{"slug":"dashboard-disease","meta":{"title":"Dashboard Disease","slug":"dashboard-disease","kind":"anti-pattern","summary":"Everything becomes a dashboard — numbers, charts and tiles — even though the people opening it came to do work, not to look at work.","problem":"The home screen is six KPIs and three charts. Every person who opens it reads none of them and clicks through to the list they actually needed. The most visited screen in the product answers a question nobody asked.","family":["orient","scan"],"principles":["surface-dont-bury","progressive-disclosure"],"status":"stable","visibility":"public","related":[{"title":"Inbox-First Application","url":"/patterns/inbox-first-application","summary":"The alternative front door — open on what needs a person, not on a summary."},{"title":"Choosing the Right Chart","url":"/cookbook/choosing-the-right-chart","summary":"For the dashboards that ARE warranted — which chart answers which question."}]},"body":"## How you get here\n\nDashboards are requested by the people who commission software and used by the\npeople who do not. A manager asks for visibility, the request is entirely\nreasonable, and the result is placed at the front door because that is where\nimportant things go.\n\nThen it spreads. Every section gets a summary strip. Every list gets tiles above\nit. Each addition is defensible and the cumulative effect is a product where the\nfirst screen of everything is a report.\n\n## How to recognise it\n\n- The home screen has no action on it.\n- Analytics show people leaving the landing screen within seconds, always\n  towards the same list.\n- Nobody can say what decision any given number would change.\n- The numbers are exact where nobody needs precision, and absent where somebody\n  does.\n- Charts have three data points.\n- Asked why a tile is there, the answer is \"so people can see it\", not \"so\n  people can decide something\".\n\n## Why it fails\n\nA number is only useful if it **changes what someone does**. Most dashboard\ntiles fail that test: revenue this month is interesting, and the person looking\nat it has no lever attached to it on that screen.\n\nMeanwhile the actual work — the seven things waiting on this person — is one\nclick away and invisible. So the product's most prominent surface is spending\nits position on the smallest audience, on every visit, for everyone.\n\nThere is a second cost. Because a dashboard is never finished, it provides no\nsense of completion. Software people work *through* feels different from\nsoftware people look *at*, and the difference shows up in whether they open it.\n\n## The way out\n\n1. **Ask what decision each number changes.** If none, it is not a metric, it is\n   trivia. Move it to a report someone visits deliberately.\n2. **Put the work first.** Most internal products should open on\n   [what needs the person](/patterns/inbox-first-application), with the numbers\n   reachable rather than resident.\n3. **Attach an action to any number that survives.** \"$21,400 overdue\" should be\n   a link into the filtered list, not a tile.\n4. **Do not chart three numbers.** Print them.\n5. **Give managers their own page.** The reporting need is real; it just is not\n   the same screen as the operator's front door, and pretending otherwise serves\n   neither.\n\n## The distinction worth keeping\n\nThis is not an argument against dashboards. A dashboard is the right screen for\nsomeone whose job **is** monitoring — an on-call view, a live operations wall, a\nweekly review. The disease is not having one; it is making it the default answer\nto \"what should this screen be\", and putting it in front of people whose job is\nsomething else.\n\n## The test\n\nWatch someone open the product. If the first thing they do is navigate away from\nthe landing screen, that screen is a toll, not a home — and it is being paid by\neveryone, every day.\n"}