Across multiple squads sharing the same catalog of cost-visible agent workflows, how does introducing cross-squad comparative dashboards (showing per-workflow usage, cost, and outcome deltas by squad) affect (a) convergence on a common set of repeatable workflows versus local ‘shadow’ variants, and (b) team-level trust in governance—does transparency help squads copy proven patterns and retire weak ones, or does it trigger defensive behavior and underreporting of high-cost exploratory usage?

Answer

Cross-squad comparative dashboards tend to increase convergence on shared, repeatable workflows and can strengthen trust in governance, but only when dashboards stay workflow-centric, are used for portfolio-level learning (not squad shaming), and explicitly protect exploratory usage. Poorly handled, the same transparency drives defensive behavior, label games, and hidden high-cost use.

(a) Convergence vs. shadow variants

• Likely benefits
  • Squads more easily spot “winning” workflows in other squads (high outcome/ cost) and copy them, increasing convergence.
  • Weak workflows with low outcome/ high cost become visible across squads and are easier to retire or redesign.
  • Local shadow variants that clearly outperform golden ones on outcomes at similar cost gain evidence for promotion, reducing long-lived forks.
• Risks
  • If dashboards are framed as league tables, squads avoid “we look expensive” by:
    • routing work through generic workflows,
    • relabeling production-like use as exploration,
    • underreporting or avoiding high-cost but needed workflows.
  • Squads may keep powerful variants off the shared catalog to avoid scrutiny, increasing hidden divergence.

(b) Trust in governance

• Trust increases when
  • Metrics and reviews are explicitly workflow/portfolio-level; leaders talk about “this workflow family” not “this squad’s spend.”
  • Dashboards pair cost with outcome metrics and exploration quotas, and reviews ask “is this worth it?” not “why so expensive?”.
  • There are clear, lightweight paths to: (1) promote strong local variants, (2) tune or retire weak workflows, and (3) allocate protected exploration budget.
• Trust erodes when
  • Comparative views are used to pressure low-spend conformity or blanket cost cuts.
  • High-cost exploratory usage leads to ad-hoc caps or negative attention, teaching squads to hide or relabel usage.
  • Only cost deltas are discussed; outcome and context are ignored.

Net: comparative dashboards are most useful when treated as portfolio-learning tools (to standardize what works and evolve the catalog) rather than performance scorecards. They then support convergence on repeatable workflows and moderate, stable trust in governance, with persistent risk of defensive behavior if cost is overemphasized.