A single-plant Hoshin Kanri deployment is hard enough. You sit with the plant manager, walk the gemba, and iterate a strategic target until the operators on the floor can translate it into concrete process steps. The catchball dialogue is direct, the feedback loop is short, and the cultural barriers are manageable because you share the same hallway. That dynamic breaks the moment you attempt to scale the same process across three, five, or fifteen manufacturing sites.
Across two decades building and transitioning quality systems in automotive and aerospace, I have watched organisations try to export their catchball process using a PDF playbook and a shared SharePoint site. The result is almost always the same. The headquarters team believes they have deployed strategy. The remote sites receive a memo. The bi-directional dialogue that makes catchball effective never materialises, and the strategic plan quietly becomes decorative wall art in a conference room nobody uses.
The failure is not a refusal to adopt lean methodology. It is a structural problem of scale. Catchball relies on rapid, honest iteration between the people who set targets and the people who execute them. As distance, product mix, and organisational layers increase, the friction in that iteration kills the dialogue. To scale catchball effectively, you must engineer the process to handle that friction deliberately.
The Volume Problem: Dilution Through Layers
When a corporate quality director announces a target to reduce the cost of poor quality by 40% across a five-plant network, the message passes through a chain of regional managers, plant managers, value-stream managers, shift supervisors, and finally operators. Each translation step risks dilution. The corporate target of 40% becomes a plant goal of 20% because the plant manager assumes the other sites will carry the weight. By the time it reaches the shift supervisor, the target is a vague instruction to reduce scrap.
In a single-site deployment, the originator of the target is physically close to the executor. The catchball bounces twice and settles. In a multi-site deployment, the same bounce requires scheduled video calls across time zones, and the operational feedback that should reshape the target gets lost in translation. The corporate team never hears that the target is structurally unrealistic for a plant running legacy equipment with a different product mix.
The solution is not to flatten the organisation overnight. It is to cap the number of translation layers between the strategic originator and the operational executor. If a target must pass through more than three management tiers before reaching the shop floor, the catchball process is already compromised. Design the deployment architecture so that the person who sets the target speaks directly to the person who owns the process, even if that means bypassing intermediate management layers for the initial iteration.
Standardising the Dialogue Format Across Sites
Different plants speak different dialects of quality engineering. One site measures defects in PPM, another in percentage, and a third uses a proprietary scoring system buried in a local MES. When you attempt catchball across these sites without a standardised dialogue format, the corporate team spends more time reconciling terminology than evaluating the substance of the feedback. The iteration stalls, and the sites revert to their silos.
The A3 report is the most effective tool for standardising the catchball dialogue across multiple sites. A single-page A3 forces every plant to articulate its interpretation of the strategic target, its proposed countermeasures, the process constraints it faces, and the resources it needs. The format is identical whether the plant is in Bratislava, Birmingham, or Baja. The corporate team can compare A3s side by side and immediately see where the targets are viable and where they need adjustment.
The discipline lies in refusing to accept alternatives. If a plant manager submits a PowerPoint deck instead of an A3, the catchball iteration does not begin. The standard format is not bureaucratic preference; it is the mechanism that allows the corporate team to process feedback from five plants simultaneously without losing structural detail. Without it, multi-site catchball degrades into a series of unrelated conversations.
Standardised Multi-Site Catchball Cycle
- 01Issue strategic directionCorporate sets 3-5 measurable objectives without specifying local methods
- 02Site-level A3 responseEach plant submits a standard A3 defining local targets, constraints, and proposed countermeasures
- 03Cross-site reconciliationCorporate reviews all A3s, identifies conflicts, and adjusts targets where operational reality demands it
- 04Commit and resourceFinal targets locked in X-Matrix with named owners and allocated budget at each site
Handling Product Mix and Capability Variation
A corporate target to achieve Cpk 1.33 across all critical characteristics sounds straightforward. But when one plant stamps body panels on a 15-year-old press line and another runs a modern servo press with in-die sensing, the starting capabilities are wildly different. A uniform target punishes the legacy plant and under-challenges the modern one. Catchball at scale must accommodate this variation or the process loses credibility with both sites.
The mechanism is tiered target setting. The corporate direction remains uniform: reduce variation, achieve process capability. But the catchball dialogue allows each site to negotiate the specific Cpk targets and timelines based on their current baseline. The legacy plant commits to Cpk 1.0 in phase one and 1.33 in phase two, contingent on a capital investment in press controls. The modern plant commits to Cpk 1.67 immediately. Both commitments are captured in their respective X-Matrices.
This is where many corporate quality teams flinch. They fear that tiered targets signal inequality or favouritism. The opposite is true. A target that ignores the operational reality of a specific plant is not a strategy; it is a fiction. Tiered targets, validated through honest catchball iteration, produce a deployment plan that is executable rather than aspirational. The X-Matrix documents the gap and the plan to close it, which satisfies both internal governance and external audit scrutiny.
The Multi-Site X-Matrix and Accountability Architecture
In a single-plant deployment, one X-Matrix on the wall of the production office is sufficient. Everyone can see it, everyone signs it, and the visual proximity reinforces the commitment. Scaling this across multiple sites requires an accountability architecture that links a master corporate X-Matrix to site-level matrices without creating a bureaucratic chain that nobody reads.
The master X-Matrix lives at the corporate level. It captures the breakthrough objectives, the strategic metrics, and the overall targets. Each site maintains its own X-Matrix that feeds into the master, documenting the site-specific tactical targets, improvement activities, and named owners. The link between the two matrices must be explicit. A site-level improvement in OEE or PPM must trace directly to a line on the master document.
During ISO 9001 and IATF 16949 audits, I have evaluated multi-site organisations where each plant had a beautiful local X-Matrix with no connection to any central strategy. The matrices were internally consistent but strategically irrelevant. The auditor's question is always the same: how does this improvement activity support the corporate quality objective? Without a traceable link, the deployment fails the test of alignment, regardless of how diligently each site executed its local plan.
A target that travels across five plants without picking up operational feedback is a memo, not a strategy.
Cultural Drift and the Distance Problem
Catchball depends on psychological safety. An operator must feel free to tell the plant manager that a target is impossible given current tooling wear rates. A plant manager must feel free to tell the corporate director that the site needs capital investment before it can commit to a Cpk upgrade. In a single site, the leadership can build that safety through daily presence and visible response to feedback. Across multiple sites, that presence is impossible to maintain, and cultural drift fills the vacuum.
Remote sites develop their own micro-cultures. If the site manager is authoritarian, operators will agree to any target during the catchball iteration and silently ignore it once the corporate team disconnects. The A3 will look perfect. The commitment will be documented. The execution will be zero. This failure mode is invisible from headquarters until the quarterly review reveals the gap.
The countermeasure is a structured gemba walk programme where corporate leaders visit each site specifically to validate the catchball process. The purpose is not to inspect the A3 reports. It is to ask operators directly whether they understand the targets, whether they contributed to the countermeasures, and whether the resources promised in the X-Matrix actually materialised. If the answers are hollow, the catchball process at that site has failed regardless of what the documentation says.
Single-Site vs. Multi-Site Catchball Failure Modes
Single-site failure
- Targets diluted by one or two layers
- Cultural barriers addressed through daily presence
- Feedback loop closes in days
- One X-Matrix visible to all participants
Multi-site failure
- Targets diluted through five or more tiers
- Cultural drift creates silo behaviour at remote plants
- Feedback loop stalls for weeks across time zones
- Disconnected matrices with no traceable alignment
Sustaining the Dialogue at Scale
The most common point of failure in multi-site catchball is the assumption that the process is self-sustaining once launched. It is not. The bi-directional dialogue requires active facilitation, and as the number of sites increases, the facilitation effort scales geometrically. A corporate quality team managing five plants can sustain quarterly catchball reviews. The same team managing fifteen plants cannot, and the process will degrade unless the resourcing changes.
Track four indicators to measure the health of multi-site catchball. Target penetration: the percentage of operators across all sites who can articulate their department's strategic target. Aim for above 80%. Feedback quality: the proportion of sites that submit specific, actionable countermeasures rather than passive agreement on their A3. Iteration speed: the time required for a target to complete the full catchball cycle across all sites. Anything over four weeks indicates structural friction. Hoshin attainment: the overall achievement rate of the strategic plan. A functional multi-site catchball process typically correlates with above 70% attainment.
When these indicators decline, the response should not be to reissue the strategy or tighten the reporting cadence. The response should be to examine the catchball architecture itself. Are there too many translation layers? Is the A3 format standardised? Are the site-level X-Matrices linked to the master? Are corporate leaders walking the gemba at remote sites? The methodology is sound. The execution at scale demands engineering discipline that most organisations underestimate.
Multi-Site Catchball Health Indicators