Most people can say roughly what they spent last month. Almost nobody can say where their attention went. Hours disappear into feeds, autoplay and half-read articles, and unlike money there is no statement at the end of the month showing where it all went.
That missing statement was the starting point for this build. If attention is genuinely finite — and a day only holds so many usable hours — then a product that pays people for attention should behave less like an open tap and more like a budget: a set amount, visible spending, and a clear record of what each hour bought.
Designing against the infinite tap
The default pattern in attention products is an endless supply: keep scrolling, keep watching, keep tapping, and the meter quietly runs in the background. That pattern is good for platforms and bad for the person, because nothing ever signals "enough."
The alternative built here has three deliberate constraints:
- A daily amount the person sets themselves, not a target the platform pushes upward.
- A running ledger, visible at any moment, showing minutes given, what they were given to, and what they earned for them.
- A hard stop when the budget is spent. Not a warning, not a streak to protect — a stop.
Each constraint sounds simple. Each one fought the rest of the design.
What worked
The ledger changed behaviour first. Once people could see a plain list — this hour went here, that twenty minutes went there, and this is what came back — the empty sessions stood out without any lecturing from the interface. Counting honestly did more than any reminder or nudge. Several patterns only became visible in the ledger: late-night sessions that earned almost nothing, and short morning sessions that were worth more than the long ones.
Setting the price and the cap in the same place also worked. In an earlier version of this project, pricing attention and limiting attention were separate decisions on separate screens, and most people only ever made the first one. Putting the daily cap next to the price turned it into one question: what is an hour worth, and how many of them are for sale today? That is a budget question, and people answered it more carefully than either half alone.
The hard stop, counterintuitively, did not hurt earnings. Sessions got shorter and more deliberate. When the supply is capped, low-value offers get skipped instead of absorbed, so the average value per hour rose even as total hours fell.
What broke, or nearly did
The first ledger was too detailed and it broke trust in the opposite direction. Minute-by-minute breakdowns with timestamps for every interaction read less like a bank statement and more like surveillance — even though it was the person's own data about themselves. The fix was aggregation: totals by day and by category, with detail available on request rather than pushed into the main view. A budget should feel like a statement, not a tracking log.
The second break was the midnight reset. A budget that resets at midnight rewards staying up late to spend the last of it and start the next day's early — the exact behaviour the cap was meant to prevent. Moving the reset to a time the person chooses, defaulting to their usual wake-up time, removed most of that gaming without adding rules.
The third problem is still only partly solved: attention spent is easy to count, attention quality is not. Two hours can look identical in a ledger and be worth completely different amounts — one focused, one half-distracted with a second screen running. The current design does not pretend to score quality. It records time and earnings plainly and leaves the quality judgment to the person reading their own statement. Adding an algorithmic quality score was tested and removed: it was confident, precise, and mostly wrong.
Lessons for builders
Give the resource a statement. Any product that consumes a finite personal resource — time, attention, data — owes the person a readable account of where it went. If the statement would embarrass the product, that is information about the product.
Put the limit where the price is. Limits buried in settings do not function as limits. A cap chosen at the same moment as the value decision becomes part of the decision.
Let a stop be a stop. Soft warnings, streaks and "are you sure?" dialogs all negotiate with the limit until it means nothing. A budget that ends the day cleanly respects the person who set it that morning.
Count what you can defend, and label the rest honestly. Time and money can be counted exactly. Quality cannot, and a dashboard that fakes that precision will eventually be caught by its own users comparing notes.
The project behind these lessons is Introworth, where people set their own price for their attention and keep a clear ledger of what it earns.
Top comments (0)