A watermelon project is green on the outside and red on the inside. It does not hide by lying. It hides because every single report is defensible on its own. The date moved by four days. The supplier has promised to come back next week.
Read one month and there is nothing to see. Read twenty-four of them in a row and there is.
So the wall is the point. Every project is a row, every month a small square, twenty-four of them across. A thin line above each row shows how far the finish date has drifted since the first check-in. A band underneath shows how sure the project manager said they were. When the line climbs and the band goes dark while the squares stay green, that is a watermelon, and nobody who has sat through a portfolio review needs it explained twice.
Ten patterns
Ten signals read the history for what a single report cannot show. Take salami slippage: a finish date that moves six times in six months, never by more than a fortnight, five weeks lost in total, and the status never once turns amber. Every individual move was reasonable. Reasonable, six times in a row, is five weeks.
Then there is the eternal finish date, where the time remaining never changes no matter how much of it passes. Zeno would have recognised it. A baseline moved three weeks before the old one would have gone red. Confidence falling three periods running while the colour sits perfectly still. The same request for help raised in three check-ins while nothing improves.
Every flag is dated to the month the pattern first became visible, not the month somebody noticed. "It was there in October and we acted in April" is the sentence that changes a steering committee. It is not a comfortable sentence. That is rather the point.
The number that gives it away
Most of what the app needs is what a project manager already reports: the forecast date, the cost, blocked dependencies, open critical risks. One field is different. Every period it asks how sure they are that they will hit the date they just entered, on a scale of one to five, and it keeps the answer.
A project manager will defend a date in front of a steering committee and then, in a small field that nobody reads aloud, rate their own confidence at two. On its own that number is worth very little. As a series, sitting next to a status that has not moved all year, it is the most useful thing on the page.
What it will not do
There is no view that ranks project managers by how often their projects raise a flag, and there will not be one. Build that, and people stop answering honestly, and what you are left with is a very expensive way of generating green. When a flag is acknowledged the app asks what was decided, not who was responsible.
Have a play
It opens with demo data already loaded. Twelve invented projects over twenty-four months, built so each of the ten signals fires at least once, and so one project stays green the whole way, because a detector that shouts at everything is no use to anyone. Clear the demo data whenever you want to start on your own. It comes back if you leave the app completely empty, which is either helpful or stubborn depending on the day.
Underneath
It runs in the browser and keeps everything on your own machine. No account, no server, and once it has loaded it makes no network calls at all, not even for fonts. Put it on a phone and it works with the aeroplane mode on, although whether a portfolio review is best conducted at 30,000 feet is not for me to say. It speaks Danish and English, and guesses which one from your machine.
Ten years of reading status reports, turned into something that reads them for me.