[{"data":1,"prerenderedAt":206},["ShallowReactive",2],{"/en/workflows/branch-signal-trust-triage":3},{"id":4,"slug":5,"locale":6,"translationGroupId":7,"localeSwitchApproved":8,"title":9,"description":10,"documentationMarkdown":11,"workflowJson":12,"category":187,"tags":188,"integrations":191,"difficulty":193,"author":194,"verified":33,"featured":33,"date":195,"modified":195,"icon":7,"imageSrc":7,"path":196,"alternates":197,"seo":198},"3bd23c4e-0dab-4a7c-8b9d-ce2382750a3c","branch-signal-trust-triage","en",null,true,"Branch Signal Trust Triage","A decision-focused assistant that helps branch teams sanity-check metrics, spot dirty signals, and choose when automation is safe—before a confident meeting goes confidently wrong.","## How it works\nThis workflow is a practical “signal triage” assistant for branch leaders and analysts. It helps teams quickly test whether a number deserves trust, spot the most common forms of dirty signal, and decide when automation is appropriate versus when human judgment is still the safety belt.\n\nIt’s designed for the messy reality: conversations, branch-level events, attribution, and dashboards that look clean right up until they’re used to justify a confident wrong decision. Users pick a decision-shaped topic, and the workflow delivers a tight checklist and next steps they can use immediately.\n\n## Key features\n- Knowledge-base answers first, so common questions get fast, consistent guidance before routing\n- A button-based menu that nudges users into decision-shaped prompts (trust, dirty signal, automation, comparisons, culture)\n- Clear “what to check next” playbooks that fit real branch operations (definitions, sampling, incentives, mix-shift)\n- Human handoff path when the user’s situation is too specific (or too risky) for a generic answer\n\n## Step-by-step\n1. **Trigger:** A user starts the conversation (incoming message).\n2. **Knowledge base pass:** The assistant answers from your knowledge base when it can; otherwise it proceeds to the triage menu.\n3. **User chooses a path:** The workflow shows a button menu:\n   - Trust this branch number\n   - Spot dirty signal\n   - Automation vs judgment\n   - Compare branches & attribution\n   - Build a signal culture\n4. **Decision playbook response:** Based on the button clicked, the workflow sends a targeted checklist and practical next steps.\n5. **Fallback to a person:** If none of the menu routes match (or inputs are inconsistent), the workflow hands off to a human team for follow-up.\n\n## Setup requirements\n- A published **Knowledge Base** in Calypso (recommended so the knowledge-base step can answer repeat questions)\n- A configured **routing destination/department** for the fallback handoff (optional but recommended)\n- No external credentials are required for this workflow",{"id":13,"teamId":14,"name":9,"version":15,"workflowVersion":16,"nodes":17,"connections":157,"routingEnabled":8,"active":33},"wf_branch_signal_trust_triage_v1","calypso-public-library","1.0.0",1,[18,34,40,52,80,90,98,104,110,116,122,128,134,140,146],{"id":19,"name":20,"type":21,"typeVersion":16,"position":22,"parameters":25,"category":32,"deletable":33,"connectable":33},"node_flow_configs","Workflow settings","flow-configs",[23,24],60,40,{"name":9,"description":26,"tags":27,"triggerType":31},"Decision-shaped guidance for branch metrics: trust checks, dirty-signal detection, automation guardrails, comparison pitfalls, and culture habits.",[28,29,30],"signal-quality","branch-metrics","decision-making","input","policy",false,{"id":35,"name":36,"type":31,"typeVersion":16,"position":37,"parameters":39,"category":31,"deletable":33,"connectable":8},"node_input","Incoming message",[23,38],180,{},{"id":41,"name":42,"type":43,"typeVersion":16,"position":44,"parameters":46,"category":51,"deletable":8,"connectable":8},"node_kb_policy","Knowledge base answers","knowledge-base-policy",[45,38],300,{"enabled":8,"fallbackToRouting":8,"sticky":8,"stickyMode":47,"activationOpener":48,"personalization":50},"default",{"enabled":8,"instruction":49},"Help the user make a trustworthy decision from messy branch signals. Be practical and plainspoken. Use checklists, warn about common traps (definitions drifting, small samples, mix-shift, incentives, attribution). Ask for clarifying details only if needed.",{"useContactName":8},"response",{"id":53,"name":54,"type":55,"typeVersion":16,"position":56,"parameters":58,"category":51,"deletable":8,"connectable":8},"node_menu","Signal triage menu","interactive-message",[57,38],560,{"messageType":59,"headerText":60,"bodyText":61,"footerText":62,"sectionTitle":63,"buttons":64,"ctaDisplayText":63,"ctaUrl":63},"button","Signal triage","Pick what you’re trying to decide. I’ll help you sanity-check the signal before it becomes a very confident slide.","Choose one:","",[65,68,71,74,77],{"id":66,"title":67},"BTN_TRUST_NUMBER","Trust this number",{"id":69,"title":70},"BTN_DIRTY_SIGNAL","Spot dirty signal",{"id":72,"title":73},"BTN_AUTOMATION_JUDGMENT","Auto vs judgment",{"id":75,"title":76},"BTN_COMPARE_BRANCHES","Branch comparisons",{"id":78,"title":79},"BTN_SIGNAL_CULTURE","Signal culture",{"id":81,"name":82,"type":83,"typeVersion":16,"position":84,"parameters":87,"category":89,"deletable":8,"connectable":8},"node_if_trust","Route: Trust check","if",[85,86],820,110,{"buttonId":66,"operator":88},"equals","routing",{"id":91,"name":92,"type":93,"typeVersion":16,"position":94,"parameters":96,"category":51,"deletable":8,"connectable":8},"node_text_trust","Playbook: Which numbers deserve trust","text-message",[95,86],1080,{"text":97},"Use this trust check before you let a number steer a decision:\n\n1) Definition stability (most failures start here)\n- Has the metric definition changed in the last 30–90 days? If yes, it’s not a trend—it's a rename.\n- Same numerator/denominator across branches? Watch “helpful” local interpretations.\n\n2) Source reliability\n- Is it system-recorded (POS/event log) or hand-entered? Manual isn’t bad—just fragile and incentive-sensitive.\n- Can you trace 3–5 records end-to-end? If you can’t audit it, don’t worship it.\n\n3) Volume & noise\n- What’s the sample size per branch per week? Small-N makes heroes and villains at random.\n- Look for confidence bands, not just rankings.\n\n4) Incentives & gaming\n- Is anyone rewarded for this number? If yes, assume it will get ‘improved’ creatively.\n- Check for suspicious smoothness or sudden step-changes after a new target.\n\n5) Operational reality check\n- Does the number move when the real-world driver moves (staffing, footfall, closures, promotions)?\n- If it disagrees with reality, the number is usually wrong—reality is annoyingly consistent.\n\nIf you share the metric name + how it’s captured, I can tell you the top 2 likely failure modes.",{"id":99,"name":100,"type":83,"typeVersion":16,"position":101,"parameters":103,"category":89,"deletable":8,"connectable":8},"node_if_dirty","Route: Dirty signal",[85,102],220,{"buttonId":69,"operator":88},{"id":105,"name":106,"type":93,"typeVersion":16,"position":107,"parameters":108,"category":51,"deletable":8,"connectable":8},"node_text_dirty","Playbook: Spot dirty signal fast",[95,102],{"text":109},"Dirty signal usually looks *reasonable*—that’s why it’s dangerous. Here are fast tells:\n\nThe ‘too clean’ problem\n- Perfectly smooth week-to-week lines on something that should be lumpy (people, branches, conversations). Suspicious.\n- Every branch improves at the same time by the same amount: that’s often a reporting change, not performance.\n\nThe ‘what changed?’ checklist (ask this before the meeting)\n- Did the dashboard logic change? (filters, dedupe rules, attribution window, timezone)\n- Did the workflow/process change? (new form, new script, new CRM field, new incentive)\n- Did the population change? (mix-shift, new segments, closures, staffing)\n\nClassic failure modes\n- Duplicates: the same customer counted twice (or five times) across channels.\n- Missingness: a branch ‘looks better’ because it stopped logging the messy cases.\n- Survivorship bias: only completed journeys counted; drop-offs disappear.\n\nPractical move: pick 10 recent records and do a reality audit. If 2–3 are ambiguous, the metric is not decision-grade yet.",{"id":111,"name":112,"type":83,"typeVersion":16,"position":113,"parameters":115,"category":89,"deletable":8,"connectable":8},"node_if_auto","Route: Automation vs judgment",[85,114],330,{"buttonId":72,"operator":88},{"id":117,"name":118,"type":93,"typeVersion":16,"position":119,"parameters":120,"category":51,"deletable":8,"connectable":8},"node_text_auto","Playbook: When to trust automation",[95,114],{"text":121},"Automation is great at repetition. It’s terrible at noticing you’ve entered a new situation.\n\nTrust automation when:\n- Stakes are low or reversible (you can undo tomorrow)\n- Volume is high (patterns are real, not one-off drama)\n- The environment is stable (same offer, same process, same definitions)\n- You have guardrails (alerts on anomalies, thresholds, and override paths)\n\nRequire human judgment when:\n- The sample is small (branch-level weekly changes, niche segments)\n- Incentives are strong (targets, bonuses, public rankings)\n- You’re launching/changing something (new campaign, new script, new channel)\n- Consequences are high (compliance, pricing, staffing cuts)\n\nSimple rule: automate the *default*, but force a human review for edge-cases and step-changes. The first place things go wrong is right after a ‘small’ process change nobody wrote down.",{"id":123,"name":124,"type":83,"typeVersion":16,"position":125,"parameters":127,"category":89,"deletable":8,"connectable":8},"node_if_compare","Route: Compare branches & attribution",[85,126],440,{"buttonId":75,"operator":88},{"id":129,"name":130,"type":93,"typeVersion":16,"position":131,"parameters":132,"category":51,"deletable":8,"connectable":8},"node_text_compare","Playbook: What teams misread in comparisons",[95,126],{"text":133},"Branch comparisons go wrong in predictable ways. Here’s the short list:\n\n1) Mix-shift masquerading as performance\n- Branch A isn’t ‘better’—it may just get easier customers, different product mix, or higher footfall.\n- Normalize where possible (per visit, per qualified lead, per staffed hour).\n\n2) Small samples create fake leaders\n- Ranking 20 branches on thin weekly data mostly ranks randomness.\n- Look at longer windows, or show uncertainty (even a simple “low/medium/high confidence”).\n\n3) Attribution is not truth, it’s a rule\n- Changing the window or dedupe rules can ‘move’ revenue without changing reality.\n- If attribution changes, re-baseline and communicate it like a definition change.\n\n4) Uneven logging\n- The best-looking branch may be the one that logs the least.\n- Compare data completeness first (missing fields, late entries, duplicates).\n\nIf you tell me what you’re comparing (conversion, revenue, CSAT, conversations), I can suggest the right denominator and the two most likely confounders.",{"id":135,"name":136,"type":83,"typeVersion":16,"position":137,"parameters":139,"category":89,"deletable":8,"connectable":8},"node_if_culture","Route: Signal culture",[85,138],550,{"buttonId":78,"operator":88},{"id":141,"name":142,"type":93,"typeVersion":16,"position":143,"parameters":144,"category":51,"deletable":8,"connectable":8},"node_text_culture","Playbook: Build a signal culture (not a slide culture)",[95,138],{"text":145},"A strong signal culture makes decisions easier—not just prettier decks.\n\nHabits that work:\n- One owner per metric: definition, data source, known caveats, and change log.\n- Decision logs: write what you decided, what signals you used, and what would change your mind.\n- “What would have to be true?” at the start of meetings (stops motivated reasoning early).\n- Pre-mortems for big calls: list how the decision could be wrong *before* committing.\n- Show the messy bits: missingness, duplicates, and edge-cases. Cleaning away the truth is not hygiene.\n\nMeeting rule with teeth: no branch ranking without (a) a denominator, (b) sample size, and (c) a note on data completeness. Confidence without those is just enthusiasm.",{"id":147,"name":148,"type":149,"typeVersion":16,"position":150,"parameters":152,"category":156,"deletable":8,"connectable":8},"node_fallback","Handoff to human support","fallback",[95,151],670,{"handoffMessage":153,"departmentId":154,"departmentName":155},"This looks specific enough that a human should sanity-check it with you. Share the metric name, timeframe, and what decision it’s meant to drive, and we’ll route you to the right teammate.","dept_data_ops","Data & Reporting","terminal",[158,161,163,165,168,171,173,175,177,179,181,183,185],{"id":159,"source":35,"target":41,"sourceHandle":160,"targetHandle":160,"type":160},"conn_input_to_kb","main",{"id":162,"source":41,"target":53,"sourceHandle":160,"targetHandle":160,"type":160},"conn_kb_to_menu",{"id":164,"source":53,"target":81,"sourceHandle":160,"targetHandle":160,"type":160},"conn_menu_to_if_trust",{"id":166,"source":81,"target":91,"sourceHandle":167,"targetHandle":160,"type":160},"conn_if_trust_true_to_text","true",{"id":169,"source":81,"target":99,"sourceHandle":170,"targetHandle":160,"type":160},"conn_if_trust_false_to_if_dirty","false",{"id":172,"source":99,"target":105,"sourceHandle":167,"targetHandle":160,"type":160},"conn_if_dirty_true_to_text",{"id":174,"source":99,"target":111,"sourceHandle":170,"targetHandle":160,"type":160},"conn_if_dirty_false_to_if_auto",{"id":176,"source":111,"target":117,"sourceHandle":167,"targetHandle":160,"type":160},"conn_if_auto_true_to_text",{"id":178,"source":111,"target":123,"sourceHandle":170,"targetHandle":160,"type":160},"conn_if_auto_false_to_if_compare",{"id":180,"source":123,"target":129,"sourceHandle":167,"targetHandle":160,"type":160},"conn_if_compare_true_to_text",{"id":182,"source":123,"target":135,"sourceHandle":170,"targetHandle":160,"type":160},"conn_if_compare_false_to_if_culture",{"id":184,"source":135,"target":141,"sourceHandle":167,"targetHandle":160,"type":160},"conn_if_culture_true_to_text",{"id":186,"source":135,"target":147,"sourceHandle":170,"targetHandle":160,"type":160},"conn_if_culture_false_to_fallback","automation",[28,29,30,189,190],"data-hygiene","automation-guardrails",[192],"Calypso Messaging","intermediate","Calypso","2026-04-30T11:03:43.933Z","/en/workflows/branch-signal-trust-triage",{"en":196},{"title":9,"description":199,"ogDescription":200,"twitterDescription":201,"canonicalPath":196,"robots":202,"schemaType":203,"alternates":204},"Turn messy branch signals into decision ready checks: verify metric trust, spot dirty data early, and choose when automation needs human judgment.","A practical signal triage assistant: detect polished noise, validate branch metrics, avoid attribution traps, and decide when automation is safe.","Decision ready signal checks for branch teams—trust tests, dirty signal spotting, automation guardrails, and comparison pitfalls in one workflow.","index,follow","HowTo",[205],{"hreflang":6,"href":196},1778614431229]