Building a small daily productivity dashboard without adding noise

A start page becomes noise the moment it starts showing things because they are available rather than because they are needed. Ten cards of accurate information is worse than three, because you stop reading any of it and the page becomes wallpaper you have to look past. Building one that keeps working means being unreasonably strict about what earns a place, and the test is not interest, it is whether the card leads to an action today.

Admit a card only if it can fail your day

Go through every candidate and ask what you would do differently if this card showed an unexpected value. If the honest answer is nothing, it is a fact rather than a signal and it belongs somewhere you visit on purpose. A card showing that a deadline is closer than you thought changes your morning. A card showing a cumulative total that only ever rises does not, however satisfying the line looks.

Prefer cards that can be wrong in a way you would notice. Something with a state, an expiry, or a threshold gives you the possibility of surprise, and surprise is the only thing a glance is good for. Anything that is the same every time you look becomes invisible within a week, and once it is invisible it is still consuming the attention budget of the cards next to it.

Put the action inside the card that raised the question

A status that tells you something is due and offers no way forward creates the exact pause you built the page to avoid. Every card should carry the one link or control you would reach for next, sitting in the card rather than in a menu at the side. The point is that noticing and acting happen in the same movement, because the gap between them is where things get postponed.

One action per card, and make it the common one. Cards with four buttons are a decision in themselves, and you will hesitate over which is right rather than doing anything. If a card genuinely has two equally likely next steps, that is usually a sign it is answering two questions and should be split.

Order by the shape of the day

Lay the page out along the day rather than by category. What you need before starting goes first, what you check mid-morning next, what you use to wrap up last. Grouping by type, so that all the numbers sit together and all the links sit together, looks tidier and reads worse, because your actual need is time-shaped and you end up scanning across groups to assemble a moment.

Set a hard cap on the number of cards, somewhere around five or six, and enforce it as a swap. Nothing new goes on the page unless something comes off. Without the cap you get the familiar drift where the page grows a little every week, each addition perfectly reasonable on its own, until the whole thing is a wall and you have gone back to using search and bookmarks instead.

Measure the page by what you did, not by how it looks

After a couple of weeks, ask which cards actually changed a decision. The strong tell is a card you look at and then immediately act on. The weak tell is a card you find yourself explaining, because a card you have to interpret is doing work your brain should not have to do at eight in the morning. Anything in the second group either needs a plainer presentation or needs to go.

Watch out for the card that exists to justify the page. Most people who build a dashboard end up with at least one element that is there because a dashboard ought to have one, often a chart or a clock. Deleting it costs nothing and immediately makes the remaining cards easier to read, which is the whole argument for keeping the page small.

Before you commit

• Limit the dashboard to today’s decisions
• Remove cards that only explain themselves
• Keep action links close to the status they affect

Admit only cards that can surprise you, keep the next action inside the card, order the page along the day, and cap the count so additions require removals. What is left is small, slightly plain, and still being used in a month, which is more than can be said for most start pages built in an enthusiastic afternoon.