[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"/en/workflows/signal-quality-checks-for-better-branch-decisions":3},{"id":4,"slug":5,"locale":6,"translationGroupId":7,"localeSwitchApproved":8,"title":9,"description":10,"documentationMarkdown":11,"workflowJson":12,"category":195,"tags":196,"integrations":198,"difficulty":201,"author":202,"verified":33,"featured":33,"date":203,"modified":203,"icon":7,"imageSrc":7,"path":204,"alternates":205,"seo":206},"eef8f1d2-6706-496e-a557-39034bdc75ad","signal-quality-checks-for-better-branch-decisions","en",null,true,"Signal Quality Checks for Better Branch Decisions","A decision-support chat flow that helps teams separate trustworthy branch signals from polished noise, catch dirty data early, and decide when automation deserves human override.","## How it works\nThis workflow gives branch leaders and operators a fast “sanity check” menu for common decision traps: numbers that look clean but lie, conversations that mislead, and attribution that flatters the wrong thing. It’s built for the moment right before a confident meeting goes off the rails.\n\nIt starts by letting your knowledge base answer specific questions (definitions, metric notes, policy context). If there’s no clear KB answer—or you want a structured check—it routes to a simple menu of decision-shaped prompts. Each choice returns a practical checklist and loops back so the team can run multiple checks in one session.\n\n## Key features\n- Knowledge-base-first support for quick “what does this metric mean?” questions before routing to the menu.\n- Button-based menu covering the most common branch signal failure modes.\n- Short, operator-friendly checklists that emphasize what breaks first (and how to notice).\n- Built-in loop back to the knowledge base and menu so users can run several checks without restarting.\n- Optional handoff to a human team when the situation needs judgment or context.\n\n## Step-by-step\n1. **Trigger:** A user starts the workflow in your Calypso Inbox channel.\n2. **Knowledge base check:** The **Knowledge Base Policy** attempts to answer the user’s question using your existing knowledge base.\n3. **Menu:** The workflow presents a button menu: **Trust this number**, **Spot dirty signal**, **Automation vs judgment**, **Compare branches & attribution**, **Build signal culture**, or **Talk to a human**.\n4. **Guided response:** Based on the button selected, the workflow sends a targeted checklist and “watch-outs” for that decision type.\n5. **Loop for next check:** After the checklist, the workflow returns to the knowledge base step and menu so the user can ask another question or run another check.\n6. **Human support (optional):** If **Talk to a human** is selected, the workflow hands off with a clear context-setting message.\n\n## Setup requirements\n- **Calypso Inbox** connected to a channel that supports interactive buttons (e.g., **WhatsApp**).\n- **Knowledge Base content** published in Calypso (recommended). If you don’t have a KB yet, the menu still works—there are no additional credentials required for this workflow.",{"id":13,"teamId":14,"name":9,"version":15,"workflowVersion":16,"nodes":17,"connections":165,"routingEnabled":8,"active":33},"wf_signal_quality_checks_branch_decisions_v1","calypso-public-library","1.0.0",1,[18,34,41,53,84,94,100,106,112,118,124,132,138,144,150,156],{"id":19,"name":20,"type":21,"typeVersion":16,"position":22,"parameters":24,"category":32,"deletable":33,"connectable":33},"node_flow_configs","Flow Configs","flow-configs",[23,23],40,{"name":9,"description":25,"tags":26,"triggerType":31},"Decision-support chat flow to separate trustworthy branch signals from polished noise, catch dirty signal early, and choose when automation needs human judgment.",[27,28,29,30],"signal-quality","branch-metrics","decision-systems","data-hygiene","input","policy",false,{"id":35,"name":36,"type":31,"typeVersion":16,"position":37,"parameters":40,"category":31,"deletable":33,"connectable":8},"node_input","Inbound Message",[38,39],160,220,{},{"id":42,"name":43,"type":44,"typeVersion":16,"position":45,"parameters":47,"category":52,"deletable":8,"connectable":8},"node_kb_policy","Knowledge Base Assist","knowledge-base-policy",[46,39],380,{"enabled":8,"fallbackToRouting":8,"sticky":33,"stickyMode":48,"activationOpener":49,"personalization":51},"default",{"enabled":8,"instruction":50},"Answer using our knowledge base when possible. If the user’s request is unclear, high-stakes, or not covered, proceed to the decision menu.",{"useContactName":33},"response",{"id":54,"name":55,"type":56,"typeVersion":16,"position":57,"parameters":59,"category":52,"deletable":8,"connectable":8},"node_menu","Decision Menu","interactive-message",[58,39],620,{"messageType":60,"headerText":61,"bodyText":62,"footerText":63,"sectionTitle":64,"buttons":65,"ctaDisplayText":64,"ctaUrl":64},"button","Decision Signal Checks","Pick the check you need. These are built for messy reality: branch numbers, conversations, attribution, and the places “clean” data hides problems.","Tip: Ask a metric question—KB tries first.","",[66,69,72,75,78,81],{"id":67,"title":68},"trust_numbers","Trust this number",{"id":70,"title":71},"dirty_signal","Dirty signal",{"id":73,"title":74},"automation_judgment","Auto vs human",{"id":76,"title":77},"compare_branches","Compare branches",{"id":79,"title":80},"signal_culture","Signal culture",{"id":82,"title":83},"talk_human","Talk to human",{"id":85,"name":86,"type":87,"typeVersion":16,"position":88,"parameters":91,"category":93,"deletable":8,"connectable":8},"node_if_trust_numbers","If: Trust this number","if",[89,90],860,120,{"buttonId":67,"operator":92},"equals","routing",{"id":95,"name":96,"type":87,"typeVersion":16,"position":97,"parameters":99,"category":93,"deletable":8,"connectable":8},"node_if_dirty_signal","If: Spot dirty signal",[89,98],200,{"buttonId":70,"operator":92},{"id":101,"name":102,"type":87,"typeVersion":16,"position":103,"parameters":105,"category":93,"deletable":8,"connectable":8},"node_if_automation_judgment","If: Automation vs judgment",[89,104],280,{"buttonId":73,"operator":92},{"id":107,"name":108,"type":87,"typeVersion":16,"position":109,"parameters":111,"category":93,"deletable":8,"connectable":8},"node_if_compare_branches","If: Compare branches & attribution",[89,110],360,{"buttonId":76,"operator":92},{"id":113,"name":114,"type":87,"typeVersion":16,"position":115,"parameters":117,"category":93,"deletable":8,"connectable":8},"node_if_signal_culture","If: Build signal culture",[89,116],440,{"buttonId":79,"operator":92},{"id":119,"name":120,"type":87,"typeVersion":16,"position":121,"parameters":123,"category":93,"deletable":8,"connectable":8},"node_if_talk_human","If: Talk to a human",[89,122],520,{"buttonId":82,"operator":92},{"id":125,"name":126,"type":127,"typeVersion":16,"position":128,"parameters":130,"category":52,"deletable":8,"connectable":8},"node_text_trust_numbers","Checklist: Which numbers deserve trust","text-message",[129,90],1120,{"text":131},"**Which branch numbers deserve trust (and which are polished noise)**\n\nUse this before you let a single KPI run the meeting:\n\n1) **Lineage test:** Can you explain where the number comes from in one breath (source → transformation → report)? If not, it’s a *story*, not a signal.\n2) **Denominator sanity:** What’s the base? Counts without context are magnets for accidental propaganda.\n3) **Stability check:** Does it move smoothly with reality, or jump when someone changes logging, incentives, or definitions?\n4) **Lag awareness:** Are you reading a leading indicator like it’s a result? (That’s how teams celebrate early… and apologize late.)\n5) **Gaming exposure:** If someone’s bonus depended on it, how would they improve it without improving the business?\n\nIf you share the metric name + where you saw it, I can help pressure-test it.",{"id":133,"name":134,"type":127,"typeVersion":16,"position":135,"parameters":136,"category":52,"deletable":8,"connectable":8},"node_text_dirty_signal","Checklist: Spot dirty signal early",[129,98],{"text":137},"**How to spot dirty signal before the meeting goes off the rails**\n\nDirty data rarely looks dirty—it looks *confident*. Quick tells:\n\n- **Too perfect:** Flat lines, round numbers, or suspiciously consistent conversion rates.\n- **One-week miracles:** Sudden jumps that conveniently align with a reporting change, staffing change, promo, or “we fixed tracking.”\n- **Missingness with a pattern:** Gaps that only happen on weekends, in one branch, or for one channel.\n- **Metric disagreements:** Two dashboards answer the same question differently. That’s not a tie—it’s a warning label.\n- **Conversation drift:** Frontline notes say one thing; the KPI says the opposite. Don’t average them—investigate the mismatch.\n\nIf you can, ask: **What would we expect to see if this were false?** Then go look for that evidence on purpose.",{"id":139,"name":140,"type":127,"typeVersion":16,"position":141,"parameters":142,"category":52,"deletable":8,"connectable":8},"node_text_automation_judgment","Guide: When to trust automation vs humans",[129,104],{"text":143},"**When leaders should trust automation (and when they still need judgment)**\n\nTrust automation when:\n- The decision is **reversible** (you can undo it cheaply).\n- The signal is **well-instrumented** (clear definitions, stable logging, monitored drift).\n- The outcome is **frequent** (lots of reps = fast learning).\n\nDemand human judgment when:\n- The cost of a wrong call is **high** (compliance, safety, reputational risk).\n- The system can be **gamed** (incentives, scripts, “checkbox behavior”).\n- You’re relying on **proxy metrics** (attribution, sentiment tags, self-reported reasons).\n\nPractical rule: **Automate the routine; escalate the weird.** If the case looks rare, emotional, or politically charged, treat the model output as a *vote*, not a verdict.",{"id":145,"name":146,"type":127,"typeVersion":16,"position":147,"parameters":148,"category":52,"deletable":8,"connectable":8},"node_text_compare_branches","Guide: Comparing branches & attribution pitfalls",[129,110],{"text":149},"**What teams misread when comparing branches, conversations, and attribution**\n\nCommon ways smart teams fool themselves:\n\n1) **Selection bias:** One branch gets harder customers, different hours, or different staffing—and the dashboard calls it ‘performance.’\n2) **Unequal measurement:** A branch that logs more diligently can look worse (more defects captured) or better (more wins recorded). Either way: apples vs pears.\n3) **Attribution vanity:** Last-touch wins, while earlier steps do the real work. If your channel ‘looks amazing,’ it may just be standing at the finish line.\n4) **Conversation counting:** More chats/calls can mean growth—or confusion. Pair volume with resolution quality.\n5) **Small numbers, big confidence:** Low sample sizes create loud nonsense. Require minimum volumes before ranking branches.\n\nTry this: when Branch A beats Branch B, ask **“What changed?”** If the answer is ‘nothing,’ you’re probably reading noise or a measurement difference.",{"id":151,"name":152,"type":127,"typeVersion":16,"position":153,"parameters":154,"category":52,"deletable":8,"connectable":8},"node_text_signal_culture","Guide: Build a signal culture that drives decisions",[129,116],{"text":155},"**How to build a signal culture that helps decisions happen (not just slides)**\n\nA few practices that beat another dashboard:\n\n- **Decision-first metrics:** Start with the decision you need to make, then pick the smallest set of signals that truly change that decision.\n- **Definitions with teeth:** Write down what a metric *means*, what it *doesn’t*, and what breaks it. Make it easy to find.\n- **Pre-mortems for numbers:** Before acting, ask ‘How could this be wrong?’ and assign someone to test that hypothesis.\n- **Two-speed reviews:** Fast weekly checks for drift + slower monthly deep dives for instrumentation and incentives.\n- **Reward truth-telling:** If surfacing bad news gets punished, you’ll get great-looking nonsense.\n\nIf you want, tell me the decision you’re trying to make and the signals you’re using—I’ll help you trim the noise without sanding off the truth.",{"id":157,"name":158,"type":159,"typeVersion":16,"position":160,"parameters":161,"category":164,"deletable":8,"connectable":8},"node_fallback_handoff","Handoff to Human Support","fallback",[129,122],{"handoffMessage":162,"departmentId":64,"departmentName":163},"Got it—this sounds like it needs human context. Share the branch, the metric or conversation thread, and what decision you’re trying to make. I’ll route this to a teammate for review.","Operations Support","terminal",[166,169,171,173,175,177,179,181,183,185,187,189,191,193],{"id":167,"source":35,"target":42,"sourceHandle":168,"targetHandle":31,"type":48},"conn_input_to_kb","output",{"id":170,"source":42,"target":54,"sourceHandle":168,"targetHandle":31,"type":48},"conn_kb_to_menu",{"id":172,"source":54,"target":85,"sourceHandle":168,"targetHandle":31,"type":48},"conn_menu_to_if_trust",{"id":174,"source":54,"target":95,"sourceHandle":168,"targetHandle":31,"type":48},"conn_menu_to_if_dirty",{"id":176,"source":54,"target":101,"sourceHandle":168,"targetHandle":31,"type":48},"conn_menu_to_if_auto",{"id":178,"source":54,"target":107,"sourceHandle":168,"targetHandle":31,"type":48},"conn_menu_to_if_compare",{"id":180,"source":54,"target":113,"sourceHandle":168,"targetHandle":31,"type":48},"conn_menu_to_if_culture",{"id":182,"source":54,"target":119,"sourceHandle":168,"targetHandle":31,"type":48},"conn_menu_to_if_human",{"id":184,"source":85,"target":125,"sourceHandle":168,"targetHandle":31,"type":48},"conn_if_trust_to_text",{"id":186,"source":95,"target":133,"sourceHandle":168,"targetHandle":31,"type":48},"conn_if_dirty_to_text",{"id":188,"source":101,"target":139,"sourceHandle":168,"targetHandle":31,"type":48},"conn_if_auto_to_text",{"id":190,"source":107,"target":145,"sourceHandle":168,"targetHandle":31,"type":48},"conn_if_compare_to_text",{"id":192,"source":113,"target":151,"sourceHandle":168,"targetHandle":31,"type":48},"conn_if_culture_to_text",{"id":194,"source":119,"target":157,"sourceHandle":168,"targetHandle":31,"type":48},"conn_if_human_to_fallback","automation",[27,28,29,30,197],"ops-leadership",[199,200],"Calypso Inbox","WhatsApp","intermediate","Calypso","2026-06-30T11:04:27.043Z","/en/workflows/signal-quality-checks-for-better-branch-decisions",{"en":204},{"title":9,"description":207,"ogDescription":208,"twitterDescription":209,"canonicalPath":204,"robots":210,"schemaType":211,"alternates":212},"Help teams spot dirty signals, validate branch numbers, and choose when automation needs human judgment—via a guided chat checklist.","A practical chat workflow to validate branch signals, catch dirty data early, and prevent confident wrong decisions. Includes KB first answers and guided checklists.","Turn messy branch signals into decision ready checks. KB first help, button menu, and practical guardrails to avoid confident wrong meetings.","index,follow","HowTo",[213],{"hreflang":6,"href":204}]