"Pretty soon websites will be irrelevant. You will simply ask your assistant to perform most tasks, like find information, place orders for products, and make payments with just your voice."
This was the dominant narrative for smart speakers back in 2016. While the devices became common, the bold expectation of a "voice commerce" revolution never materialized. The reason is a fundamental principle of user experience: the interface must match the task.
High-Information Task, Low-Bandwidth Interface
Voice is a low-bandwidth, sequential interface. Shopping, on the other hand, is a high-bandwidth, parallel task. When you are researching and buying a new product, you are scanning dozens of options, comparing specifications side-by-side, and visually verifying details. A voice-only interface, where you must listen to a list of options one by one, is incredibly inefficient for this.
The Exception That Proves the Rule
This isn't to say voice transactions never happen. But when they do, they are almost exclusively for repetitive orders of known supplies. A user might say, "Alexa, re-order my usual brand of paper towels."
This is the exception that proves the rule. This is not a discovery or decision-making process. The hard work of the initial purchase—researching brands, comparing price-per-sheet, reading reviews, and entering payment information—was done on a visual interface, likely a website or mobile app.
The voice command is simply a convenient, low-friction shortcut to repeat a decision that has already been made. It is an API call, not a shopping experience.
The Echo of Hype in 2025
This is highly aligned with today's AI hype. We are hearing the exact same narrative, just with a new subject: "In the future, we don't need websites; users will just ask their voice or chatbot."
This logic ignores the lesson we just learned. A conversational interface is a poor substitute for a graphical user interface when a user needs to process a large amount of information to make a considered decision. Asking is not the best interface unless you prefer to do all of your complex work by making phone calls instead of using a computer.
Before you invest in replacing a perfectly good, high-information interface with a trendy, low-information one, ask the hard questions:
- Is this a task of discovery and comparison, or a task of simple, repetitive execution?
- Does this new interface actually make the core task easier, faster, and more reliable for the user?
We don't need to reinvent the wheel to move forward. Optimize for possibilities, but don't discard the basics.