B2B Data Product
One of NZ's main Food Safety & Auditing Companies wanted to provide their customers with a suite of Data Products to enable them to track, analyse, and improve on their auditing data.
The project was developed along 12 months, including multiple dashboard screens to help Quality Assurance and Food and Safety Managers across the country to check their auditing reports, points of improvement, and track their Food Safety standards performance over time.
This is a paid data product now in offer to their customers as a premium service alongside their auditing contracts.
*all figures, measures, categories and names displayed here are based on fake mock data and not in any way representative of real performance or information from the client's data.
Year
2025
Type
Client Project
Category
Dashboard, Data Product, Advanced Analytics, Interactive, Business Intelligence
Tools
Power BI, Figma, Snowflake
Client
Large Food Safety & Auditing company in New Zealand
Services
Data Design, Consulting, Training, User Experience, Development
Every complex data product development is a long list of compromises, and this was a great example of how challenges and compromises can still make for a great data product.
I joined the team a little later than ideal, after client interviews and market research had already been conducted. This made the project a little more challenging, since I didn't have direct contact with the client's themselves, having all usability decisions made in partnership with the Project Manager assigned to this development.
While having less information than ideal to kick start, we opted to make the process as agile as possible, so we could gather feedback directly from potential users.
The first step was to understand the business brand and how they wanted this data product to reflect that for their clients. The resulting process was a comprehensive design system, including specific branding guidelines for data visualisation. Colour palettes, typographic scales and combinations are only a small part of this work. The design system also includes options for layouts - still allowing for flexibility in design and placement of elements, but keeping spacing and flow consistent. It also includes navigation and UX guidelines, with clear instructions on how to use buttons, filters, breadcrumbs, and other page navigation cues. The idea was to have a set of clear, per-established rules to allow non-designers to maintain or further expand the data product, while maintaining it consistent visually.
this design system also includes components to allow the team to create backgrounds to be used in Power BI, to enhance the interface and experience of the dashboards, making them look more consistent, polished, less like a standard Power BI dashboard, and more like a robust data product.
It also includes components to help non-designers to create mockups and prototypes, to bounce potential ideas more quickly, before going into the final tooling for solutions. The idea of prototyping without data before jumping into development in Power BI was a new way of working for the team, but they quickly realised the advantage in both speed and alignment this could bring to the project.
The Design System was created in Figma, and I ran weekly training sessions teaching them how the tool works, as well as how to maintain the design system long-term.
The development was done in Power BI. The tooling decision was made based on their own internal expertise and what they would be able to maintain long-term. It was a challenging choice, since Power BI is not ideal for customised web interfaces, like their own requirements suggested they would need.
Multiple tricks later, they had a series of 3 dashboard suites, each made of several screens, displaying audit data from multiple perspectives, according to the audience personas defined during the discovery stage. Each dashboard dynamically populates data controlled via Row-Level security for each client separately, allowing for there to be the maintenance of just one group of screens, which would work for their entire portfolio of clients.
The biggest technical challenge was to build the data model in such a way as to allow for different clients to have their metrics and measures anchored to different Financial Years - a requirement we were only made aware of once we already had a robust prototype in place, half-way through the project. This meant re-building the data model taking into account that each potential Row-Level Security coordinated client access would have all of their calculations, data and visualisations display different financial years. This was a complex build, but in the end it works seamlessly, even if with quite a maintenance cost associated with the dynamic interactivity requirements.
The trickies dataviz challenge arose almost at the end of the project: The need for custom visualisations to display complex information. Often times, bar and line charts will do. But sometimes, the best and most effective way to display an industry specific insight to an expert audience, is to use a custom visualisation. Power BI is not widely known for allowing a lot of creative takes in their visualisation layer. And that's when DENEB comes to the rescue! Deneb is a third party visual for Power BI which allows the build of declarative dataviz. This means full control and flexibility to develop what is required.
The Deneb viz is a custom grouped jitter plot. It has two average lines - one for each group of sites and a national one, across the entire chart. These are important points of comparison to enable Food Safety experts to quickly understand, across potentially hundreds of sites, where they should focus. What sites have audit results above national averages? What groups of sites have their averages higher than others? This chart works as a prioritisation tool. It is also dynamic, so the user can click on either a group, or a specific site, and get more details to allow them to follow up directly with their regional or site managers.
All in all, this was a complex project, which ended on a high note. The data product is now available for clients who can subscribe to the service as a premium offer.
