Data-driven design grounds design decisions in measured data (A/B testing, analytics, real behaviors) rather than the designer's intuition alone.
The promise is super healthy: moving past the opinion of the highest-paid person in the room, and deciding based on proof rather than opinion. Testing two versions, keeping the one that performs best. When well-executed, this culture has corrected countless ego-driven mistakes and made design truly accountable for its impact.
But it has a limit that has since become famous. In 2009, designer Douglas Bowman left Google, sharing that he had been asked to test "41 shades of blue" to choose a link color: when everything is decided by measurement, judgment no longer has a seat at the table. This is the symptom of a much deeper challenge.
Data optimizes what already exists; it doesn't invent what has yet to be imagined. An A/B test helps you climb the hill you are already on (the local maximum), but it will never leap to a higher hill located elsewhere (the conceptual jump). And Goodhart's law warns: "when a measure becomes a target, it ceases to be a good measure." Optimizing click-through rates can easily degrade the user experience while temporarily boosting short-term figures.
Data is fantastic at answering "which of these two options?" and terrible at answering "what is the right question?". Inspired intuition, fueled by field research, experience, and deep domain expertise, proposes the hypotheses; data then helps decide between them. One without the other is either blind or sterile.
Our methodology: data clarifies, intuition proposes, and field experience arbitrates. We are just as skeptical of "gut feeling" as we are of absolute measurement. A metric is a spotlight: it illuminates one area while leaving the rest in the dark. The designer’s true role is to understand what the measurement cannot see.
What is happening elsewhere: cognitive biases (on both the intuition and data-reading sides), insights (qualitative data), user testing, creativity (the leap beyond the local maximum), and artificial intelligence (which industrializes data).
References: D. Bowman, departure from Google and the "41 shades of blue" episode (2009); Goodhart's law.
Rémi
Gréau
Director of Design
Related Articles
Insights: Data is not the same as understanding
An insight is neither a simple data point nor a mere observation: it is the hidden "why" that unlocks true action. One can accumulate data endlessly without ever producing a single, powerful insight.
Information Design: Elevating Data into Meaningful Insights
Information design beautifully transforms raw data into clear understanding (as championed by Wurman and Tufte). Yet, elevating data into true insight is an art of interpretation: clarifying complexity without compromising its integrity is where real design differentiation shines.
Cognitive biases: to empower, or to exploit?
Cognitive biases are shortcuts in judgment (Kahneman). Design can leverage them to either empower or exploit. And as creators, we too carry our own biases.
User Testing: Reassuring Ourselves, or Truly Learning?
User testing observes real people to pinpoint exactly where they get stuck (Nielsen's rule: five users are more than enough!). Its true and exciting challenge: are we testing simply to reassure ourselves, or to pioneer new learning?
Generative Design: Who True Creator?
Generative design empowers systems and AI to craft innovative shapes and solutions based on your constraints. Since the machine does the producing, who is the true creator? The designer, who strategically guides and selects the winning concepts.
Our Associated Offers
User and client testing
Evaluate a service, product, or device in real-world conditions to measure its effectiveness and desirability.
Discover →
Digital Product UX Design
Aktan designs digital products that are useful and simple, based on the real usage and journeys of your users.
Discover →
AI Empowering Operational Excellence
Design services where humans and AI collaborate effectively to maximize value.
Discover →