April 29, 2022
How to drive value from your data strategy?
In late March 2022, YOOI’s CEO, Nicolas Averseng joined the LightsOnData show to discuss with George Firican the value of D&A, break down common challenges organizations face today in D&A value management and share what can be done to become value-driven.
The LightsOnData show is one of the top self-media that invites global leaders to talk about anything related to data and how to treat data as an asset.
This is a journey about how organizations, regardless of the industries and sizes, can gain more value through their D&A investments. Here are the key takeaways of the interview and happy reading!
What do we mean by value?
Value is everything that serves the overall business strategy.
In a broad sense, value is everything that serves the overall business strategy. So to determine "value", it is key to start by clearly identifying what matters for the organization.
Data can be used for a large variety of use cases, from launching new products to optimizing processes, reducing anomalies, improving environmental impact, etc. As such, it brings a wide variety of impacts.
- Financial value is the most obvious and direct type, it includes incremental revenue or costs reduction. And this is probably the main thing enterprises are looking for in the long run.
- But there are a lot of other types of value. They ultimately and sometimes indirectly have a financial impact, but do not necessarily have a monetary value in the short term: customer satisfaction, churn rate, employee happiness, environmental impact, etc.
Most importantly, when we talk about the value, we also need to look after the associated costs and risks that are required to be assessed, tracked and managed. It is because they can either negatively impact the expected value, or even have a much broader negative impact (financial, legal, image, etc.).
Data assets are powerful and valuable only if they are correctly and properly used. But the analysts usually report, year after year, that more the 80% of data & analytics investments fail to deliver any value.
So as part of the data strategy, organizations need to define the criteria and standards to measure value, and actually start tracking it over time.
What are the current challenges that companies are having to drive value?
There are certainly many challenges. Overall, the problem is that two-thirds of the cases that fail to deliver value are because of organizational or project management issues, instead of technology-related ones.
- It can be the projects that fail to obtain access to the required data,
- Fail due to data quality issues,
- Fail to pass the PoC phase because industrialization was not anticipated,
- Fail to get use cases adopted,
- Fail because compliance and regulations limits were not properly identified,
- Or even fail to address the right problem from the beginning.
Still, the main focus of many organizations remains on technology. Building capabilities without clear or prioritized business use cases is not one of the success criteria.
What is so often missing is:
- the clarity on what the organization wants to achieve (a.k.a. how does the organization deliver value, and is data supposed to enable that)
- the actual measurement of outcomes afterwards
In a recent discussion with a CDO, we were joking about how much effort goes into the budgeting process, to justify investments. Yet, none of the original assumptions that went into the approval are actually tracked or even measured afterwards.
What is your take on data-driven vs. value-driven?
That's something that has been quite a debate lately, and it may look like experts are picky and playing with words. But in the end, I think it matters to convey the right message.
What organizations need is to be able to make better decisions, based on data-supported insights. So the foundation is indeed the data, but this is not what it is in the driving seat. E.g. if you don't understand very well what you want to achieve, your current context, then you have no way to use data the right way, perform the right analysis, etc.
There are specific cases where you can directly sell and monetize data. Otherwise, monetization and value will come from what you will do with the data.
The term "value-driven" really captures that imperative to reverse the thinking and drive your strategies and operations by starting with the value you want to achieve.
Any advice for companies who want to become value-driven?
There are 3 key principles that I can repeat over and over and that are key to becoming value-driven :
- The first and the most critical one is to always start from the expected outcomes / what you want to achieve, and avoid the common trap of thinking about technology first. This is probably the most frequent source of struggles.
- Second, achieving data-driven value creation at scale is a long journey for organizations, thus you need to also have a clear assessment of your current maturity. As this will be a driver to assess prerequisites and change management needs. For this matter, we have a quick maturity test available for free that provides the first level of answers.
- And the third one is that the best way to deliver value and to get the transformation in motion is to deliver initiatives that have what we call the 3 Us: Useful, Useable and Used.
A data strategy has to be a very pragmatic way to enable the business strategy, with iterative delivery and tangible outcomes.
By achieving concrete operational benefits and having users onboard, you can animate the transformation to extract more and more value from data assets.
What are "data assets"?
I tend to have a broad definition for data assets: “any data-related item owned by an organization, that, when exploited adequately and efficiently, can generate value for the organization.”
Obviously, the first kind of data asset is the data sources themselves. This is what most people talk about and refer to since it is rightly the foundation.
But when looking at the full data value chain, there are different kinds of assets at each stage, and they all need to be properly managed, as they all:
- carry some dimensions of potential and realized value,
- correspond to specific risks
- and have associated costs.
For example, we also have:
- ML models
- Data pipelines
- Dashboards and reports
- APIs, Streams
It is important to not only consider this complete view of assets and understand how they are connected together but also know how they are contributing or could contribute to value creation.
YOOI is a value management solution, can you explain how portfolio management can help with value creation?
We have created YOOI because organizations struggle with creating value with their Data & Analytics investments. Generally, they don't really track the existing efforts and assets which makes it difficult to justify further investments.
Managing Data & Analytics investments as a portfolio is a powerful way to build and animate the transformation.
To give a better definition, portfolio management means being able to manage all assets and initiatives on their full lifecycle. And by connecting it to the business strategy and outcomes along that lifecycle it becomes value management.
- The lifecycle starts with ideation, to capture the right ideas and demands from across the organization, and also capture the link with required assets.
- Then, in order to define the ideal roadmap, initiatives need to be assessed and then prioritized in terms of risks, costs, expected outcomes and dependencies.
- During development, the challenge is to ensure alignment and to track progress and operational performance, across the value chain - which involves many different actors and many different specialized tools.
- Ultimately, when D&A use cases are deployed in production, the operational, financial and qualitative performance need to be tracked over time. This is how effective value can be assessed, and to further decide on which portfolio to take.
What makes YOOI unique, and specifically suited for Data & Analytics, is the ability to manage the portfolio of assets and of initiatives in a combined way. The solution makes it possible to track which value is delivered by which initiative which in turn relies on which assets.
This kind of "business lineage" is very powerful to really understand how the various assets contribute in terms of value, but also to better understand the associated risks. For example, assessing the potential impact of a data quality issue and justifying investments.
By having this 360° view and the ability to track the value of both initiatives and assets all along their lifecycle, enterprises can really ensure that their Data & Analytics investments have an impact and deliver tangible value.
What is the place of data governance in this holistic approach to data?
Data governance is an important part of being able to use data in a safe and efficient manner, by :
- Maintaining an accurate inventory of data sources
- Documenting and curating the corresponding metadata,
- Consolidating data quality information
- Helping with data discovery
- Enforcing policies and protecting data access, ensure compliance on how and when data can be used (sometimes facilitating those policies with workflows, to get approvals efficiently and track those decisions)
But a common pitfall of data governance programs is that they are viewed as inhibitors and not enough as enablers. The reason is that once again when this is built with a bottom-up approach, teams get overwhelmed with the complexity and the processes - potentially losing sight of the actual goals.
One thing I learned from processes over the years is that too often in organizations the processes become the goal instead of being means to achieve business objectives. And that's the moment when it becomes an administrative burden, even when it was initially seen as a way to be more effective, to have a better alignment & communication, or to keep risks under control.
This is something data governance is facing. And then we hear data governance people complaining about the difficulty to engage users or executives in those initiatives, or about the fact that there is not enough realization of how important data governance is.
So the need is to reverse the thinking here as well. Really build or rebuild the data governance program and processes as part of the data strategy with a value-driven, collaborative and user-centered approach.
Is there a difference between Data & Analytics governance and Data governance?
Yes! When you look at the whole value chain, there is more governance than "just" data governance.
- Project governance, with specific gates, validations and reporting,
- Models governance, in order to track models performance or drift, across various versions,
- Infrastructure governance, in order to track cloud costs but also ensure proper locations are used
- Analytics governance, to cover the access, usage and lifecycle of reports and dashboards
- Data pipelines governance, APIs governance, Security governance, etc.
In almost every case, we can identify the corresponding "Ops" approach to investing in automation as much as possible (DataOps, DataGovOps, MLOps, AIOps, DevSecOps, FinOps, etc.). And each of these practices is operating with specific experts and its own toolset, therefore creating new silos of information.
If the organization doesn't pay close attention, they might all be going in slightly different directions, chasing their own specific targets.
This means you really can't have each of those programs operate independently.
The real challenge is to find a way to connect and align all these pieces together, and to ensure they all serve the same set of business goals:
- To ensure that the coverage in terms of policies and compliance is complete (avoid holes in the racket)
- To ensure that decisions taken are coherent across those different aspects (avoid silos)
- To ensure that the overall workflow is streamlined for the teams and various stakeholders
To achieve all these goals, there are two important things to look after :
- Always start with the end goal and expected business value in mind - I am repeating myself, but this is really the key to not building a byzantine set of processes,
- Having a way to build visibility on that full value chain, to be able to drive alignment and productivity toward those goals
And this is where Data & Analytics portfolio management is powerful: it makes the connection between the business strategy (which defines how value is measured), initiatives that are vehicles to deliver value, and those different kinds of assets.
And to operate that end-to-end view is where a more holistic approach to governance can be efficiently defined and deployed.
And how measure/evaluate the value of an organization’s investments?
It all goes back to the starting point, looking at how the goals, metrics and KPIs identified during the definition of the roadmap are actually evolving. In other words, this is the time to evaluate expected vs realized value.
Depending on the maturity of the organization and the nature of the selected value types, the way to measure is fairly different:
- Manual assessments for each initiative. For example, introduce user interviews when a complex measurement has to be deployed. it potentially needs to be peer-reviewed and rely on approval workflows to avoid massive biases.
- Crowdsourced evaluation. For example, rely on polls which usually require some animation during the poll and manual post-processing afterwards
- It can sometimes be automated. For example, tracking adoption or operation metrics.
What is really important is:
- Start doing it and using pragmatic even if manual approaches initially
- Be able to gather both qualitative and quantitative evaluations
- Be able to consolidate all this information easily, to then be able to provide extended visibility and support investment decisions
And maybe it is a new "Ops" (ValueOps) on the way. You need to invest in automation to streamline those processes and scale value monitoring and be able to gain agility on investment decisions.
How Culture is key in this journey?
Culture is key because this is what will enable adoption first and further drive scale afterwards.
Culture is how people behave: how they collaborate, how they take decisions, etc. So the whole point of the data culture is not to bring knowledge but to drive a change - and that's why it is so complex.
We know that deriving insights and making decisions based on data can be difficult, with many potential biases. We have seen that a lot lately at scale with COVID: just by looking at the same statistics, different people manage to make very different analyses leading to almost opposite conclusions. Usually, it is because of the way they incorporate other contextual information.
I think that many upfront approaches toward data literacy or other forms of purely training-oriented approaches typically fail because they are too theoretical and based on knowledge. They don’t have enough practical approaches and are based on behavior change.
To change behaviors, you obviously need to have training support to acquire the right knowledge, but you first and foremost need to have the opportunity to put into action your operational tasks and processes.
What comes first, data strategy or data culture?
I like those chicken or egg questions.
Data strategy first is definitely the key for me, as this is the only way to define a common goal. But you need to be aware of your maturity - and that includes the cultural aspect, to be able to prioritize efforts accordingly.
Feel free to replay the webinar with George Firican on the LightsOnData show where we answered some other spontaneous questions from our audiences and discussed how they can maximize the value from their D&A.
Don’t hesitate to contact our team to learn more and schedule a demo and we’re always here to help!