Dashboards, texts or prompts?
How to present data insights.
For years, the go-to response to complex business data has been to build bigger, denser and more complex dashboards. But as cognitive overload rises and conversational Ai redefines user expectations, designers must ask a fundamental question: Should this insight be a chart, a textual explanation, or a prompt? In this article I will present my approach to transforming raw metrics into effortless action.
Linnea Forss
Senior service designer
Date published:
Length:
5 min. read
We have access to more data than ever before. Business leaders use customer data to optimize experiences and increase sales, and in our everyday lives, we can measure, track, and analyze everything from our electricity consumption to pension savings.
But more data does not automatically equal more clarity.
From working on client projects with “numbers as outputs”, I have often seen users presented with high volumes of data in complex graphs or unintuitive self service solutions with sliders and calculations that leaves them confused or calling customer service instead.
A study from McKinsey & Company shows that despite massive investments in self-service analytics and dashboards, up to 64% of users abandon automated visual interfaces when faced with complex decisions, citing information overload and uncertainty about how to interpret the figures.
— McKinsey & Company (Reimagining Digital Customer Experience)
As designers, we’re responsible for giving meaning to data and choose the right communication medium —whether that is a dashboard, a text message, or a response to a prompt.
Begin with the end
Before diving into data visualization, we should start with the end goal of the user.
What are they actually trying to achieve? Are they exploring broad trends, looking for a quick status update, or trying to make a specific, high-stakes decision?
Dashboards: Exploration and pattern recognition
Visual dashboards excel at exploration and pattern recognition.
A well-designed visual dashboard allows the brain to process large amounts of information simultaneously. Graphs and heat maps are ideal to spot visual anomalies and spatial trends without knowing exactly what you are looking for in advance.
An example: Visualizing physical foot traffic
Working with spatial analytics and foot-traffic platforms showed the power of visual exploration.
When analyzing customer movement across a physical store, a spreadsheet of coordinates tells you absolutely nothing. However, when that movement data is designed as a visual heat-map or a sequential flow, an operational manager can see an unexpected "cold zone" in five seconds.
The dashboard did not answer a predefined question, but enabled the user to explore a physical or digital space visually and discover an opportunity they didn’t know existed.
Visualizing foot-traffic movement as a heatmap allows store managers to spot unexpected spatial bottlenecks and unengaged zones in seconds.
Words: Explanation and proactiveness
While dashboards are great for visual exploration and at-a-glance overviews, design is equally about knowing when visual charts create unnecessary cognitive load.
In some scenarios, the best data visualization is no visualization at all. Plain language can be just as good - or even better. In cases where users face high-stakes personal decisions or when the math gets very complex, charts often confuse more than they clarify.
An example: When the best visualization is no visualization
In a project for a pension provider, the initial idea was to design visual graphs and multi-variable sliders to explain future retirement income.
But user tests showed that complex visual graphs left people feeling overwhelmed and uncertain. Besides struggling to decode the charts and pension jargon, they did not trust their own interpretation in an area that had economic consequences. Instead, they wanted clear, direct answers from the source.
The solution was to replace the charts and calculators with a simple written narrative, sent in an SMS to users whenever a change in their life impacted their pension.
“We can see [event]. This means [impact for the specific user]. Here’s what to do [action].”
Translating raw numbers into plain language eliminated the cognitive friction of analyzing figures and eliminated the risk of getting them wrong.
Prompts: On-demand answers and conversational depth
With the rise of large language models and conversational AI, another design paradigm has emerged: the prompt.
When a user has a hyper-specific question, forcing them to navigate through a complex dashboard, adjust date pickers, or isolate columns creates friction. This is where prompts are great.
A prompt allows users to skip UI navigation entirely and ask for direct status updates or cross-variable analysis. For example:
• "Based on weather this afternoon and historic foot traffic in the store, how many extra staff members should we call in to cover the peak between 14:00 and 17:00?”
• "If I switch to a 4-day workweek for 3 years and plan to retire at age 68, how much should I increase my voluntary contribution to get the same net monthly payout as now?
Instead of forcing you to check three different systems, it does the math instantly by combining weather forecasts, shift schedules, or complex pension rules into one clear answer. A prompt can quickly test out different "what-if" ideas without needing to merge spreadsheets or call customer service for help.
Prompts (or conversational UIs) are not a total replacement for dashboards. While an operational manager still needs a visual dashboard for real-time status monitoring, they will use a prompt when they have a targeted question and need an instant answer.
