The short answers to how Weather for Dads works, why it sometimes disagrees with your usual app, and what all the buttons do. If you would rather be shown the screen itself, start with how to read it.
Most weather apps show you one forecast and never tell you whether it was right. We do the opposite. Every few hours we record what each of our five sources predicted for every hour ahead, then check it against an independent record of what the weather actually did. That is well over a million forecasts scored so far, and it is all published on the accuracy page.
The short version: our combined forecast currently beats or matches every one of the five sources it is built from, on temperature, wind, rain and sky. Not because we have a better weather model, we do not have one at all, but because we know which source has actually been right lately and lean on it accordingly.
Put in the way that actually matters when you are deciding whether to go out: if we say it will be dry, we are right about nineteen times in twenty. If we say it will rain, we are right about three times in five.
That gap is not us being careless with the rain. Pinning rain to one particular hour in one particular garden is genuinely hard, showers move in minutes and miss by streets, and the forecasts everyone else uses are right about half the time when they call it. So the honest advice is to trust the dry and treat the rain as a reason to take a coat rather than a promise that you will need it.
You do not have to take our word for any of it. The accuracy page shows the workings, including where we come off worse, and you can break it down by how far ahead the forecast was looking.
It means that if today happened ten times over, rain would reach your spot on four of them. That is all. It is a chance, not an amount, not a share of the sky, and not a portion of the hour. "Rain" has a threshold behind it too: at least 0.2mm in that hour, so a few specks that darken the pavement do not count.
There is a lot of nonsense written about this online, including a popular claim that the percentage is the share of the water in the cloud that falls out. It is not, and never has been. There is one genuine source of confusion, though: American forecasters multiply their confidence by the area they expect it to cover, so their 50% can mean "certain, over half of you". British and European services, which is what we use, mean the chance at your location.
What the percentage will never tell you is how much. An 80% chance of 0.1mm is a damp five minutes; a 40% chance of 6mm will soak you. They are different questions and we are working on showing the second one properly.
The part we can promise is that ours are checked. We bucket every forecast by the chance we published and count how often it then rained, so when we say 30% we can show you that it rained about 30% of the time. Very few forecasters test that about themselves, and none of the ones we know of publish it.
Because most apps show a single source. The BBC uses one provider, your iPhone uses Apple, and so on. Weather for Dads reads five and works out one answer from them.
Not by averaging, which just gives you a wishy-washy middle. Each measure leans towards whichever source has actually been most accurate at it: the sky call goes to the source with the best record on cloud, and temperature and wind lean towards whoever has been closest lately. When the sources disagree we say so on the front page rather than quietly picking one.
Small differences are normal and usually come down to which exact spot the forecast is for. We use your precise location and show you the area and postcode we resolved it to, so you can check we have got the right place.
There is something in that, but not for the forecasts we blend. The Met Office licenses its data widely, so a fair few UK apps and websites are showing you Met Office numbers with their own styling on top. That is real, and it is probably where the idea comes from.
Our five are not that. YR comes from the Norwegian Meteorological Institute, DWD is the German national service's ICON model, which we reach through Open-Meteo, and Apple and Tomorrow.io each run their own. Different organisations, different models, which is the whole reason blending them is worth doing in the first place.
You do not have to take our word for it, because we publish the marking. If all five were the same data reinterpreted, they would score the same on the accuracy page. They do not, and not by a little. The Met Office is not top of that table on every measure, which simply could not happen if everyone else were re-presenting their forecast.
Worth being clear on one thing: those scores compare the forecast data each provider sends us through their API, which is what our blend is built from. They are not a comparison against anyone's own app.
Where it genuinely is all Met Office is warnings. They are the only body that issues official UK weather warnings, so when an amber warning shows up in the app, that came from them and we read it straight from their own feed rather than waiting for someone to pass it on.
Because they are good at different things, and not always the ones you would expect. Rather than guess, we measure it: some sources shout about rain and catch more of it at the cost of crying wolf, others are quietly better on temperature, and Apple is the only one giving true minute-by-minute rain for the next hour. The current standings are on the accuracy page and they do move.
Stack five together, weight them by what they have actually got right, and the gaps mostly cover for each other. That is why the combined answer scores better than any single source in it.
Because their licence does not allow what we do, and not on a technicality. Three of its conditions rule it out on their own. We may not store their forecasts in order to score them later, which is the whole method behind this app. We may not combine their data with other weather information, which is what a blend is. And we may not use their data to rate, rank or evaluate them, which is precisely what the accuracy page does to every source on it.
So the honest answer is that we do not know whether AccuWeather is any good, because the terms we would have to sign forbid us from finding out and telling you. That is true at every price they offer. It is not a question of paying more.
We would rather carry five forecasts we are free to mark in public than six where one of them cannot be questioned.
The day, from six in the morning until ten at night. The line is the temperature, with the real degrees printed on it every couple of hours, and a little weather icon riding along it. The dotted line is now.
Drag it, or drag the picture, and everything follows the hour you land on: the big temperature, the sky, the wind and the painted scene. The handle tells you which hour you are reading. Nothing you do to it changes the forecast, so poke about.
How much rain, as opposed to how likely it is. The percentage above them answers "will it rain?", and the marks answer "how wet will I get?", which are two different questions that most weather apps roll into one.
One mark is a shower, two is wet, three is a soaking. On an hour, three marks means at least 1.5mm in that hour; across a whole day it means at least 6mm, because an amount that would be a downpour in one hour is an ordinary wet Tuesday spread over a day. We never write millimetres in the app, because nobody thinks in them.
Green is calm, under 12kph: put the parasol up. Cream is breezy, 12 to 25, which is an ordinary British afternoon and not worth colouring. Amber is windy, 25 and over: bring the washing in.
Gusts count too. A gust can push an hour up a band on its own, because it is the gust that takes the washing stand over, not the average. Where that happens we print the gust next to the average in the hourly, so the number that caused the colour is always findable.
Because that is the answer, and a row of numbers is not. Each day gets a few words worked out from that day's own hours: "Rain from ten" means the rain arrives at ten, "Dry from eleven" means it clears then, and "Showers about" means it is scattered and no hour is safe.
Green ones are days worth going outside for. The words never contradict the icon or the arrow beside them, because they are all worked out from the same numbers.
Instead of making you read numbers, we turn the forecast into a plain answer for the things a dad actually wants to do. Each one says when: NOW if it is good to go, the time it opens up if that is later today, or the next good day if today is a no. It is based on temperature, rain and wind across the day.
They also watch the live rain, not just the hourly forecast. If it is actually raining on your head, they will not cheerfully say NOW for the barbecue just because the model for that hour said dry. Beer is never cancelled by weather, it just says OUT or IN, and before midday it only tells you when it opens, on principle.
The one signal no model has: actual people standing in their actual gardens. Tap Report, top left, to say what it is really doing, including how heavy the rain is, and dads nearby see it as one line under the forecast, in your own words if you add a note. That matters most when rain is heading their way.
We also score the forecasts against those reports, as a second opinion kept deliberately separate from the station-based scores. It does not quietly change your forecast; it is there so you can see what the dads made of it alongside what the instruments said. Reports are anonymous and approximate; see the privacy page.
With your permission, the app uses your device location to fetch the forecast, then shows you the neighbourhood and postcode district it resolved, for example "New Malden Village, KT3", so you can confirm it is right. You can also type a postcode or place name instead. We never keep a history of where you have been. The full detail is on the privacy page.
Free, with no account, no sign-up, and no name or email required. There is nothing to log into.
On an iPhone, get it from the App Store: search for Weather for Dads, or tap this link.
Everywhere else, and on an iPhone if you would rather not bother with the App Store, open the site in your phone's browser and add it to your home screen. On Android, tap the "Add to phone Home Screen" button in the app itself; on an iPhone tap Share and then "Add to Home Screen". It opens like a normal app, straight to the forecast, and it is the same app either way. There is nothing in the App Store version you miss out on by using the website.
One Android quirk worth knowing, because it catches people out. Use the button inside the app, or your browser's "Install app" option if you see one. Avoid "Add to Home screen" in the browser menu. Installing properly gives you a real app: it appears in your app drawer and you can drag it into a folder alongside your other weather apps. "Add to Home screen" can instead leave you a plain shortcut, which sits on the home screen but cannot be filed away anywhere or found in the drawer. If you have ended up with one you cannot move, delete it and add it again using the button. Browsers other than Chrome, such as Samsung Internet or Firefox, tend to only ever make the shortcut kind.
One a day, because a wet forecast goes down easier with a groan. Rate it, or submit your own and it goes up against everyone else's, with the week's winner revealed on Sunday evening. You will find it under Other stuff, in the bottom right. Standards are low and the bar is not rising.
You can send one on, too, using the Share button on the joke. If it came from another dad, his name goes with it.