Introduction - The Myth of the "Average User"

The other day, I asked my smart tv to play "Maalai Neram", a Tamil song I love. It paused for a second before confidently replying, "Okay, playing 'My Little Love' now". I tried again, enunciating clearly. It offered me a different English song. After the third try, I gave up and just used the remote, wondering how a device so "smart" can be so completely dense.

While voice recognition has gotten remarkably good at parsing different English accents, it often hits a wall when you cross linguistic or cultural boundaries. This wasn't a random glitch. The device wasn't designed for a world where "Maalai Neram" is as valid a request as a Billboard hot 100 hit (I hope they are still a thing :D). It was designed for a mythical "average user" whose cultural and linguistic landscape is limited. By chasing this imaginary center, we create products that, at best, serve no one perfectly, and at worst, make brilliant technology feel exclusionary and sometimes, a complete idiot.

This essay argues for a more powerful and effective approach. The solution isn't to abandon data-driven design, but to challenge its oversimplification. It's about recognizing that our "average user" is a myth, not because a target user doesn't exist, but because they exist in countless variations. True innovation comes from designing for the rich diversity within our target demographic: the expat, the new parent, the user with low vision. By building for these real people, we don't just create functional features; we uncover our own blind spots, challenge our assumptions, and build more robust, creative, and ultimately more human-centered products for everyone. And frankly, it's how you build a tv(voice assistants) that actually plays the music you want to listen to.

The Problem - The Hidden Costs of Our Blind Spots

A smart tv getting your song wrong is annoying. But what happens when the same design philosophy is applied to products with higher stakes? The consequences quickly escalate from inconvenient to actively harmful. These aren't just edge cases; they are massive, costly failures hiding in plain sight.

Take the Dutch childcare benefits scandal, a national disaster driven by a single piece of software. The government deployed a self-learning algorithm to identify potential fraud, but its "risk indicators" were deeply flawed. Factors like having dual nationality were weighted heavily, meaning the system learned to equate an immigrant background with a higher risk of fraud. As a result, tens of thousands of families were wrongly accused, forced into crippling debt to repay benefits they were entitled to, and had their lives destroyed. This wasn't a bug; it was discrimination automated and executed with ruthless efficiency.

This problem isn't limited to the public sector. When Amazon built an AI tool to streamline hiring, it taught the system using a decade's worth of its own resume data. Because that historical data was dominated by male applicants, the AI developed a systemic bias against female candidates. It penalized resumes that included the word "women's" and downgraded graduates from all-women's colleges. Amazon had to scrap the entire project. Beyond the development costs, the true loss was the immense talent the algorithm was designed to miss. When we design for a narrow average, we are choosing to inherit all of the biases of the past and code them into our future.

The Solution - Inclusive Design as a Product Strategy

Okay, so we've established that designing for the "average user" could result in products that alienate other potential users. So what's the alternative? While business realities of time and money often mean we must start with a primary target user, the solution is to build with responsibility and foresight into the process from day one. This is inclusive design. It's not about being nice; it's about being smart.

This strategy begins by enriching our understanding of that core user. Let's take a common persona: "Alex, a skilled software developer". Alex isn't a monolith. What if he's an expat living in Germany, navigating a complex interface in a language he's still learning? This is a very common scenario. Alex is me :) For years, many online banking and insurance apps in Germany were available only in German, creating significant barriers for a large part of their customer base. While many have now added English options which is a positive step, some core services often remain German-only for legal reasons. By considering this "stress-case" persona from the start, we are forced to think about clearer navigation, simpler language, and better user guidance. These aren't just features for non-native speakers; they are improvements that make the product less intimidating and more intuitive for everyone, including native German speakers.

True accessibility is much broader than a single feature. It isn't a one-size-fits-all solution; it’s context-dependent. For example, the needs for a complex data visualization tool are completely different from those for a ride-sharing app. For the data tool, accessibility might mean high-contrast modes for charts, keyboard shortcuts for filtering data, and screen reader compatibility that clearly announces data points and trends. The focus is on complex information clarity. For the ride-sharing app, the priorities are different: large, easy-to-tap buttons for one-handed use, clear audio cues for ride status, and a simple map interface for quick glances in a distracting environment. Understanding this distinction and building responsibly for your specific target demographic is the core of a mature product strategy.

The Unforeseen ROI - How Inclusive Features Become Mainstream

Here’s the beautiful, almost magical thing about designing inclusively: when you solve a problem for a specific group, the solution almost always ripples outward to benefit everyone. Consider video captions, they were created as an essential tool for the deaf and hard-of-hearing community. Today, a huge majority of people watch videos on social media with the sound off, relying entirely on captions. They are now an indispensable feature for anyone on a noisy train, or in a quiet office.

The same is true for "Dark Mode". Initially introduced as an accessibility feature to reduce eye strain for people with light sensitivity and certain visual impairments, it has become one of the most popular customization options on our phones and apps. Millions of users who have no visual impairment use it to save battery life, reduce glare at night, or simply because they prefer the aesthetic.

These features designed for a specific need with inclusion at the forefront became mainstream expectations. The simple language we use for a non-native speaker is clearer for everyone. The one-click checkout we design for someone with limited mobility becomes the preferred option for a busy parent trying to buy something with one hand while holding a baby. While capturing a new market segment is a clear benefit, the true ROI of inclusion lies in discovering unforeseen innovations that make your core product more usable, more resilient, and more valuable to every single user.

Conclusion - A Call to Action for Product Builders

In a world driven by data, it's easy to believe that numbers tell the whole story. We build our products on metrics and analytics, trusting them to guide us. But data is a mirror reflecting a complex and often biased world. The real pitfall isn't using data, but oversimplifying it into a single, monolithic "average user", which erases the rich diversity within our own target markets.

The solution isn't to abandon data-driven personas, but to enrich them. Instead of one "average", we should build a set of well-researched personas that represent the true breadth of our user base. What does the product experience look like for the power user, the brand-new user, the user with accessibility needs, or the user from a different cultural background? Building for these multiple "averages" is the essence of a resilient and responsible product strategy.

This is how we move forward. We use data cautiously, we challenge its hidden biases, and we celebrate the complexity of our users by representing them authentically in our design process. So the next time you build a user persona, don't stop at one. Try asking, "Which averages matter?" and "Whose story is our data not telling?" The answers will lead to not just more inclusive products, but fundamentally better ones.

References

Boffey, Daniel. "Dutch government resigns over child benefits scandal". The Guardian. 15 Jan. 2021. 23 July 2025. https://www.theguardian.com/world/2021/jan/15/dutch-government-resigns-over-child-benefits-scandal.

Dastin, Jeffrey. "Amazon scraps secret AI recruiting tool that showed bias against women". Reuters. 10 Oct. 2018. 23 July 2025. https://www.reuters.com/article/us-amazon-com-jobs-automation-insight/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-women-idUSKCN1MK08G.