SuperhumanCraft6 min readSelling speed as the whole product
An email client where the mouse is the slow path.
- The surface
- Processing a morning inbox: triaging with the keyboard, the command bar, snoozing, and the empty inbox screen.
- What the user wants
- I want to clear my inbox in one pass every morning instead of letting it become a background anxiety.
Onboarding as instruction
Superhuman built its early onboarding around a guided session with a person who taught the shortcuts, rather than a self-serve signup that dropped the user into an inbox.
Teach the skill the product depends on
The value only exists for a user who has the shortcuts in muscle memory, and nobody acquires that from a tooltip. Paying for human instruction is unscalable on purpose, because the alternative is a churned user who thought they bought an inbox. It also produced an unusually direct feedback channel during the early years.
The command bar
One key opens a searchable command list covering actions, people and settings, so any function is reachable without knowing where it lives.
A flat namespace over a menu tree
Menu hierarchies force the user to model the designer's taxonomy before they can act. A search box over every command replaces that model with a word they already have. It also means new features ship without renegotiating the navigation.
Latency as a design constraint
Superhuman markets a strict interaction latency target and holds features to it, with local search and prefetching doing much of the work.
Perceived performance is a feature, not an optimisation
Below roughly a tenth of a second an interface feels like an extension of the hand, and above it the user notices the software. Treating that threshold as a requirement rather than a nice-to-have changes architectural decisions rather than just polish passes. It also gives the team a single number to reject features against.
Triage verbs
Core actions are a single key each: mark done, snooze until a chosen time, reply, and move to a split of the inbox.
Small vocabulary, applied repeatedly
Inbox processing is one decision repeated a hundred times, so the cost of that decision dominates everything else. Four verbs is few enough to become automatic within a week. Snooze in particular converts a deferral, which is otherwise a source of guilt, into a scheduled object.
Split inbox
The inbox is divided into named sections by rule, such as messages from real people versus newsletters and calendar traffic, so different categories are triaged in different passes.
Batch by decision type
Switching between how do I answer this and do I even care about this is a context switch that happens silently and costs real time. Grouping by the kind of judgement required means the user makes one kind of decision at a time. The rules are the product's opinion about what deserves attention.
The empty inbox screen
Reaching zero reveals a full-bleed image and a calm end screen rather than an empty list.
peak-end rule: People judge an experience by its most intense moment and its ending, not its average.
People remember an experience by its most intense point and its ending, so the last thing an inbox shows determines how the chore is remembered. Ending on a reward makes tomorrow's session easier to start. It costs nothing to build and does most of the emotional work in the product.
Where the effort went
What not to copy
- The price sets the audience, and the audience sets the roadmap. A tool priced for executives will keep optimising the executive's morning, which limits how widely the design ideas can travel.
- Speed depends on aggressive local caching and prefetching, which means a large mail archive on a weak device or a poor connection gets the opposite of the promise.
- Keyboard fluency is an investment with no partial payoff. A user who only half learns the shortcuts has bought an expensive ordinary email client, and the product cannot tell the difference from the inside.
- Inbox zero is a philosophy the product enforces rather than offers. For someone whose work genuinely does not sort into done and snoozed, the interface keeps asking a question they cannot answer.
The takeaway
Find the last screen in a repeated task and spend disproportionate effort there.
Finished the teardown? Bank it and the day counts toward your run.
Where the principles come from
- Thinking, Fast and Slow (peak-end rule), Daniel Kahneman
- Response Times: The 3 Important Limits, Jakob Nielsen, NN/g
- Badass: Making Users Awesome, Kathy Sierra
- The Humane Interface, Jef Raskin
Written from public behaviour of the product, not from inside it. Interfaces change often, so treat the flow described here as of the time of writing and check the live product before quoting it.