[{"data":1,"prerenderedAt":226},["ShallowReactive",2],{"/en/workflows/signal-to-decision-coach-branches-conversations-attribution":3},{"id":4,"slug":5,"locale":6,"translationGroupId":7,"localeSwitchApproved":8,"title":9,"description":10,"documentationMarkdown":11,"workflowJson":12,"category":205,"tags":206,"integrations":209,"difficulty":212,"author":213,"verified":33,"featured":33,"date":214,"modified":214,"icon":7,"imageSrc":7,"path":215,"alternates":216,"seo":217},"3b7c58ae-fdb2-41d9-a5cc-9732bdf2b0e6","signal-to-decision-coach-branches-conversations-attribution","en",null,true,"Signal-to-Decision Coach (Branches, Conversations, Attribution)","A practical guided chat that helps teams pressure-test branch numbers, spot dirty signals, and decide when automation is safe—before confident meetings go wrong.","## How it works\nThis workflow is a guided “decision hygiene” coach for branch and ops teams. It takes messy signals—branch counts, conversation summaries, attribution claims—and turns them into decision-shaped checks people can actually use in the moment.\n\nOperators pick the situation they’re in (numbers, dirty signal, automation vs judgment, comparisons, or culture). The workflow responds with concise guardrails and a few sharp questions designed to catch the most common failure mode: data that looks clean right up until someone makes a confident wrong decision.\n\n## Key features\n- Starts with a Knowledge Base pass to answer straightforward questions, then routes into a guided coaching menu.\n- Button-based routing so operators can quickly pick the decision they’re trying to make.\n- Practical checklists for “trustworthy vs polished noise” branch numbers.\n- A fast “dirty signal” screen to stop bad evidence before it becomes meeting-slide truth.\n- Optional human handoff for ambiguous cases or high-stakes decisions.\n\n## Step-by-step\n1. **Trigger:** A new inbound message starts the workflow.\n2. **Knowledge Base Policy:** The assistant first attempts a Knowledge Base answer; if it can’t, it continues into routing.\n3. **Guided menu:** The user chooses what they need help with:\n   1) Which branch numbers deserve trust\n   2) How to spot dirty signal before a meeting\n   3) When to trust automation vs human judgment\n   4) Comparing branches, conversations, and attribution\n   5) Building a signal culture that drives decisions\n   6) Talk to a human\n4. **Decision coaching:** The workflow sends a focused, operator-friendly response matching the selected option.\n5. **Escalation (optional):** If the user selects **Talk to a human**, the workflow hands off to the configured department.\n6. **If nothing matches:** The workflow asks the user to pick an option and returns to the menu.\n\n## Setup requirements\n- **Calypso Inbox / Web chat** connected to receive inbound messages.\n- (Optional) A configured **Knowledge Base** for the Knowledge Base Policy node to draw from.\n- (Optional) A defined **handoff department** (“Analytics Ops”) to receive escalations.\n- No additional credentials are required beyond your Calypso workspace access.",{"id":13,"teamId":14,"name":9,"version":15,"workflowVersion":16,"nodes":17,"connections":170,"routingEnabled":8,"active":33},"wf_signal_decision_coach_v1","calypso-public-library","1.0.0",1,[18,34,40,52,84,94,102,107,113,118,124,130,136,142,148,153,163],{"id":19,"name":20,"type":21,"typeVersion":16,"position":22,"parameters":24,"category":32,"deletable":33,"connectable":33},"node_flow_configs","Workflow settings","flow-configs",[23,23],80,{"name":9,"description":25,"tags":26,"triggerType":31},"Guided coaching to pressure-test branch metrics, spot dirty signals, and decide when automation is safe.",[27,28,29,30],"signal-quality","branch-metrics","decision-systems","attribution","input","policy",false,{"id":35,"name":36,"type":31,"typeVersion":16,"position":37,"parameters":39,"category":31,"deletable":33,"connectable":8},"node_input","Inbound message",[23,38],220,{},{"id":41,"name":42,"type":43,"typeVersion":16,"position":44,"parameters":46,"category":51,"deletable":8,"connectable":8},"node_kb_policy","Knowledge Base first","knowledge-base-policy",[45,38],300,{"enabled":8,"fallbackToRouting":8,"sticky":33,"stickyMode":47,"activationOpener":48,"personalization":50},"default",{"enabled":8,"instruction":49},"If the user is asking for definitions, known procedures, or established metrics rules, answer from the Knowledge Base. If the question is ambiguous or decision-shaped (\"Should we trust this?\") route to the coaching menu.",{"useContactName":8},"response",{"id":53,"name":54,"type":55,"typeVersion":16,"position":56,"parameters":58,"category":51,"deletable":8,"connectable":8},"node_menu","Choose what to sanity-check","interactive-message",[57,38],540,{"messageType":59,"headerText":60,"bodyText":61,"footerText":62,"sectionTitle":63,"buttons":64,"ctaDisplayText":83,"ctaUrl":83},"button","Signal-to-decision coach","Pick what you’re deciding. I’ll help you spot the “looks fine” traps before they become a confident slide.\n\nWhat do you want to sanity-check?","High stakes? Tap “Talk to a human”.","Options",[65,68,71,74,77,80],{"id":66,"title":67},"btn_trust_numbers","Trust numbers",{"id":69,"title":70},"btn_dirty_signal","Dirty signal",{"id":72,"title":73},"btn_automation_vs_judgment","Automation vs human",{"id":75,"title":76},"btn_compare_branches","Compare branches",{"id":78,"title":79},"btn_signal_culture","Signal culture",{"id":81,"title":82},"btn_human","Talk to a human","",{"id":85,"name":86,"type":87,"typeVersion":16,"position":88,"parameters":91,"category":93,"deletable":8,"connectable":8},"node_if_trust_numbers","If: Trustworthy branch numbers","if",[89,90],780,140,{"buttonId":66,"operator":92},"equals","routing",{"id":95,"name":96,"type":97,"typeVersion":16,"position":98,"parameters":100,"category":51,"deletable":8,"connectable":8},"node_msg_trust_numbers","Coach: Which numbers deserve trust","text-message",[99,90],1040,{"text":101},"Here’s the uncomfortable truth: the shiniest branch number is often the least useful.\n\n**Numbers that usually deserve trust (or at least a serious look):**\n1) **Counts tied to an operational event** (e.g., “cash-in completed”, “appointment attended”) with a clear timestamp.\n2) **Metrics that reconcile**: the total matches independent traces (system logs vs settlement vs inventory).\n3) **Stable definitions**: same inclusion rules week-to-week (no quiet re-labeling).\n4) **Denominators you can explain**: per staff-hour, per open-day, per eligible customer—pick one and stick to it.\n\n**Polished noise (common offenders):**\n- **Averages without distribution** (one weird day can cosplay as performance).\n- **Rate improvements from shrinking denominators** (“conversion up” because fewer leads were counted).\n- **Vanity completeness**: 99% filled fields that are 99% copy-paste.\n\n**Two questions to ask before you trust it:**\n- “If this number doubles tomorrow, what *physical thing* would we see in-branch?”\n- “What’s the most likely way this metric can be wrong while still looking correct?”\n\nWant to check a specific metric? Tell me the metric name and what decision it’s being used to justify.",{"id":103,"name":104,"type":87,"typeVersion":16,"position":105,"parameters":106,"category":93,"deletable":8,"connectable":8},"node_if_dirty_signal","If: Dirty signal screen",[89,38],{"buttonId":69,"operator":92},{"id":108,"name":109,"type":97,"typeVersion":16,"position":110,"parameters":111,"category":51,"deletable":8,"connectable":8},"node_msg_dirty_signal","Coach: Spot dirty signal before the meeting",[99,38],{"text":112},"Dirty signal is sneaky because it’s usually **consistent**, not random. Here’s a fast screen you can run before a meeting goes off the rails.\n\n**The 7-minute dirty-signal check:**\n1) **Definition check:** “What exactly is counted?” If two people give two answers, stop.\n2) **Time window trap:** “What changed in the last 2–4 weeks?” (policy, staffing, hours, channel mix, incentives).\n3) **Backfill & late-arrival:** Are events arriving late and being rewritten? If yes, don’t treat yesterday as final.\n4) **Missingness pattern:** If missing data clusters by branch, shift, or agent, it’s not missing at random—it’s a story.\n5) **One-system dependency:** If the metric relies on a single UI step, it measures compliance as much as reality.\n6) **Outlier audit:** Pull the top 5 and bottom 5 branches—do the extremes make operational sense?\n7) **Narrative smell test:** If the conclusion is too tidy (“everything improved everywhere”), it’s probably instrumentation.\n\n**Meeting-stopper line (useful and polite):**\n“Before we debate the story, can we confirm the definition, the time window, and whether the data is final or still backfilling?”\n\nTell me what claim the meeting wants to make (one sentence). I’ll point out the most likely failure mode.",{"id":114,"name":115,"type":87,"typeVersion":16,"position":116,"parameters":117,"category":93,"deletable":8,"connectable":8},"node_if_automation_vs_judgment","If: Automation vs judgment",[89,45],{"buttonId":72,"operator":92},{"id":119,"name":120,"type":97,"typeVersion":16,"position":121,"parameters":122,"category":51,"deletable":8,"connectable":8},"node_msg_automation_vs_judgment","Coach: When to trust automation",[99,45],{"text":123},"Automation is great at **repeatable judgments** on **stable signals**. It’s terrible at “novel situations that look like old situations.”\n\n**Trust automation when:**\n- The input is **instrumented** (events are logged, not inferred from text or memory).\n- The decision is **low-regret** (a wrong call is annoying, not expensive).\n- The rules are **auditable** (you can explain why it acted).\n- You have a **fallback path** (humans can override and learn from failures).\n\n**Require human judgment when:**\n- The metric is a proxy for something emotional/complex (customer intent, complaint severity).\n- Incentives exist to game the signal (targets, bonuses, rankings).\n- The situation is changing (new product, new script, staffing shifts, policy changes).\n- The cost of a wrong action is asymmetric (false positives hurt more than false negatives, or vice versa).\n\n**Practical rule:**\nIf the decision needs a paragraph to justify, don’t let a single metric make it. Use automation to **queue** the work; use humans to **close** the judgment.\n\nTell me what you want to automate (the action), and what signal would trigger it. I’ll tell you what usually breaks first.",{"id":125,"name":126,"type":87,"typeVersion":16,"position":127,"parameters":129,"category":93,"deletable":8,"connectable":8},"node_if_compare_branches","If: Compare branches & attribution",[89,128],380,{"buttonId":75,"operator":92},{"id":131,"name":132,"type":97,"typeVersion":16,"position":133,"parameters":134,"category":51,"deletable":8,"connectable":8},"node_msg_compare_branches","Coach: What teams misread in comparisons",[99,128],{"text":135},"Branch comparisons and attribution fail in the same way: **we compare outputs without comparing inputs**.\n\n**What teams repeatedly misread:**\n1) **Footfall / lead mix:** One branch gets easier customers; another gets the hard cases. Performance isn’t apples-to-apples.\n2) **Capacity constraints:** A branch at 100% utilization can look “worse” because it can’t absorb demand spikes.\n3) **Channel substitution:** Improvements may be customers switching channels, not true growth.\n4) **Measurement drift:** One branch is strict about logging; another is “creative.” You’re ranking compliance.\n5) **Attribution optimism:** “This campaign drove X” is often “X happened while the campaign existed.”\n\n**How to compare without lying to yourself:**\n- Normalize by a meaningful denominator (open hours, staffed hours, eligible customers).\n- Compare **distributions**, not just averages (median + 10th/90th is a great start).\n- Look for **leading indicators** that should move *before* the outcome (e.g., queue time before NPS changes).\n- Treat attribution as **directional** unless you have a clear counterfactual.\n\nSend me: (a) the two branches you’re comparing, (b) the metric, and (c) the decision riding on it. I’ll suggest the minimum fair comparison.",{"id":137,"name":138,"type":87,"typeVersion":16,"position":139,"parameters":141,"category":93,"deletable":8,"connectable":8},"node_if_signal_culture","If: Build a signal culture",[89,140],460,{"buttonId":78,"operator":92},{"id":143,"name":144,"type":97,"typeVersion":16,"position":145,"parameters":146,"category":51,"deletable":8,"connectable":8},"node_msg_signal_culture","Coach: Build a signal culture that decides",[99,140],{"text":147},"A healthy signal culture doesn’t produce more dashboards. It produces **faster, cleaner decisions**.\n\n**Three habits that work (and reduce slide-debate):**\n1) **Pre-commit the decision rule:** “If X moves by Y for Z weeks, we do A.” No improvising after seeing the data.\n2) **Make definitions boring and enforced:** One owner per metric; changes require a note in plain language.\n3) **Reward truth over winners:** If people get punished for bad news, they’ll manufacture good news.\n\n**Two lightweight rituals:**\n- **Metric obituary (monthly):** “Which metric misled us? How? What do we change?”\n- **Signal handoff checklist:** When a number enters a decision meeting, it travels with: definition, time window, completeness/backfill status, and known biases.\n\n**A little wit, because it helps:**\nIf a metric can’t survive one skeptical question, it shouldn’t be driving a budget.\n\nTell me what decision you’re trying to make this week. I’ll suggest the 2–3 signals worth trusting—and the 1–2 you should treat as vibes.",{"id":149,"name":150,"type":87,"typeVersion":16,"position":151,"parameters":152,"category":93,"deletable":8,"connectable":8},"node_if_human","If: Talk to a human",[89,57],{"buttonId":81,"operator":92},{"id":154,"name":155,"type":156,"typeVersion":16,"position":157,"parameters":158,"category":162,"deletable":8,"connectable":8},"node_fallback_handoff","Handoff to Analytics Ops","fallback",[99,57],{"handoffMessage":159,"departmentId":160,"departmentName":161},"Got it—looping in a human for this. Please share the decision you’re making, the metric(s) in question, and what would change if the number is wrong.","analytics-ops","Analytics Ops","terminal",{"id":164,"name":165,"type":97,"typeVersion":16,"position":166,"parameters":168,"category":51,"deletable":8,"connectable":8},"node_msg_no_match","No match: prompt again",[99,167],620,{"text":169},"I didn’t catch a selection. Please tap one of the buttons above. If you don’t see the buttons, send another message and I’ll show the menu again.",[171,175,177,179,182,185,187,189,191,193,195,197,199,201,203],{"id":172,"source":35,"target":41,"sourceHandle":173,"targetHandle":174,"type":47},"conn_input_to_kb","out","in",{"id":176,"source":41,"target":53,"sourceHandle":173,"targetHandle":174,"type":47},"conn_kb_to_menu",{"id":178,"source":53,"target":85,"sourceHandle":173,"targetHandle":174,"type":47},"conn_menu_to_if1",{"id":180,"source":85,"target":95,"sourceHandle":181,"targetHandle":174,"type":47},"conn_if1_true_to_msg","true",{"id":183,"source":85,"target":103,"sourceHandle":184,"targetHandle":174,"type":47},"conn_if1_false_to_if2","false",{"id":186,"source":103,"target":108,"sourceHandle":181,"targetHandle":174,"type":47},"conn_if2_true_to_msg",{"id":188,"source":103,"target":114,"sourceHandle":184,"targetHandle":174,"type":47},"conn_if2_false_to_if3",{"id":190,"source":114,"target":119,"sourceHandle":181,"targetHandle":174,"type":47},"conn_if3_true_to_msg",{"id":192,"source":114,"target":125,"sourceHandle":184,"targetHandle":174,"type":47},"conn_if3_false_to_if4",{"id":194,"source":125,"target":131,"sourceHandle":181,"targetHandle":174,"type":47},"conn_if4_true_to_msg",{"id":196,"source":125,"target":137,"sourceHandle":184,"targetHandle":174,"type":47},"conn_if4_false_to_if5",{"id":198,"source":137,"target":143,"sourceHandle":181,"targetHandle":174,"type":47},"conn_if5_true_to_msg",{"id":200,"source":137,"target":149,"sourceHandle":184,"targetHandle":174,"type":47},"conn_if5_false_to_if6",{"id":202,"source":149,"target":154,"sourceHandle":181,"targetHandle":174,"type":47},"conn_if6_true_to_handoff",{"id":204,"source":149,"target":164,"sourceHandle":184,"targetHandle":174,"type":47},"conn_if6_false_to_no_match","automation",[27,28,29,30,207,208],"data-hygiene","ops-coaching",[210,211],"Calypso Inbox","Web chat","intermediate","Calypso","2026-06-27T11:04:23.920Z","/en/workflows/signal-to-decision-coach-branches-conversations-attribution",{"en":215},{"title":218,"description":219,"ogDescription":220,"twitterDescription":221,"canonicalPath":215,"robots":222,"schemaType":223,"alternates":224},"Signal to Decision Coach for Branch Metrics","Guide teams to trust the right branch numbers, spot dirty signals, and choose automation vs judgment with a routed coaching chat.","A practical routed chat that pressure tests branch metrics, conversation signals, and attribution—so decisions are based on evidence, not polished noise.","Turn messy branch signals into decision ready checks. Spot dirty data early, compare branches sanely, and know when automation needs a human.","index,follow","HowTo",[225],{"hreflang":6,"href":215},1785947667744]