From raw tables to mobile UI: how Kiếm Tiền Chợ shortens the path to signal clarity.

I.The raw table problem

Our first internal tool was a table. It was accurate, dense, complete — and unusable on a phone during market hours. Every column was defensible on its own, but reading twenty of them to answer a single question (is this worth my attention right now?) is a design failure, not a data problem.

The redesign started from that question instead of from the schema. Rather than asking what data we have, we asked what decision the reader is about to make and what the shortest honest path to it looks like.

II.What we kept and what we cut

Cutting is the hard part, because everything in the table matters to someone. The rule we settled on is simple: the first screen answers the decision, and everything else moves one tap away instead of disappearing.

• On the first screen: the ticker, the direction, the conviction level, and the one-line reason.
• One tap away: the full report, the score breakdown, and the historical context.
• Out of the default view: intermediate calculations that explain the model but not the decision.
• Never removed anywhere: the risk level and the timestamp, because a signal without a date is a rumour.

III.Designing for a decision made in thirty seconds

Most users open the app between other things — a commute, a break, the ten minutes before the open. The interface has to stay readable at arm's length, with one thumb, in bad light.

• One primary action per screen, sized for a thumb rather than a cursor.
• Numbers aligned and formatted identically everywhere, so a glance is enough to compare two ideas.
• Colour reserved for direction and risk — never decoration, so it keeps its meaning.

IV.Speed is part of the content

A signal that arrives late is a different signal. We treat load time as an editorial constraint: the list renders first, charts stream in afterwards, and nothing blocks the first meaningful read.

The same applies to the data itself. Where a value cannot be fresh we label it as of a time rather than presenting a stale number as though it were live.

V.The flow, end to end

1. The model produces a ranked list with a conviction score and the drivers behind it.
2. The research desk reviews the list, discards what does not survive scrutiny, and writes the reasoning.
3. The app delivers the short form first: ticker, direction, conviction, and a one-line thesis.
4. The full report — score breakdown, risk level, and history — stays one tap away for anyone who wants it.

VI.What we are building next

The next round of work is about continuity: making it obvious what changed since you last looked, so returning after two days does not mean starting over. Clarity is not a redesign you finish, it is a constraint you keep re-applying.