Pipeline Stages, Lifecycle States, Events, and Tasks: What Each One Means
Many pipeline failures begin with one overloaded status field. A stage is a managed process position; a lifecycle state is a validated condition of a lead; an event is a timestamped occurrence; and a task is accountable work. Each object has different rules, history, ownership, and reporting consequences.
DEFINITION
What is this about?
A pipeline stage shows where a scoped work item sits in an operating process. A lifecycle state records the validated condition of a lead using SellersGPT's lead, qualified-lead, and active-lead definitions. A signal is a timestamped event or evidence object that can change interpretation or action without becoming a stage. A task is assigned work with an owner, deadline, and completion evidence. Link these objects; never make one field perform every job.
KEY TAKEAWAYS
- A stage describes managed process position; a lifecycle state describes validated lead condition.
- An event records what happened; a task records what someone must do.
- One event can leave state unchanged, while one transition can create or cancel several tasks.
- History belongs in immutable events and decisions, not overwritten status text.
- Reporting must name the object, scope, effective time, and denominator.
Who this explainer is for
This explainer is for product, sales, operations, data, and CRM teams designing or repairing a B2B GTM workflow. It provides an editorial domain model, not a vendor-specific database schema.
Contents
1 Why one status field fails 2 Define pipeline stage 3 Define lifecycle state 4 Define event 5 Define task 6 Connect the four objects 7 Model exceptions, reversals, and parallel work 8 Report each object correctly 9 Illustrative example 10 Common mistakes 11 What this does not mean 12 Common questions 13 Sources and methodology
1. Why one status field fails
Consider a field containing Follow-up sent. It might mean:
• an activity occurred;
• a workflow step completed;
• a lead remains in an outreach stage;
• a task is complete;
• the buyer progressed;
• an automation is waiting;
• the record needs no action. Those meanings are incompatible. The field cannot tell whether the message delivered, whether a buyer responded, whether qualification changed, whether another task is due, or whether outreach is now stopped. Formal process and modeling standards distinguish behavior, state, and occurrences in their own domains. The Object Management Group publishes BPMN for representing business-process behavior. OMG, "Business Process Model and Notation, Version 2.0.2" OMG's UML specification includes state-machine concepts for modeling state-dependent behavior. OMG, "Unified Modeling Language, Version 2.5.1" These specifications do not define a sales pipeline. They demonstrate why process position and state are separate modeling concerns. SellersGPT uses four linked objects so that each field has one job.
2. Define pipeline stage
A pipeline stage is the current managed position of a scoped work item in an operating process. Stages organize work; they do not by themselves prove buyer condition. Examples of process stages include:
• preparation review;
• approved for controlled outreach;
• outreach in progress;
• response review;
• qualification review;
• active-response coordination;
• human-owned follow-up;
• stopped or closed. A stage contract needs:
• process name and stage identifier;
• work-item scope;
• entry and exit criteria;
• allowed actions and prohibited actions;
• current owner or queue;
• time boundary and escalation;
• exception and re-entry rules;
• completion or terminal consequences. An account, campaign, lead, and task can each participate in different processes. Avoid one universal stage field across every object. A campaign can be outreach in progress while one lead is active, another remains qualified, and a third is suppressed.
3. Define lifecycle state
A lifecycle state is the currently validated condition of one lead record under a versioned definition. The SellersGPT lead lifecycle is: lead -> qualified lead -> active lead The arrow is conceptual, not a claim that buyers progress linearly.
• Lead: an identifiable person or organization has a recorded engagement or response event satisfying the applicable lead-creation rule.
• Qualified lead: applicable fit and qualification criteria are supported by required evidence, with no blocking exclusion.
• Active lead: meaningful human interaction or a required tenant action creates visible active work; route acceptance remains a separate control. Signal is not a lifecycle state. A lead can accumulate many timestamped signal events or evidence statements while its lifecycle state remains unchanged. A negative timing update can be a signal without moving the record closer to purchase. Use the current lifecycle state plus a signal collection or derived signal_present indicator, never a single overwritten signal state. A qualified lead can return to lead when new evidence disproves a criterion. An active lead can become inactive under a human-owned disposition while retaining its history. Each state needs effective time, decision time, criteria version, evidence links, and transition reason.
4. Define event
An event is an immutable record that something occurred or was asserted at a specific time. Examples include:
• outreach approved or sent;
• provider accepted or rejected delivery;
• click, reply, or form received;
• unsubscribe or company stop recorded;
• identity resolved or records merged;
• qualification evaluated;
• state transition decided or reversed;
• signal recorded;
• task assigned, accepted, completed, canceled, or overdue;
• handoff offered, accepted, rejected, or recalled. The Cloud Native Computing Foundation's CloudEvents specification standardizes common metadata for describing events across services. CloudEvents, "CloudEvents Specification" It is a technical interoperability specification, not a GTM event taxonomy. SellersGPT events should nevertheless have stable identifiers, source, type, subject, time, data reference, schema version, and correlation information. Google Analytics documents an event-and-parameter model for measurement. Google Analytics, "About events" A page event in that product does not carry SellersGPT commercial meaning. Preserve the occurrence and then evaluate it under the applicable pipeline rules. Delivery events also need underlying semantics. RFC 3463 provides enhanced status-code classifications for mail systems. IETF, RFC 3463 Store both provider data and normalized event type rather than losing the original diagnostic.
5. Define task
A task is a bounded unit of work assigned to an accountable owner. A task should include:
• task ID, type, and subject;
• reason or decision gap;
• required action and completion evidence;
• owner, contributors, and fallback;
• created, due, accepted, completed, and canceled times;
• priority based on consequence and time, not just activity;
• allowed and prohibited actions;
• linked events, state, stage, and context version;
• terminal disposition. Tasks are not evidence that work succeeded. Verify application can complete with fit confirmed, fit disproved, or evidence still Unknown. Completion must record the result and source. Avoid tasks that merely duplicate timers. Wait three days is better represented as a scheduled review event or stage deadline. Create a human task only when a person must evaluate, decide, communicate, approve, or resolve an exception.
6. Connect the four objects
Use explicit relationships: From Relationship To Example Event triggers evaluation of Lifecycle state substantive reply initiates qualification review Transition decision changes Lifecycle state evidence satisfies qualified-lead criteria Lifecycle state constrains Pipeline stage active lead pauses automated outreach stage Stage creates or cancels Task active-response coordination creates an owner-routing task Task completion produces Event engineer confirms application requirement Event closes or changes Task unsubscribe cancels queued follow-up task This creates a causal graph rather than an overloaded status. The minimum shared keys are subject, pursuit scope, correlation ID, causation ID, effective time, and version. If a form creates a lead and a route, the form event is the cause; the lead-state decision and route task are separate consequences. If the task later produces a technical confirmation, that completion event can support another decision.
7. Model exceptions, reversals, and parallel work
Real pipelines are not one-directional. Model:
• exception: an event cannot be processed normally and enters a review stage;
• reversal: new evidence changes a prior lifecycle state;
• re-entry: a closed process resumes under an explicit rule;
• parallel task: two independent actions are needed without creating two owners;
• cancellation: a hard stop invalidates scheduled or open tasks;
• merge: duplicate subjects combine while preserving both event histories;
• split: one record is discovered to represent multiple people, sites, or pursuits. NIST's log-management guide addresses policies and practices for managing event records. NIST SP 800-92 It does not prescribe pipeline history, but it supports retaining reviewable records rather than silently overwriting material changes. Use effective time and processing time. A stop received at 09:00 but processed at 09:07 should invalidate a follow-up scheduled for 09:05 even if the system executed it; the incident record must show the ordering and remediation. Do not rewrite the stop time to make operations appear correct.
8. Report each object correctly
For stages, report current work-item count, entries, exits, age, time in stage, blocked items, and terminal disposition. For lifecycle states, report current count, transitions, reversals, state age, criteria version, and evidence completeness. For events, report occurrences, distinct subjects, duplicates, processing delay, failures, and provenance. For tasks, report created, accepted, completed, overdue, canceled, result, and owner. Always name the unit and denominator. 42 replies is an event count. 38 people replied is a distinct-subject count. 14 qualified leads is a current or cohort state count depending on definition. 10 response tasks completed is work output. These cannot be placed in one conversion equation without cohort logic. Recommended integrity checks:
• current state equals the latest valid transition decision;
• every transition references a trigger and rule version;
• every active lead has a current accepted owner or explicit exception;
• every open task references a current decision gap;
• hard stops cancel incompatible tasks and stages;
• merged records preserve original events and identifiers.
9. Illustrative example
Illustrative example - explicitly fictional. CinderRock Automation receives one form from a fictional plant engineer at 10:00. The system records these separate objects: Time Object Record 10:00 Event form received, event E1 10:01 Event identity resolved to lead record L1, event E2, caused by E1 10:01 Lifecycle decision L1 created as lead under rule LC-3 10:02 Pipeline stage work item enters qualification review 10:02 Task verify application voltage, owner Engineering, due 14:00 11:20 Event engineer records 480 V confirmation with buyer-provided source 11:21 Lifecycle decision qualification criteria satisfied; state becomes qualified lead 11:21 Event request for drawing recorded as material signal 11:22 Task account owner must respond with approved drawing by 15:00 11:23 Lifecycle decision meaningful human response required; state becomes active lead There is one form event, one lead record, two lifecycle advances after creation, one material signal, and two tasks. It would be incorrect to count ten leads because the table has ten rows. It would also be incorrect to say that the drawing task caused qualification; the voltage evidence did. This example is fictional and does not predict a sale.
10. Common mistakes
• Using one status field for process, state, event, and task.
• Naming an event as a lifecycle state.
• Assuming a task completion proves a positive result.
• Treating a campaign stage as every lead's state.
• Overwriting the latest event instead of preserving history.
• Counting event rows as distinct leads.
• Creating human tasks for timers that need no judgment.
• Losing causation and correlation during a merge.
• Reporting current-state counts as cohort conversion.
11. What this does not mean
This explainer does not mandate a specific database technology, event bus, CRM, or process notation. It does not claim that every business needs the same stages or tasks. The technical standards cited are analogies for disciplined modeling, not SellersGPT implementation claims. The object model also does not define qualification pricing, market selection, ICP criteria, buying roles, or AI controls. It provides the vocabulary needed to implement whatever reviewed business rules apply.
12. Common questions
Is signal a state or an event? A signal is a timestamped material event or evidence object, not an exclusive lifecycle state. Implement a signal collection plus an optional derived signal_present indicator so history is not overwritten while the lead retains its independently derived lifecycle state. Can one lead be in two pipeline stages? A lead can participate in separate processes, such as qualification review and technical-response work. Within one exclusive process, define one authoritative current stage and use parallel tasks for concurrent work. Should every event create a task? No. Create a task only when an accountable person or controlled process must perform work. Duplicate telemetry, routine acknowledgments, and completed automated checks may need events without tasks. What is the source of truth for current state? The latest valid state-transition decision for the scoped lead, reconstructed from immutable decisions and corrections. A cached current-state field can improve performance but must reconcile with that history.
13. Sources and methodology
The four-object definitions, relationships, SellersGPT lifecycle semantics, and reporting checks are SellersGPT editorial tools. Sources provide formal or first-party examples of process notation, state modeling, event envelopes, measurement events, logs, and mail-status events. They do not define a B2B sales ontology. 1 PP06-OMG-BPMN - Object Management Group, "Business Process Model and Notation, Version 2.0.2." https://www.omg.org/spec/BPMN/2.0.2/PDF Accessed August 25, 2026. 2 PP06-OMG-UML - Object Management Group, "Unified Modeling Language, Version 2.5.1." https://www.omg.org/spec/UML/2.5.1/PDF Accessed August 25, 2026. 3 PP06-CLOUDEVENTS - Cloud Native Computing Foundation, "CloudEvents Specification, Version 1.0.2." https://github.com/cloudevents/spec/blob/v1.0.2/cloudevents/spec.md Accessed August 25, 2026. 4 PP06-GA4-EVENTS - Google Analytics Help, "About events." https://support.google.com/analytics/answer/9322688 Accessed August 25, 2026. 5 PP06-NIST-LOGS - National Institute of Standards and Technology, SP 800-92, "Guide to Computer Security Log Management." https://csrc.nist.gov/pubs/sp/800/92/final Accessed August 25, 2026. 6 PP06-RFC3463 - Internet Engineering Task Force, RFC 3463, "Enhanced Mail System Status Codes." https://www.rfc-editor.org/rfc/rfc3463 Accessed August 25, 2026.
Sources and Methodology
- omg.org: Pipeline Stages, Lifecycle States, Events, and Tasks: What Each One Means
- omg.org: Pipeline Stages, Lifecycle States, Events, and Tasks: What Each One Means
- github.com: Pipeline Stages, Lifecycle States, Events, and Tasks: What Each One Means
- support.google.com: Pipeline Stages, Lifecycle States, Events, and Tasks: What Each One Means
- rfc-editor.org: Pipeline Stages, Lifecycle States, Events, and Tasks: What Each One Means
- csrc.nist.gov: Pipeline Stages, Lifecycle States, Events, and Tasks: What Each One Means
Methodology
This article separates cited official or first-party facts from SellersGPT calculations, examples, frameworks, and recommendations. Source links and scope are retained from the publication pack.
Limitations
This material supports planning and decision-making. It is not legal, financial, regulatory, or performance advice, and it does not guarantee leads, meetings, customers, revenue, or return on investment.