can a Data Platform return value on it’s investment? You shouldn’t build / buy what you don’t need - consider the option of doing nothing as a baseline to compare against.
This isn’t just a binary decision too: you can look at refactoring or optimising existing Data Platforms to maximise value while keeping costs low.
Return on Investment (ROI) can be discovered by researching the value added of a data platform in your organization and the cost to make one. Though there is a tradeoff here: you want to spend long enough to build a strong business case, but not so long the business has changed rendering your forecasts irrelevant.
This is usually the point at which you’ll look at your Data Strategy (or even write one!) to help you find and improve the value in your Data Platform or, more generally, in all your organisational data. We won’t go into depth here on Data Strategy or its tools and techniques such as Data Maturity Assessments and Data Value Chains, but if you’re interested, you can read our “Ultimate Guide to Data Strategy“ for more details.
You can also de-risk and get a better idea of potential ROI / value by building Proof of Concepts (PoCs) and/or Minimum Viable Products (MvPs) for a subset of your future Data Platform: better to spend a tenth of the investment on a failed PoC than all of the investment on a failed Data Platform!
Bring Value as Early as Possible
Building a Data Platform will at least take months and likely years to build completely, even if you buy a off the shelf product, as you still need to learn it and configure it to your needs. However, planning in detail that far ahead can be wasted resources due to a number of factors that means your plans will change every month, week or even day:
The average CDO stays in their role for only 30 months - usually a new director likes to change direction from the old director.
Business plans are changed every 3 to 12 months
Data technology is always changing and improving, which can outdate existing designs
Feedback from existing work has meant you have to change your designs.
Spending too long in strategy and design or building the foundations can increase your chances you’re designing the wrong solution even if you started on the right path.
So you should look to build in stages and/or in thin slices so you can gather feedback that you are building the right thing as early as possible. That said, spend not enough time on strategy and design and you’ll build wrong solution from the start - it is a balance, though try to keep your scope for the initial build as small as possible to reduce up front design time while delivering enough so the Data Platform actually provides value when released to its end users.
One way to keep scope low and keep focus on what brings value is to start the discovery and design at the data outputs (Reports, Dashboards, etc.) and work backwards:
Looking at raw data first is not advisable as until you know what outputs you are building you don’t what exact data you’ll need for it.
What’s the Best Place to Start?
Picking your initial use cases is very important and can be difficult - picking a unsuitable first use case or product can lead to issues that doom your whole Data Platform effort and the issues may take awhile to become apparent. Sometimes you only get one opportunity to show how a new / upgraded Data Platform will add value, so great care should be taken to pick the right place to start.
One simple way to determine where you start is to score your use cases by value created and level of risk out of 5, with the risk level being determined by a number of factors: forecasted time to build, software complexity, business enthusiasm, access to source data, business change and more. We won’t say how to weigh each of the risks here, as it will be different in each organisation. You can factor in forecasted cost of use case as well, though be careful with giving a exact cost at this stage - there will be a lot or variability in your forecast.
Then take away your risk from your value created and rank use cases from high to low, showing your most appealing use cases at the top.
Note there are many other frameworks of various complexity for prioritising use cases, choose the framework that suits your planning process.
Set Goals to Aim For
So you now know the value of your current Data Platform, you’ve limited the scope of the initial phase to absolute minimum and know where to start: but when you start designing and building how do you check the Platform is meeting it’s goals?
You’ll want to set goals for your Data Platform and for each of your Data Product(s) in a Data Platform that roll up to your Data Platform goals. The Data Platform goals should also rollup into your Data Strategy and Business goals so you can build a strong case why your Data Product and/or Platform is bringing value now and in the future.
Their are various frameworks for setting goals, with SMART being one of the most adopted, though adopt the goals framework that suits you and your organisation. The important point is to have something to aim for that be tracked over time.
You’ll also want to collect baseline metrics when you launch your Data Platform, such as number of users per day, so you can track trends, which can then be used to build evidence on where next best to improve your platform.
If you’re looking to go down the path of Data Products, should have a look at Product Thinking so you make sure you are focusing on your customer problems and not features
Feel Free to Ignore all the Above!
If feel strongly about what direction you should go in, then move on to designing and building your Data Platform, especially if you already have a clear Data Strategy in place; don’t try and find value if you already know where it is.
That said, we would at least recommend writing down why you’re building or updating an Data Platform and what added value it will bring, because with most big investments such as this you’ll audited on the decisions you made at some point and a quick way to do that is write a Product One Pager.
Sponsored by The Oakland Group, a full service data consultancy.
Photo by Alexander Grey on Unsplash