Issue #8 - Deliver on the Data Needs, not the Data Desires
How to master the art of knowing what the business needs
Read time: 8 minutes
Data people often think they are the smartest people in the room.
This is their downfall…
In reality, they might be, but it doesn’t matter. What matters is who has the power in the room. This does not boil down to who is most senior, confident or has the experience; this is about who carries the most influence to make an impact.
When it comes to the corporate world, business stakeholders are the ones directly driving revenue and, therefore, they often have the most influence.
And when it comes to building data products and solutions, this fact changes the overall equation.
Where Data Goes to Die
Despite all the great approaches to building data solutions—agile methodology, design thinking, user testing, quality assurance, etc.—the amount of failed data projects is astronomical.
Gartner says 85% of big data projects fail and only 20% of analytical insights deliver business outcomes. VentureBeat says 87% of data science projects don’t make it to production. These numbers are a few years old, but from my experience, they are still fairly accurate (especially with the advent of premature AI initiatives).
Why is this?
Let’s take the typical process. Maybe a business stakeholder mentions something off-hand about wanting access to a random KPI faster. Or potentially marketing will submit a business case for a new tool that they think they should have. The data team will instantly think, I got a solution for that, then go and build it.
This build often goes one of two ways:
Done without the stakeholder’s input – Input at the beginning is different than input throughout the process. After the use case is submitted, the data team will build it without requesting further information or perspectives. The problem here is the solution presented by the stakeholders may not be exactly what they need. It may be a cool dashboard or tool, but it may lack the right data, the right context, be too advanced for the end user, etc. Moreover, the definition of what a ‘simple solution’ means will differ between data and business teams, leading to a misinterpreted output.
Hyper attention to stakeholder’s input on the one question – The flip side is building the most unique and specific data solution possible, and not thinking about the wider context or scalability. An example of this is an organisation that has 800 dashboards that are accessed 1-2 times per month or maybe 1-2 times in their lifetime. Sometimes data teams cater too much to what the stakeholder is saying and don’t challenge their perspectives/ solutions (often due to a lack of business acumen).
Logical Progression of the Stakeholder Needs
Everybody talks about getting the needs of the business, but unfortunately most people/ teams do it the wrong way:
They start by asking “What data tools do you want?”
The business, in response, usually answers, “I want a tool that can do x, y, and z” with the answers usually in the automated, predictive and ML/ AI realm.
The data team says “okay” and the overall interaction creates unrealistic expectations of what the team can deliver.
The teams then don’t talk again until a prototype or tool is built, by which point the business stakeholders have forgotten about the conversation and what they said they wanted.
While people are well-intentioned here, these types of interactions and inquiries can do more harm than good.
Understanding the correct business needs is an art and a science. It involves assessing the situation, getting the perspectives of the user, using data expertise to chart the right path, establishing the relationship, and bringing the right people along for the ride. This is a process, not a one-time event!
Now before we get into what you should ask of the business stakeholders, I first want to outline the logical progression of steps a data team should aim to deliver on in the business’s ‘data-driven’ journey.
Step 1 – Build the Data Foundations
Start with ensuring there are strong data foundations across the business
This includes data governance, the right platform/ tooling, robust management processes, a clear data direction/ strategy, and having access to the right data
These don’t all need to be done, but there needs to be a level of completeness to avoid technical debt and inefficiency in delivery
Step 2 – Create the BI & Reporting Tooling
Start building timely, accurate, and actionable insights to support day-to-day decision-making via tools like Power BI, Tableau, Looker, etc.
This can be fairly standard tooling/ solutions that enable business stakeholders to access and interpret data easily, empowering them and the overall business
It leads to saved time, more growth opportunities and the creation of a more knowledgeable workforce (and delivers immediate, tangible value)
Step 3 – Incorporate Data Science & Predictive Analytics
Evolve the reporting & BI tools through analytical add-ons that uncover patterns, make predictions, and provide deeper insights
Done through advanced analytical methods, machine learning, and automation
Examples can be adding regression options, cluster analysis, AB testing potential, personalisation tooling, etc.
NOTE – This is where everybody wants to start, but skipping to this step is useless if the data is poor or if the original BI/ analytics tools aren’t working/ embedded properly
Step 4 – Explore Optimization, AI, and Prescriptive Analytics
The most advanced output that involves applying sophisticated algorithms and models to solve complex problems, optimizing processes, and making prescriptive recommendations
In normal speak, it means automating ML decision-making through AI or technological integration so that optimal decisions are easy to come by and ingrained in the business's day-to-day
This progression is something I will go into a lot more detail about when we talk about data use cases (a future article), but having that logical progression implanted in your brain will serve you well in the meantime.
Business Needs Interviews
Okay, we know why data projects fail and how initiatives should progress through the analytical lifecycle. Now, how do we ensure we are building the right things?
Well the first step is to ask the right questions to truly understand the business needs. Here are 8 questions in the order you should ask them:
What is your role and key stakeholders? – Understand their place in the organisation and who they report to/ work with to comprehend the wider picture of their role/ purpose
What are your business goals? – Get the context on what type of results they are looking to achieve and why they do what they do every day. This helps put you in their shoes
How do you measure your performance? What are your top 5 KPIs? – Be strict about limiting them to five areas of measurement, because realistically if they go beyond that, your solution will not deliver as required
How does your job deliver on the overall business strategy? – This is the link between their role and the business strategy. If you are building something for them, you want to make sure it has a larger overall impact on the organisation. What is the value that this solution will create?
How do you currently use data to achieve your goals? – The first data question about understanding their baseline literacy, uses, and how they view data within their role and team
Where do you think data can help with your job? – People often jump to this question, but the reason it is asked here is because now you have led them through the process of what their job is and what they want to achieve. This is also a place where you can jump in and lend your ideas/ expertise
What are the key challenges to making this a reality? – Understand why things have failed in the past and why data doesn’t currently play the role it should
What keeps you up at night? – My favourite question! What is it that worries them the most and hinders progress? This builds on the previous question but puts it more bluntly. Given this is a data-focused interview, you will uncover the most pertinent blockers to data progress with this one inquiry
Get answers to these eight questions and you will have the right ammunition to get the business needs.
Also notice how I didn’t ask about data until the 5th question?
By focusing on the business user, what they know and what they want before you starting talking about the D word you get them on your side and connect the dots between the business direction and potential data solutions.
In the end, bridging the gap between the business and data teams through this process comes down to:
Listening to their needs
Understanding the role they play
Determining how they create value
Collaboratively developing data solutions
Bringing the business user along on the journey
Doing it all with a logical and value-focused approach
Next week we dive back into the data-focused content with a hotly contested topic—how do we define data? Here we will get into the difficulties of defining data, the implications from that, what categories of data terminology exist, and different data structures. Basically, we are going to try and demystify what data actually means. See you then, and comment below on any questions or feedback you have!
Thanks for the read! Comment below and share the newsletter/ issue if you think it is relevant! Feel free to also follow me on LinkedIn (very active) or Medium (not so active). See you amazing folks next week!
It seems like muscle memory for data people to start with thinking and talking about the data and method for accomplishing a project.
Its like you have to essentially unlearn that and join the user in the business arena first in order to be effective, and then you can go back to your data and be the analyst once you have what you need!