Building Trust through Data Product Design

Business Challenges

A lack of trust in data often arises when business teams feel that the data they are using is unavailable, unreliable, inconsistent, or difficult to interpret. This mistrust can stem from several issues:

Data Quality Problems: Inaccurate, outdated, or incomplete data can lead to misleading insights, which creates doubts about the reliability of data.

            Impacts:

• Credibility of the analytics team is diminished.
• Reluctance of business stakeholders to act on analytics as they are sceptical of their validity.
• Shadow IT – without trust in the analytics team, business teams will rely on anecdotal insights and alternative data sources.

Lack of transparency and documentation: When the methods used by analytics teams to collect, process, or analyse data are not transparent, business teams may question the validity of the results.

Impacts:

• Business users are unable to utilise data or insights that are available to them due to a lack of understanding on how it has been derived, or how best to use it or the tools available.
• Unmaintained analysis that no longer reflect business language or requirements.

Misalignment of Analytics: If the KPIs or metrics being analysed do not align with business objectives, stakeholders may perceive the analysis as irrelevant or out of touch with their needs.

Impacts:

• Wasted Resources due to lack of trust. Business teams will challenge reporting outputs, requiring repeated validation and rework due to a lack of documentation and collaboration.
• Strategic goals are hindered by a lack of alignment to business and data strategy.

Poor Rapport: Lack of clear communication between analytics and business teams can lead to misinterpretations or unmet expectations about what the data represents or how it should be used.

Impacts:

• Business teams will become or remain siloed if there is no relationship or trust with the Analytics team.
• Missed opportunity for collaboration and innovation to support a data driven organisation.

Rebuilding Trust with Data Products

By addressing the root causes of mistrust in data, organizations can foster stronger relationships between analytics and business teams, enabling more effective collaboration and trust in analytics. This helps to establish robust data-driven decision-making throughout the organisation, unlocking innovation and the foundation of a healthy data culture.

What’s a Data Product?

Data Products are the foundational output of the analytics team, that support activities within the business, while also providing a collaborative framework on which to build trust. In its simplest form, a data product is one of more curated tables, typically exposed through a semantic model (dataset), that can be used to provide:

• Data Exploration – Data Analysts investigate and visualise data to understanding it’s characteristics and patterns.
• Data Experimentation – Data Scientists experiment with data and test hypotheses.
• Centralised Reporting (traditional BI) – Reporting is build according to end user specifications by the centralised analytics team, and pushed out to the business for consumption.
• Self-Service Analytics – Business users create their own insights or utilise the data product to answer their own questions.

A Data Product can provide value to different personas within the business, allowing them to utilise their skillset to interact with the data product at the appropriate level.

Data Product Design

Data Product design enables analytics teams to give the business an opportunity to explain their requirements, increasing understanding for both parties, and enabling the business to contribute to, and take responsibility for, the analytical outputs that will support their decision-making and efficiencies. Through this approach both teams agree that a Data Product is viable, feasible, usable and most importantly, valuable.

SunBeam

Data Product design is achieved through the SunBeam framework, Advancing Analytics’ method for gathering data modelling requirements for analytical data models which uses the strengths of the analytics and business teams to enhance the data modelling experience for all stakeholders. SunBeam (Blog - An Introduction to SunBeam). SunBeam provides three pillars of collaboration that underpin a new way of working between analytics and business teams.

Requirements Gathering

Through workshops with a focus on collaboration, the analytics team leads the business through deep dives in to events that occur within their domain, creating a ‘map’ or model of what is measured and how it is contextualised (metrics and dimensions). This then forms that basis of what is expecting to be physically built in the analytics platform.

Figure 1 - SunBeam Model illustrating the metrics and dimensions required to understand the customer sales journey

User Questions

Upon creation of a SunBeam model, a great way to validate its value is to capture that questions that the business would have of the data product. In Figure 1, we have a model relating to Customer buying a product. To validate this model, and provide some tangible understanding for business stakeholders, we can interrogate it’s ability to answer user questions. The business may ask questions such as:

• What is the Margin % of Product Category ‘shoes’ YTD and YoY?
• Which Sales Person has the highest Revenue per State?
• Which customer age band has bought the highest quantity of nerf guns in the fiscal quarter?

All user questions can be answered with the Data Product, providing confidence that what is proposed is going to provide value to the business.

Delivery Agreement

A delivery agreement is an opportunity for the analytics team and the business to assign ownership, understand responsibilities, enable individuals to utilise the Data Product and therefore support a successful launch and maintenance of the Data Product. The delivery agreement is a more official way to share responsibility for the Data Product, fostering collaboration and underpinning the iterative process that SunBeam and Data Products require. The delivery agreement can cover:

Figure 2 - Data Product Delivery Agreement - assigns clear context and ownership

Maintaining Trust

Through our use of collaborative and iterative frameworks like SunBeam, and shared responsibility clearly defined through the Data Product Delivery Agreement, the Analytics team and business teams throughout the organisation can learn to build a relationship, and ultimately support each other in achieving shared business objectives.

As Data Product Design and Implementation matures within the organisation, so can the processes that help prevent the erosion of trust, and support shared success.