Ask “how?” again and again until a big goal breaks down into actions you can start this week.
Big goals are hard to act on. “Grow the business” or “get fit” doesn't tell you what to do on Monday morning. The How-How Diagram fixes that by asking one question, “How?”, over and over, and writing each answer to the right of the card that prompted it.
Your goal sits on the left. Ask how you could achieve it and you get a few broad routes. Ask how for each route and you get more specific means. Keep going and the tree spreads out into concrete actions, things a named person could start this week. Along the way you mark which branches are alternatives and which must all happen together, then choose the paths worth following.
Use it when you have a clear goal but no plan, when a plan feels vague or full of gaps, or when a team needs to see every route to a target before deciding where to put its effort.
Each column answers “how?” for the one to its left. Read it right to left and each card answers “why?”
What exactly do you want to achieve, written so that asking “how?” makes sense?
How could this be achieved? What are all the different routes, not just the first one?
For each answer, how would that be done? Keep asking until the cards turn concrete.
Which cards are alternatives you could choose between, and which must all happen together?
Point at any card and ask “why?”: does the card to its left give a clear, convincing answer?
Which branches give the most progress for the least effort, and who will start each one, when?
A goal and an action live at different levels. The How-How Diagram builds the staircase between them, one small, answerable question at a time.
We stall on goals that are too big to act on. Each “how?” takes one step down from aims to means, until you reach something specific enough to put in a calendar. Progress starts with that first concrete card.
Asking “how?” rather than “what's the plan?” invites several answers at every level. You see alternatives you'd otherwise skip, so the plan you choose is the best of several, not just the first that came to mind.
Because each action hangs from the card it serves, you can always trace it back to the goal. That keeps busywork out of the plan, and it helps people see why the task in front of them matters.
A card with nothing to its right isn't a plan yet; it's a hope. On a tree, unfinished branches stand out at a glance, so you can see exactly where more thinking is needed before anyone starts work.
Put the goal on a card at the far left of a wall or page. Phrase it as something to achieve, with a measure if you can: “Fill 15 more repair slots a week by spring,” not just “repairs.”
Ask “How could we do this?” Write each answer on its own card in a column to the right. Aim for three to five genuinely different routes.
Take each card in turn and ask how again. Put the answers in the next column. Keep going until the cards on the right are actions a named person could start.
Label each fork. At an OR fork you can pick the cheapest or fastest branch; at an AND fork every branch must get done. Use two colors so the difference is easy to see.
Go from right to left. Each card should explain why the cards to its right exist. Fill gaps, merge duplicates and cut any card that doesn't serve the goal.
Pick the branches with the best mix of impact and effort. For each chosen action, name an owner, a first step and a date.
Sticky notes or index cards, a wall or large sheet, and markers in two colors for AND and OR. With AI, the prompt below can build the tree with you.
Who: a family-run café near a business park. Goal: 40 more lunch orders a day within three months. Constraint: two staff at lunchtime and no money for advertising.
40 more lunch orders a day by the end of the quarter, without adding staff or advertising.
How? A group pre-order form for teams. Lunch cards that office managers can buy in bulk. One delivery run at 12:15.
How? A grab-and-go fridge stocked by 11:30. A shorter noon menu. Online orders paid ahead and waiting on a shelf.
How? A weekly special announced every Monday. A loyalty stamp that counts double at lunch. A quiet table for laptop workers.
The fridge AND the short menu go together, since both free up the two staff. Pre-orders OR delivery: start with one, not both.
Monday: the owner emails five office managers. Wednesday: the fridge is stocked. Results reviewed after two weeks.
A How-How Diagram means asking the same question many times without settling for the first answer. AI does that tirelessly: it suggests routes you missed, pushes vague cards down to real actions and checks the logic back up the tree. You know which routes are realistic, affordable and worth your time.
Ask for routes that are genuinely different, not variations on one idea.
Give me 5 truly different ways to get 40 more lunch orders a day. No two can rely on the same thing.
Take one vague card and keep asking how until it becomes an action.
Take the card 'win nearby offices' and ask how twice more, until each card is a job for one person.
Let AI read the tree from right to left and flag the weak links.
Read my tree from right to left. Flag any card that doesn't clearly serve the card to its left.
Score the final actions so you start with the ones that matter most.
Score each action in my tree 1-5 on impact, ease and speed. Which three should I start this week?
Describe your goal, pick a mode, then copy the prompt into Claude, ChatGPT, Gemini or Copilot. In coach mode the AI asks “how?” with you one level at a time and waits for your answers before adding its own.
The “Open in” links put the prompt in the web address. If your challenge includes anything sensitive, copy and paste instead.
After the session, click one to copy it, then paste it into the same chat.