[{"data":1,"prerenderedAt":220},["ShallowReactive",2],{"/en/workflows/how-to-vet-branch-signals-before-you-decide":3},{"id":4,"slug":5,"locale":6,"translationGroupId":7,"localeSwitchApproved":8,"title":9,"description":10,"documentationMarkdown":11,"workflowJson":12,"category":200,"tags":201,"integrations":204,"difficulty":207,"author":208,"verified":32,"featured":32,"date":209,"modified":209,"icon":7,"imageSrc":7,"path":210,"alternates":211,"seo":212},"0108698b-5895-4149-895a-3cd02eb45e80","how-to-vet-branch-signals-before-you-decide","en",null,true,"How to Vet Branch Signals Before You Decide","A decision-support chat workflow that helps teams sanity-check branch metrics, spot dirty signals, and choose when to trust automation vs human judgment—before the meeting goes confidently wrong.","## How it works\nThis workflow turns “messy signals” into decision-ready guidance through a short, structured chat. It starts by enforcing your Knowledge Base policy (so terms like *active customer*, *qualified lead*, or *branch conversion* mean what your team thinks they mean), then offers a menu of common decision traps.\n\nOperators use it when someone drops a shiny dashboard number, a confident attribution claim, or a “the conversations say…” conclusion right before a decision. The workflow responds with practical checks that catch polished noise early—without scrubbing away inconvenient truth.\n\n## Key features\n- Applies a **Knowledge Base policy** first, keeping definitions and guidance consistent across teams.\n- Uses a **button-based triage menu** so people can quickly pick the decision risk they’re facing.\n- Provides **decision-shaped checklists** for branch metrics trust, dirty signal detection, automation vs judgment, comparisons, and signal culture.\n- Includes a **human handoff path** to route complex cases to Analytics Ops.\n\n## Step-by-step\n1. **Trigger:** A user starts the workflow (Input).\n2. **Grounding step:** The workflow applies the **Knowledge Base policy** to keep responses aligned with your internal metric definitions and preferred guidance.\n3. **Choose your situation:** The user gets a button menu with six options (Interactive Message).\n4. **Route by selection:** The workflow checks which button was pressed via a sequence of **IF** nodes.\n5. **Deliver guidance:** The workflow sends a tailored message (Text Message) for:\n   1) which branch numbers deserve trust vs polished noise,\n   2) how to spot dirty signal before the meeting,\n   3) when to trust automation vs human judgment,\n   4) what teams misread when comparing branches/conversations/attribution,\n   5) how to build a signal culture that enables decisions.\n6. **Escalate when needed:** If the user selects **“Route to Analytics Ops”** (or no selection matches), the workflow hands off to **Analytics Ops** (Fallback).\n\n## Setup requirements\n- **Calypso Knowledge Base** populated with your metrics glossary and reporting rules (recommended so “truth checks” match your org’s definitions).\n- A messaging channel that supports interactive buttons (for example, **WhatsApp**) connected to Calypso.\n- No additional credentials are required beyond your normal Calypso channel and Knowledge Base access.",{"id":13,"teamId":14,"name":9,"version":15,"workflowVersion":16,"nodes":17,"connections":166,"routingEnabled":8,"active":32},"wf_signal_vetting_001","calypso-public-library","1.0.0",1,[18,33,39,51,82,91,100,106,112,118,124,130,137,143,150,156],{"id":19,"name":20,"type":21,"typeVersion":16,"position":22,"parameters":24,"category":31,"deletable":32,"connectable":32},"node_flow_configs","Workflow settings","flow-configs",[23,23],80,{"name":9,"description":25,"tags":26,"triggerType":30},"Decision-support triage for branch metrics, dirty signals, and automation judgment calls.",[27,28,29],"signal-quality","branch-metrics","decision-systems","input","policy",false,{"id":34,"name":35,"type":30,"typeVersion":16,"position":36,"parameters":38,"category":30,"deletable":32,"connectable":8},"node_input","Start",[23,37],220,{},{"id":40,"name":41,"type":42,"typeVersion":16,"position":43,"parameters":45,"category":50,"deletable":8,"connectable":8},"node_kb_policy","Knowledge base policy","knowledge-base-policy",[44,37],300,{"enabled":8,"fallbackToRouting":8,"sticky":8,"stickyMode":46,"activationOpener":47,"personalization":49},"default",{"enabled":8,"instruction":48},"Use our Knowledge Base definitions first. Keep advice practical, decision-shaped, and non-academic. If a user asks for numbers we don’t have, give checks to validate them rather than inventing data.",{"useContactName":32},"response",{"id":52,"name":53,"type":54,"typeVersion":16,"position":55,"parameters":57,"category":50,"deletable":8,"connectable":8},"node_menu","Decision triage menu","interactive-message",[56,37],540,{"messageType":58,"headerText":59,"bodyText":60,"footerText":61,"sectionTitle":62,"buttons":63,"ctaDisplayText":62,"ctaUrl":62},"button","Signal triage","Pick what you’re trying to decide (or what feels ‘too clean’ to trust). I’ll give you a fast set of checks before the meeting runs away with the story.","Tip: if a number looks flawless, ask what it’s blind to.","",[64,67,70,73,76,79],{"id":65,"title":66},"trust_numbers","Trust branch nums",{"id":68,"title":69},"spot_dirty_signal","Spot dirty signal",{"id":71,"title":72},"automation_vs_judgment","Automation vs human",{"id":74,"title":75},"compare_branches","Compare & attribute",{"id":77,"title":78},"signal_culture","Signal culture",{"id":80,"title":81},"talk_to_analytics_ops","Analytics Ops",{"id":83,"name":84,"type":85,"typeVersion":16,"position":86,"parameters":88,"category":90,"deletable":8,"connectable":8},"node_if_trust","If: trust numbers","if",[87,37],760,{"buttonId":65,"operator":89},"equals","routing",{"id":92,"name":93,"type":94,"typeVersion":16,"position":95,"parameters":98,"category":50,"deletable":8,"connectable":8},"node_txt_trust","Trust check: branch numbers","text-message",[96,97],990,140,{"text":99},"Here’s the fast way to tell *decision-grade* branch numbers from polished noise:\n\n1) **Definition lock:** Can two people independently explain what’s counted (and what’s excluded)? If not, it’s a vibe, not a metric.\n2) **Denominator reality:** Ask “out of what?” Branches can ‘improve’ by shrinking the denominator (fewer opportunities) while the chart smiles.\n3) **Volatility test:** If the branch volume is low, week-to-week changes are mostly weather. Demand confidence bands or at least a 4–8 week view.\n4) **Lag & revision:** Does the number get corrected later (chargebacks, returns, late CRM updates)? If yes, don’t treat today’s value as truth.\n5) **Incentive smell:** If someone’s bonus depends on it, assume it will be gamed—usually unintentionally at first.\n\nIf you tell me the metric name and time window, I’ll suggest the 2–3 most common failure modes to check first.",{"id":101,"name":102,"type":85,"typeVersion":16,"position":103,"parameters":105,"category":90,"deletable":8,"connectable":8},"node_if_dirty","If: spot dirty signal",[87,104],340,{"buttonId":68,"operator":89},{"id":107,"name":108,"type":94,"typeVersion":16,"position":109,"parameters":110,"category":50,"deletable":8,"connectable":8},"node_txt_dirty","Dirty signal checks",[96,44],{"text":111},"Dirty data rarely looks dirty. It looks *plausible*—until it ruins a confident decision.\n\nRun these checks before the meeting:\n\n1) **Edge-case scan:** Look at the weird tails (top 1%, bottom 1%). Dirty pipelines fail at the edges first.\n2) **Duplicate & identity drift:** Same customer counted twice? Branch transfers? Reopened cases? Quiet duplication is a classic ‘growth’ generator.\n3) **Timestamp sanity:** Sudden spikes at midnight, end-of-month cliffs, or perfect hourly patterns usually mean batching, not behavior.\n4) **Source breakpoints:** Did tracking, scripts, forms, or call routing change? If yes, treat before/after comparisons as suspect.\n5) **Missingness map:** What’s not being captured (unknown channel, untagged conversations)? Missing data is not random—it's biased.\n\nIf a chart looks too smooth, ask what got rounded, bucketed, or filtered away.",{"id":113,"name":114,"type":85,"typeVersion":16,"position":115,"parameters":117,"category":90,"deletable":8,"connectable":8},"node_if_auto","If: automation vs judgment",[87,116],460,{"buttonId":71,"operator":89},{"id":119,"name":120,"type":94,"typeVersion":16,"position":121,"parameters":122,"category":50,"deletable":8,"connectable":8},"node_txt_auto","Automation vs judgment guidance",[96,116],{"text":123},"When to trust automation (and when to keep a human in the loop):\n\n**Trust automation when:**\n- The decision is **frequent**, **reversible**, and the cost of being wrong is low.\n- Inputs are stable and definitions don’t change every quarter.\n- You can measure outcomes quickly (tight feedback loop).\n\n**Require human judgment when:**\n- The decision is **one-way** (pricing, staffing, closures) or reputationally sensitive.\n- The model is fed by **proxy signals** (engagement ≠ intent) or missing key context.\n- You’re seeing regime change: new product, new incentive, new channel, new policy.\n\nPractical rule: automate the *recommendation*, not the *accountability*. Make someone sign the exception list.",{"id":125,"name":126,"type":85,"typeVersion":16,"position":127,"parameters":129,"category":90,"deletable":8,"connectable":8},"node_if_compare","If: compare branches & attribution",[87,128],580,{"buttonId":74,"operator":89},{"id":131,"name":132,"type":94,"typeVersion":16,"position":133,"parameters":135,"category":50,"deletable":8,"connectable":8},"node_txt_compare","Comparison & attribution traps",[96,134],620,{"text":136},"What teams repeatedly misread when comparing branches, conversations, and attribution:\n\n1) **Mix shift:** Branch A isn’t ‘better’—it got easier leads. Compare like-with-like (segment, product, channel, seasonality).\n2) **Capacity effects:** A staffed branch converts better *until* it saturates. Volume and conversion often trade places.\n3) **Conversation bias:** The loudest conversations aren’t the most important. Escalations overrepresent pain; silent churn is invisible.\n4) **Attribution as a story:** Last-touch is tidy, not true. If channels overlap, treat attribution as a *hypothesis generator*, not a scoreboard.\n5) **Local process changes:** A new script, referral habit, or manager can move metrics without any ‘market change’ at all.\n\nIf you want one high-leverage move: build a comparison table that shows **volume, conversion, and outcome lag** side-by-side. Most bad calls ignore at least one of the three.",{"id":138,"name":139,"type":85,"typeVersion":16,"position":140,"parameters":142,"category":90,"deletable":8,"connectable":8},"node_if_culture","If: build signal culture",[87,141],700,{"buttonId":77,"operator":89},{"id":144,"name":145,"type":94,"typeVersion":16,"position":146,"parameters":148,"category":50,"deletable":8,"connectable":8},"node_txt_culture","Signal culture playbook",[96,147],780,{"text":149},"A signal culture that helps decisions happen (not just slides) looks like this:\n\n1) **One metric, one owner:** If nobody owns the definition, everyone owns the confusion.\n2) **Decision logs:** Record the decision, the signals used, and what you expected. You can’t learn from outcomes you didn’t predict.\n3) **Red-team the number:** Make it normal to ask “how could this be wrong?” before asking “how do we present it?”\n4) **Guardrail metrics:** Pair every ‘go’ metric with a ‘don’t break the business’ metric (quality, complaints, returns, SLA).\n5) **Kill vanity fast:** If a metric can go up while customer reality goes down, label it *non-decisioning*.\n\nWitty but useful rule: if the dashboard answers every question instantly, it’s probably answering the easy ones.",{"id":151,"name":152,"type":85,"typeVersion":16,"position":153,"parameters":155,"category":90,"deletable":8,"connectable":8},"node_if_handoff","If: route to Analytics Ops",[87,154],820,{"buttonId":80,"operator":89},{"id":157,"name":158,"type":159,"typeVersion":16,"position":160,"parameters":162,"category":165,"deletable":8,"connectable":8},"node_fallback","Handoff to Analytics Ops","fallback",[96,161],880,{"handoffMessage":163,"departmentId":164,"departmentName":81},"Got it. I’m routing this to Analytics Ops. Please share: (1) the metric name, (2) time window, (3) branch(es), (4) what decision you’re making, and (5) the chart/source link if you have it.","analytics-ops","terminal",[167,170,172,174,177,180,182,184,186,188,190,192,194,196,198],{"id":168,"source":34,"target":40,"sourceHandle":169,"targetHandle":169,"type":169},"conn_input_to_kb","main",{"id":171,"source":40,"target":52,"sourceHandle":169,"targetHandle":169,"type":169},"conn_kb_to_menu",{"id":173,"source":52,"target":83,"sourceHandle":169,"targetHandle":169,"type":169},"conn_menu_to_if_trust",{"id":175,"source":83,"target":92,"sourceHandle":176,"targetHandle":169,"type":169},"conn_if_trust_true_to_txt","true",{"id":178,"source":83,"target":101,"sourceHandle":179,"targetHandle":169,"type":169},"conn_if_trust_false_to_if_dirty","false",{"id":181,"source":101,"target":107,"sourceHandle":176,"targetHandle":169,"type":169},"conn_if_dirty_true_to_txt",{"id":183,"source":101,"target":113,"sourceHandle":179,"targetHandle":169,"type":169},"conn_if_dirty_false_to_if_auto",{"id":185,"source":113,"target":119,"sourceHandle":176,"targetHandle":169,"type":169},"conn_if_auto_true_to_txt",{"id":187,"source":113,"target":125,"sourceHandle":179,"targetHandle":169,"type":169},"conn_if_auto_false_to_if_compare",{"id":189,"source":125,"target":131,"sourceHandle":176,"targetHandle":169,"type":169},"conn_if_compare_true_to_txt",{"id":191,"source":125,"target":138,"sourceHandle":179,"targetHandle":169,"type":169},"conn_if_compare_false_to_if_culture",{"id":193,"source":138,"target":144,"sourceHandle":176,"targetHandle":169,"type":169},"conn_if_culture_true_to_txt",{"id":195,"source":138,"target":151,"sourceHandle":179,"targetHandle":169,"type":169},"conn_if_culture_false_to_if_handoff",{"id":197,"source":151,"target":157,"sourceHandle":176,"targetHandle":169,"type":169},"conn_if_handoff_true_to_fb",{"id":199,"source":151,"target":157,"sourceHandle":179,"targetHandle":169,"type":169},"conn_if_handoff_false_to_fb","automation",[27,28,29,202,203],"data-hygiene","automation-governance",[205,206],"WhatsApp","Calypso Knowledge Base","intermediate","Calypso","2026-08-02T11:03:28.116Z","/en/workflows/how-to-vet-branch-signals-before-you-decide",{"en":210},{"title":9,"description":213,"ogDescription":214,"twitterDescription":215,"canonicalPath":210,"robots":216,"schemaType":217,"alternates":218},"Guide teams through quick signal checks: which branch numbers to trust, how to spot dirty data, and when automation needs human judgment.","A practical chat workflow to sanity check branch metrics, catch dirty signals early, and decide when to trust automation vs human judgment.","Stop polished noise from driving confident wrong decisions. Triage branch signals, attribution claims, and automation risk with quick checks.","index,follow","HowTo",[219],{"hreflang":6,"href":210},1785947666103]