The website could answer "who in the pack is overdoing it?". The phone only showed load one dog at a time. Not any more: the phone's Pack reads its load the way the website does.
What you see
At the top of the Pack, three choices: Faces (the screen you know, unchanged), Load and Summary.
Load opens on the Load and plan, dog by dog scatter: to the right, the share of this week's plan already run; up, recent load relative to the usual one, with the "caution" and "danger" bands of your format. A dashed line marks where you should be today. Each dog is a circle, a square or a triangle according to its band, and the list below says the same figures in words — "72 % of plan · ratio 1.94 · danger". Further down, the load per dog since the season started, with the kennel median, then team form: base fitness, recent fatigue, and their balance above or below zero.
Summary plays "Why this number?" on every dog: one row per dog, its number, and the reason in one line ("Same recent load, lower base"). Tap the row: a sheet gives everything else — when the number was last computed, the kilometres over 7 and 28 days against the pack median, the runs, the longest gap without a run while the pack went out, the last rest — and opens the dog's page.
And "Understand this number", on a dog's page, now shows the facts on the phone too in its Why tab: "From 22/09 to 27/09: 6 days without a run, 5 pack outings meanwhile", "Rest declared since 01/10".
On a dog's page, Training tab, the Fitness model card finally tells the phone what we know about that dog: a bar that grows with its rated sessions.
What others see, and what they don't
Every member of the kennel sees what you see: it is the kennel, not your private log. Nothing leaves it — no public page, no club. A dog that has left is not there; a guest dog has its row, but never counts in "the pack" it is compared with.
What it does not do
It does not plot a dog too early. A dog enters the scatter only with 28 computed days of load and a plan week. Before that, the scatter counts it ("2 dogs not plotted") rather than drawing it where the arithmetic would throw it: on day nine of a season, everyone would read "danger".
It never puts a dog at zero. A dog with no computed day is left out of the ranking, and the screen says so; a day covered by a single dog is dropped from the team curve, and counted; a gap in the season stays a gap, never bridged by a straight line.
It does not read without signal. These numbers come from the server, as on the website: offline, the screen says so in a sentence. The Fitness model card, though, reads offline.
Confidence is not accuracy. The bar says how many rated sessions we have for this dog, not that the model is right: it keeps population defaults until vets validate them. It is not medical advice.
No big table. Nine columns don't fit on a phone, let alone with large text: one row per dog, and the sheet for the detail. You sort by load, name, kilometres, runs or gap, ascending or descending.
Three choices we made
Shape rather than verdict colour. On the website, a scatter dot takes the colour of the dog's availability. On the phone it takes the shape of its band — circle, square, triangle — because that is what still reads on a screen washed out by snow; availability already has its view, Faces.
Curves and bars in blue. Orange is kept for the one action of the moment; data reads in blue, as everywhere else in the app.
The switch only for a kennel that trains. A pure breeding kennel keeps its Pack as it was: it has no load to read.