Project summary
Scope
Redesign flexible reporting so users can answer their own commercial questions across metrics, dimensions, filters and time periods.
Challenge
The existing Tableau experience was slow and difficult to configure. At the same time, moving reporting in-house risked recreating the same complexity in a different interface.
Solution
I designed Freeform around how users form a reporting question, setting deliberate boundaries on flexibility and creating patterns that could scale as the reporting platform grew.
7.5×
Faster load time
~89%
Fewer clicks to a report
~90%
Fewer data requests
Reframing Freeform
Design around the question, not the configuration
Tableau was too slow to update the view on every change, so all selections sat behind an Apply button. Users often made their changes, forgot to apply them, and assumed the report had already updated. Errors, such as metrics that couldn’t be shown together, only appeared after clicking Apply and waiting for the view to reload.
Decision
Performance was the first priority of the redesign. Once Freeform was fast enough. I then structured the flow around how users form a reporting question: What do I want to measure? How do I break it down? When? Analyse. Rather than asking users to configure everything up front, I let the report build step by step, so problems are visible as soon as they happen.
Outcome
Users build reports through 1 guided flow with no Apply step, and see the result after every change instead of waiting to find out if it worked.

Designing the entry point
Help users get to something meaningful faster
Stakeholders worried an empty screen might feel broken, but loading a default report would assume what users wanted to see and add unnecessary load time. At the same time, starting from 19 metrics plus slicers and filters could feel overwhelming for less experienced users.
Decision
I turned the entry state into a guide rather than a blank screen. Returning users see their Saved Views first, so they can jump straight back into reports they already use. If they don’t have any, the empty state suggests commonly used metric, slicer and filter combinations as a starting point, while still leaving the option to build from scratch.
Outcome
Getting to a meaningful report dropped from 9 clicks to 1, while avoiding unnecessary data loading and still giving experienced users full control over how they start.

Setting boundaries
Give users flexibility without letting reports become unmanageable
User interviews showed that people often wanted to compare several metrics at once, but very large combinations quickly became harder to interpret. Performance was also one of the most consistent pain points in Tableau, so unlimited configuration would work against one of the core goals of the redesign.
Decision
I capped each report at 5 metrics. This covered the comparisons users commonly needed, while setting a clear boundary around readability and the amount of data the platform had to load and render. Rather than maximising flexibility, I deliberately balanced user need, readability and performance.
Outcome
Users can compare up to 5 related metrics in 1 report, reducing the need to build separate analyses while keeping report complexity and load time under control.

Designing for exploration
Let users decide when a chart becomes too complex
Our initial idea was to remove the chart once it showed more than 10 results, assuming it would become unreadable.
User interviews challenged that. Some users said they only had 11 or 12 accounts they wanted to break down, and could still want to read the chart. A fixed cut-off would remove a useful view just because they were slightly over the limit.
Decision
I challenged the fixed 10-result cut-off. User interviews showed that readability depends on context. Instead of automatically removing the visualisation, I kept it available and let users narrow the data with filters when needed.
Outcome
0 fixed chart limit, users can explore the full result set and filter it down when needed, rather than losing the visualisation after an arbitrary 10-result threshold.

Comparing over time
Show the comparison users expect by default
User interviews showed that senior users rarely look at the current year in isolation — they usually want to understand performance against the previous year.
Decision
I defaulted the time period to current year + previous year, so once a metric is selected, users immediately see a useful comparison. They can still change the timeframe when they need a different view.
Outcome
Users get 2 years of context by default, removing the extra setup needed to create the most common year-on-year comparison.

Extending with AI
Use the AI experience users already know
My first idea was to add an AI chat to Freeform, so users could ask questions like “How is this account performing?” instead of building the report themselves.
But that would only help users who were already in Freeform. Most reporting questions come up during other work, such as preparing for a client call or checking a team’s numbers. The company also already had an internal AI agent that users opened from anywhere in the product. A second chat inside Reporting Universe would give users two places to ask the same question.
Decision
Rather than building a separate AI experience, I chose to expose Reporting Universe through MCP and connect it to the existing agent. Users can ask their question wherever they are working. The agent uses Freeform’s metrics, slicers, filters and date ranges to answer, and can open the same view in Freeform when users want to look deeper. This made reporting part of users’ everyday workflow, rather than another destination they had to go to.
Outcome
1 AI entry point, available across the product. Users get reporting answers where they already work, and we didn’t have to build and maintain a separate chat.

Measuring the impact
Make reporting faster, simpler and easier to return to
The redesign wasn’t just about adding more flexibility. The biggest opportunity was reducing the effort required to get from a reporting question to something useful.
Decision
Across Freeform, I focused on removing unnecessary setup, setting clear limits where complexity started to hurt usability, and keeping the experience fast enough that users could explore data without hesitation.
Outcome
~7.5× faster report loading, 100% adoption onto Reporting Universe, and common reports reduced from 9 clicks to 1 from the entry screen.
Reflection
Flexibility needs clear boundaries
The hardest part of Freeform wasn’t giving users more control, it was deciding where that flexibility should stop, where the product should guide them, and where users should stay in control.
Decision
Throughout the redesign, I used user behaviour and performance constraints to challenge assumptions: when to show guidance, how much configuration to allow, when not to hide data, and when to reuse existing product capabilities instead of building new ones.
Outcome
Freeform became a more focused self-service reporting experience: faster to enter, easier to explore, and flexible enough for different analysis needs without recreating the complexity of the old Tableau experience.


