Catchball is the two-way conversation at the center of Hoshin. A manager throws a draft objective to a direct report. That person tests it and throws back an answer. The two go back and forth until they agree on the work, the resources and the date.
Without catchball, you have a list of orders. The plan can look good on paper and still fail on the floor, because the people who must carry it never tested it. If they didn't test it, they don't own it.
The five things catchball tests
Each round of catchball tests the draft objective against five questions. If you receive an objective, you work through them. If you hand one down, expect answers on all five.
| Test | The question to ask |
|---|---|
| Feasibility | Can this be done at all, given what you know about the work? |
| Ownership | Will one named person answer for this, and does that person accept it? |
| Resources | What people, money or machine time does this need that you don't have now? |
| Timing | Does the date fit with everything else that happens this year? |
| Dependencies | Which other departments must do something for this to work? |
How to run a round
Follow these steps for each objective at each level of the org chart.
- Send the draft ahead of time. Give the person the objective in writing, with the reason for it and the date. Give them a few days to look at their own data before you meet.
- Ask them to answer the five tests. Have them bring their answers in writing. This changes the meeting from a briefing into a conversation about facts.
- Hear the counter before you respond. A counter is an answer that changes something: a smaller target, a later date or an ask for resources. Let the person finish and show their reasoning.
- Decide on each ask. Say yes, say no or change the objective. If you say no to an ask, say what comes off the list or what changes in the target. Saying no and leaving the target alone isn't catchball.
- Write down what you agreed. Record the final words of the objective, the owner, the date and every ask, with the answer to each one.
- Repeat one level down. The person who received the objective now hands parts of it to their own direct reports, and the same conversation starts again. Any change that the lower level needs comes back up.
A worked example
The company in this example is fictional. It's a metal parts maker with 180 employees, one plant and two shifts. The numbers are made up to show the method. They aren't results.
The plant manager holds this annual objective: cut Line 2 scrap from 4% to 3% this year. She drafts a piece of it for the production supervisor: "Bring Line 2 scrap to 3% by September 30."
The production supervisor looks at his data and comes back with these answers:
| Test | The supervisor's answer |
|---|---|
| Feasibility | Most Line 2 scrap comes from the press brake. The team can likely fix the part that starts at changeover. The rest has no pattern yet. |
| Ownership | He accepts it, and he wants the second-shift supervisor to own the press brake work. |
| Resources | He needs each lead operator off the line for four hours to write setup sheets. |
| Timing | September 30 is too early. The plant's maintenance shutdown takes Line 2 down in May, and that costs a month of trials. He proposes December 31. |
| Dependencies | Some scrap starts with bad castings. That part needs the quality manager to rebuild the check on purchased castings. |
Now the plant manager has decisions to make. Here is how the round goes:
- Date. She has the same shutdown on her calendar, so she accepts a later date. They settle on November 30, because the annual objective needs time to show in the full-year number.
- Off-line hours. She says yes to the four hours per lead operator, on the condition that the hours come after the month-end shipping rush.
- Castings. She doesn't decide this one alone. She calls in the quality manager. This is the horizontal part of catchball: peers at the same level depend on each other. The quality manager already has a project on purchased castings, and agrees to share the check results with Line 2 each week.
- Ownership. She agrees that the second-shift supervisor can own the press brake work, but she keeps the production supervisor as the owner of the 3% objective.
The final objective reads: "Bring Line 2 scrap to 3% by November 30. Second shift owns the press brake A3. Lead operators get four hours each after month-end. Quality shares castings results each week." The production supervisor now has a different objective from the one he received, and he can explain why every word is there.
Then he starts the same conversation with the second-shift supervisor.
What goes wrong
Watch for these signs that you aren't really running catchball:
- Nothing ever changes. If every objective comes back with the same words, people aren't testing it, or they don't believe that they can push back.
- Asks have no answers. A list of asks with no yes or no leaves the receiver holding the target without the means. Answer every ask.
- The talk happens only once. People find problems after the year starts. Keep the same conversation open in your one-on-ones. If the target has to change, change it in the open and keep the reason.
- The plan moves only down the tree. If nobody asks about dependencies, peers find out about each other's plans when something breaks.
- It runs too long. Catchball is a means to a plan that people own. Set a date when planning closes, and hold the remaining questions for the first monthly review.
Check that it adds up, by hand
Before you close a round, ask one more question: if every action on this list succeeds, does it close the gap? In the example, the production supervisor should show how much of the 1-point scrap gap the press brake work can take out, and how much depends on the castings. Hoshin's standard forms don't test this for you. An X-matrix shows that two items link. It doesn't show that the work adds up to the goal. You have to ask.
How Run Hoshin handles this
Run Hoshin has no catchball screen. The negotiation itself happens in your conversation, and the app doesn't track offers and counters. It does record the result. The top seat writes objectives in words and hands them to direct reports. A hand-down goes one level at a time, to a direct report. The database refuses a hand-down to a peer, or one that skips a level. The app has no place to record a dependency between peers, so you agree on those outside the app. When the words of a goal change, the seat above writes a new version and gives the reason, and the earlier words stay in the history. The app doesn't check that the goals below add up to the goal above. See what the app does not do yet.
Sources
- Lean Enterprise Institute, Strategy Deployment, which describes the dialogue that is often called catchball
Try handing an objective down in words, in a trial with sample data.