[{"data":1,"prerenderedAt":246},["ShallowReactive",2],{"/en/workflows/branch-signal-trust-filters":3},{"id":4,"slug":5,"locale":6,"translationGroupId":7,"localeSwitchApproved":8,"title":9,"description":10,"documentationMarkdown":11,"workflowJson":12,"category":226,"tags":227,"integrations":231,"difficulty":233,"author":234,"verified":33,"featured":33,"date":235,"modified":235,"icon":7,"imageSrc":7,"path":236,"alternates":237,"seo":238},"47baef89-170f-4815-8f2d-aac5cabf0063","branch-signal-trust-filters","en",null,true,"Branch Signal Trust Filters","A guided chat workflow that helps leaders and analysts separate trustworthy branch signals from polished noise, spot dirty data early, and decide when automation should (and shouldn’t) drive decisions.","## How it works\nThis workflow turns “we have a lot of signals” into “we can make a decision without embarrassing ourselves later.” It starts with a knowledge base answer when possible, then offers a simple menu of decision-shaped topics that diagnose which branch numbers to trust, how to spot dirty signal, and what teams misread when comparing branches and attribution.\n\nIt’s designed for the moment right before a confident meeting goes off the rails: you’ll get short, practical guardrails and checklists—without scrubbing away the messy truths that often explain what’s really happening.\n\n## Key features\n- Knowledge-base-first responses, with automatic fall-through to a guided menu when the KB can’t help.\n- Button-based routing to keep conversations decision-focused (trust, hygiene, automation vs judgment, comparisons, culture).\n- Repeatable “filters” for branch metrics that look clean but behave suspiciously.\n- A built-in escape hatch to hand off to a human team when the situation needs judgment.\n\n## Step-by-step\n1. **Trigger:** A user sends a message to start the conversation.\n2. **Knowledge Base policy:** Calypso attempts to answer from your Knowledge Base. If confidence is low or the question needs triage, it continues to the menu.\n3. **Interactive menu:** The user picks what they’re trying to decide (e.g., which numbers to trust, dirty signal checks, automation vs judgment).\n4. **Routing:** The workflow matches the selected button and sends the corresponding decision guidance.\n5. **Loop:** After each answer, the menu is shown again so the user can run another filter.\n6. **Human handoff (optional):** If the user selects “Talk to a person,” the workflow routes to your designated department.\n\n## Setup requirements\n- **Calypso Knowledge Base:** Populate a few short articles (or snippets) on branch metric definitions, attribution rules, and known data caveats. No external credentials required.\n- **Calypso Department for handoff (optional but recommended):** Create/identify an “Insights Ops” (or similar) department and paste its ID into the fallback node.",{"id":13,"teamId":14,"name":9,"version":15,"workflowVersion":16,"nodes":17,"connections":187,"routingEnabled":8,"active":33},"wf_branch_signal_trust_filters","calypso-public-library","1.0.0",1,[18,34,41,53,87,97,105,110,116,122,128,134,140,146,152,158,164,170,180],{"id":19,"name":20,"type":21,"typeVersion":16,"position":22,"parameters":25,"category":32,"deletable":33,"connectable":33},"fc_1","Workflow settings","flow-configs",[23,24],-420,-40,{"name":9,"description":26,"tags":27,"triggerType":31},"Guided decision filters for branch signals: trust, hygiene, comparisons, and automation vs judgment.",[28,29,30],"signal-quality","branch-metrics","decision-systems","input","policy",false,{"id":35,"name":36,"type":31,"typeVersion":16,"position":37,"parameters":40,"category":31,"deletable":33,"connectable":8},"in_1","Inbound message",[38,39],-240,140,{},{"id":42,"name":43,"type":44,"typeVersion":16,"position":45,"parameters":47,"category":52,"deletable":8,"connectable":8},"kb_1","Knowledge Base: answer or route","knowledge-base-policy",[46,39],-20,{"enabled":8,"fallbackToRouting":8,"sticky":8,"stickyMode":48,"activationOpener":49,"personalization":51},"ai_sticky_release",{"enabled":8,"instruction":50},"Use the Knowledge Base to answer questions about branch metrics, attribution, and signal reliability. If the user’s situation is unclear or the KB doesn’t directly apply, proceed to the menu and help them choose a decision filter.",{"useContactName":8},"response",{"id":54,"name":55,"type":56,"typeVersion":16,"position":57,"parameters":59,"category":52,"deletable":8,"connectable":8},"im_1","Choose a decision filter","interactive-message",[58,39],220,{"messageType":60,"headerText":61,"bodyText":62,"footerText":63,"sectionTitle":64,"buttons":65,"ctaDisplayText":64,"ctaUrl":64},"button","Make branch signals decision-ready","Pick what you’re trying to decide. I’ll give you a quick filter that catches the usual failure modes before you bet a meeting on the wrong number.","Tip: Always-up charts often mean the definition changed.","",[66,69,72,75,78,81,84],{"id":67,"title":68},"trust_numbers","Trust numbers",{"id":70,"title":71},"spot_dirty","Spot dirty signal",{"id":73,"title":74},"auto_vs_human","Auto vs judgment",{"id":76,"title":77},"messy_evidence","Use messy evidence",{"id":79,"title":80},"compare_branches","Compare branches",{"id":82,"title":83},"signal_culture","Signal culture",{"id":85,"title":86},"talk_human","Talk to a person",{"id":88,"name":89,"type":90,"typeVersion":16,"position":91,"parameters":94,"category":96,"deletable":8,"connectable":8},"if_1","If: trust numbers","if",[92,93],460,40,{"buttonId":67,"operator":95},"equals","routing",{"id":98,"name":99,"type":100,"typeVersion":16,"position":101,"parameters":103,"category":52,"deletable":8,"connectable":8},"tx_1","Trust filter: branch numbers","text-message",[102,93],720,{"text":104},"**Trust filter for branch numbers (quick, ruthless):**\n1) **Definition stability:** Has the metric’s definition, eligibility, or window changed in the last 30–60 days? If yes, treat trends as *new metric, old label*.\n2) **Denominator sanity:** Are we looking at counts when we should be looking at rates (or vice versa)? Branch size differences quietly fake “performance.”\n3) **Lag + backlog:** If the work clears later (claims, onboarding, approvals), today’s dip might be tomorrow’s catch-up.\n4) **Gaming resistance:** If people can hit the number by changing behavior that doesn’t help customers, the number will become a sport.\n5) **Cross-check:** Trust metrics that reconcile with an independent signal (e.g., revenue *and* fulfillment *and* customer callbacks).\n\nIf you can’t answer #1–#3 in one breath, the number is probably polished noise—use it for questions, not conclusions.",{"id":106,"name":107,"type":90,"typeVersion":16,"position":108,"parameters":109,"category":96,"deletable":8,"connectable":8},"if_2","If: spot dirty signal",[92,39],{"buttonId":70,"operator":95},{"id":111,"name":112,"type":100,"typeVersion":16,"position":113,"parameters":114,"category":52,"deletable":8,"connectable":8},"tx_2","Dirty signal checks",[102,39],{"text":115},"**How to spot dirty signal before the meeting gets confident:**\n- **Too smooth to be real:** Perfectly steady lines often mean rounding, batching, or missing data—not “stability.”\n- **Spikes at weird times:** End-of-week/month cliffs usually indicate process artifacts, not customers.\n- **Nulls that look like zeros:** A missing feed can quietly become “0,” and everyone nods.\n- **Sudden cross-branch convergence:** When every branch becomes identical overnight, you didn’t find best practices—you found a pipeline issue.\n- **New top performer with no story:** If a branch ‘wins’ but no one can explain what changed on the ground, suspect tagging/attribution.\n\nPractical move: ask for a **one-page lineage**—where it comes from, when it refreshes, and the top 3 ways it breaks. If no one owns that, don’t let the metric chair the meeting.",{"id":117,"name":118,"type":90,"typeVersion":16,"position":119,"parameters":121,"category":96,"deletable":8,"connectable":8},"if_3","If: automation vs judgment",[92,120],240,{"buttonId":73,"operator":95},{"id":123,"name":124,"type":100,"typeVersion":16,"position":125,"parameters":126,"category":52,"deletable":8,"connectable":8},"tx_3","Automation vs judgment guidance",[102,120],{"text":127},"**When to trust automation (and when to keep a human in the loop):**\n**Good for automation:**\n- High-volume, low-stakes decisions with clear labels (routing, reminders, basic eligibility).\n- Cases where error is cheap and easily reversible.\n\n**Keep human judgment:**\n- Anything with **reputation risk**, **customer harm**, or **policy ambiguity**.\n- When incentives exist to game the inputs.\n- When the model/logic is trained on yesterday’s world (new product, new branch process, new channel behavior).\n\nRule of thumb: automate **execution**, not **interpretation**. Let systems do the busywork; make humans own the definition, exceptions, and the “are we sure this means what we think it means?” moment.",{"id":129,"name":130,"type":90,"typeVersion":16,"position":131,"parameters":133,"category":96,"deletable":8,"connectable":8},"if_4","If: messy evidence",[92,132],340,{"buttonId":76,"operator":95},{"id":135,"name":136,"type":100,"typeVersion":16,"position":137,"parameters":138,"category":52,"deletable":8,"connectable":8},"tx_4","Using messy evidence well",[102,132],{"text":139},"**Turn messy evidence into usable insight—without cleaning away the truth:**\n- **Don’t over-sanitize:** The weird edge cases often explain the main pattern (especially in branch ops).\n- **Bucket before you polish:** Group issues by *type of failure* (timing, eligibility, channel, training) before you chase precision.\n- **Prefer ranges over single numbers:** When inputs are messy, false precision is how teams get confidently wrong.\n- **Capture context, not just outcomes:** A ‘bad’ conversation tagged correctly can be more valuable than a ‘good’ metric with unclear sourcing.\n\nA practical output that leaders can use: **“If this stays true, we should do X. If it flips, we should do Y.”** That’s a decision system, not a spreadsheet.",{"id":141,"name":142,"type":90,"typeVersion":16,"position":143,"parameters":145,"category":96,"deletable":8,"connectable":8},"if_5","If: compare branches & attribution",[92,144],440,{"buttonId":79,"operator":95},{"id":147,"name":148,"type":100,"typeVersion":16,"position":149,"parameters":150,"category":52,"deletable":8,"connectable":8},"tx_5","Comparisons & attribution traps",[102,144],{"text":151},"**What teams misread when comparing branches, conversations, and attribution:**\n1) **Mix-shift:** One branch gets harder cases; another gets walk-ins. Raw comparisons punish the branch doing the hard work.\n2) **Channel drift:** Calls vs in-person vs chat create different conversion shapes—don’t compare without normalizing.\n3) **Attribution is a story, not a fact:** Last-touch is tidy and often wrong. If incentives depend on it, it will get ‘optimized.’\n4) **Small numbers lie loudly:** A branch can ‘lead’ on a metric with 20 cases. Confidence should scale with volume.\n5) **Process changes masquerade as performance:** New scripts, staffing, or queue rules can move metrics without changing underlying demand.\n\nBest practice: compare branches on **rate + volume + case mix**, and always ask: “What would have to be true for this to be an artifact?”",{"id":153,"name":154,"type":90,"typeVersion":16,"position":155,"parameters":157,"category":96,"deletable":8,"connectable":8},"if_6","If: signal culture",[92,156],540,{"buttonId":82,"operator":95},{"id":159,"name":160,"type":100,"typeVersion":16,"position":161,"parameters":162,"category":52,"deletable":8,"connectable":8},"tx_6","Signal culture guidance",[102,156],{"text":163},"**Build a signal culture that makes decisions happen (not just slides):**\n- **Name metric owners:** Every metric needs someone who can explain definition, refresh timing, and failure modes.\n- **Publish “how it breaks”:** A short, honest list of known caveats beats a long, ignored dashboard.\n- **Decision-first reporting:** Start with the decision to make, then pick the minimum signals needed.\n- **Reward truth-telling:** If people get punished for bad news, you’ll get good-looking numbers and bad outcomes.\n\nWitty but true: the best metric is the one that survives contact with reality *and* still changes what you do on Monday.",{"id":165,"name":166,"type":90,"typeVersion":16,"position":167,"parameters":169,"category":96,"deletable":8,"connectable":8},"if_7","If: talk to a person",[92,168],640,{"buttonId":85,"operator":95},{"id":171,"name":172,"type":173,"typeVersion":16,"position":174,"parameters":175,"category":179,"deletable":8,"connectable":8},"fb_1","Handoff to Insights Ops","fallback",[102,168],{"handoffMessage":176,"departmentId":177,"departmentName":178},"Got it—this sounds like it needs a human read. I’m handing this over to Insights Ops. Share the branch, timeframe, and the metric(s) you’re deciding on, and what decision you need to make.","dept_insights_ops","Insights Ops","terminal",{"id":181,"name":182,"type":173,"typeVersion":16,"position":183,"parameters":185,"category":179,"deletable":8,"connectable":8},"fb_2","Fallback: couldn’t match selection",[102,184],740,{"handoffMessage":186,"departmentId":177,"departmentName":178},"I didn’t catch a menu selection. If you tell me which branch metric (and timeframe) you’re looking at, I can route you—or I can connect you with Insights Ops.",[188,192,194,196,199,202,204,206,208,210,212,214,216,218,220,222,224],{"id":189,"source":35,"target":42,"sourceHandle":190,"targetHandle":190,"type":191},"c_1","main","default",{"id":193,"source":42,"target":54,"sourceHandle":190,"targetHandle":190,"type":191},"c_2",{"id":195,"source":54,"target":88,"sourceHandle":190,"targetHandle":190,"type":191},"c_3",{"id":197,"source":88,"target":98,"sourceHandle":198,"targetHandle":190,"type":191},"c_4","true",{"id":200,"source":88,"target":106,"sourceHandle":201,"targetHandle":190,"type":191},"c_5","false",{"id":203,"source":106,"target":111,"sourceHandle":198,"targetHandle":190,"type":191},"c_6",{"id":205,"source":106,"target":117,"sourceHandle":201,"targetHandle":190,"type":191},"c_7",{"id":207,"source":117,"target":123,"sourceHandle":198,"targetHandle":190,"type":191},"c_8",{"id":209,"source":117,"target":129,"sourceHandle":201,"targetHandle":190,"type":191},"c_9",{"id":211,"source":129,"target":135,"sourceHandle":198,"targetHandle":190,"type":191},"c_10",{"id":213,"source":129,"target":141,"sourceHandle":201,"targetHandle":190,"type":191},"c_11",{"id":215,"source":141,"target":147,"sourceHandle":198,"targetHandle":190,"type":191},"c_12",{"id":217,"source":141,"target":153,"sourceHandle":201,"targetHandle":190,"type":191},"c_13",{"id":219,"source":153,"target":159,"sourceHandle":198,"targetHandle":190,"type":191},"c_14",{"id":221,"source":153,"target":165,"sourceHandle":201,"targetHandle":190,"type":191},"c_15",{"id":223,"source":165,"target":171,"sourceHandle":198,"targetHandle":190,"type":191},"c_16",{"id":225,"source":165,"target":181,"sourceHandle":201,"targetHandle":190,"type":191},"c_17","automation",[28,29,30,228,229,230],"data-hygiene","attribution","leadership",[232],"Web chat","intermediate","Calypso","2026-05-28T11:03:36.097Z","/en/workflows/branch-signal-trust-filters",{"en":236},{"title":9,"description":239,"ogDescription":240,"twitterDescription":241,"canonicalPath":236,"robots":242,"schemaType":243,"alternates":244},"Route teams to practical checks for branch metrics, dirty signals, attribution comparisons, and when to trust automation vs judgment.","A guided workflow to pressure test branch numbers, spot dirty signal before meetings, and choose when automation is safe—or when humans should decide.","Practical filters for branch signals: what to trust, how to spot dirty data, how to compare branches, and when automation needs human judgment.","index,follow","HowTo",[245],{"hreflang":6,"href":236},1780761213312]