Skip to content
Brandon Olander
← All tools

Interactive tool · Synthetic data

Revenue Orchestration Lab

The same evidence should not always produce the same action.

VantageGrid sells cloud-operations software. Its revenue process still reads a content download as commercial priority and a production deployment as noise. This lab runs twelve situations through a model that keeps evidence, entity state, readiness and permission to act as separate things — and shows what changes when they stop being the same field.

People engage. Accounts buy.

A person's behaviour, an account's position, a product relationship and an opportunity are four different states. Atlas Logistics can be an Observe customer, a Control trialist and a closed-lost Control prospect at once, and collapsing those into one lifecycle field is how a system starts giving confident wrong answers.

Evidence is not inference.

A title string suggesting someone is an economic buyer is not the same fact as a human confirming it. Inferences carry confidence and can be withdrawn. Repeating an inference — even with a higher vendor score — produces the same evidence again, not a verification.

Readiness is not permission.

An account can be entirely ready to buy and entirely off limits, at the same moment, for good reasons. A hard block removes permission to contact. It does not remove the commercial case, and it does not delete the evidence that produced it.

VantageGrid, its two products, every account, every person and every event in this lab are invented. Nothing here is a real customer, a real outcome or a real performance claim.

Scenario controls

The failure it exposes

The case the legacy rule misses entirely: a healthy Observe customer validating Control in production, with a confirmed budget owner engaged on the business case.

Watch for

Both validation axes reaching meaningful, and the owner set becoming a coordinated group rather than a handoff.

Timeline

Step 1 of 7 · Day 1 · Atlas Logistics

Stepping forward replays the scenario from the beginning to the chosen event and re-derives every value on the page. Nothing is remembered between steps, so any step can be reached from any other and produce the same result.

S03 Expansion-ready customer. Step 1 of 7, day 1. Control trial started from inside Observe. Readiness: product interest. Permission to act: allowed. Next action owned by Marketing / Growth.

What changed

Day 1. Every dimension below is compared against the previous step.

  • Control relationship

    Prospect changed to Trialing

  • Technical validation

    None changed to Emerging

  • Buying-group coverage

    Absent changed to Emerging

  • Trial momentum

    No product usage changed to Active

  • Engagement breadth

    0 known people changed to 1 known person

  • Commercial readiness

    Not warranted changed to Product interest

  • Permission to act

    No action warranted changed to Allowed

Orchestration

ReadinessProduct interest

Evidence exists. No meaningful validation on either axis.

Permission to actAllowed

No constraint applies. The action can execute under existing rules.

No hard block, no competing owner, and no warning that raises the bar.

Next action

Hold in Control nurture and keep building account context. Do not occupy a seller with this yet.

Nurture

Evidence exists, but neither technical nor commercial validation is meaningful yet. Missing: Economic or business authority.

Recipients

  • Jonah Reyes

Owner

Marketing / Growth

Anonymous-to-known conversion, nurture, air cover, account engagement.

Participating

  • Product / PLG

Evidence timeline

Every event the system has seen, in order, with its source and whether it was observed or inferred. Inferences carry a confidence and can be withdrawn; observations cannot be upgraded by repetition.

  1. 01
    Day 1Product telemetryObservedCurrent step

    Control trial started from inside Observe

    Jonah Reyes launches a Control workspace from the Observe console. Existing customer, existing tenant, no new contract.

    Associated entities: Jonah Reyes · Control

Entity and assessment state

Six entity axes and seven assessments, each held separately. Any of them can move while the others stand still, which is the whole reason they are not one field.

Account

Name
Atlas Logistics
Relationship
Customer
Fit
Icp strongRecorded context. No orchestration rule reads it.
Engagement breadth
1 known person with recorded activityAccount-wide. Activity on any product counts here.
In the Control motion
1 personOnly these can hold a seat, or be targeted, for this motion.

Product relationships

  • Observe

    Customer

    Monitoring and observability. The installed base.

  • Control

    Trialing

    Automation and remediation. Sold as an expansion onto Observe.

    • Workspace createdDay 1 · Jonah Reyes

People

  • Jonah Reyes

    Director, Platform Engineering

    Technical leaderInferred

    Participating in: Control

  • Sofia Marchetti

    VP Infrastructure

    Unknown roleInferred

Buying group

CoverageEmerging
Technical seat
Jonah Reyes
Authority seat
Empty
Missing
  • Economic or business authority

Opportunity and customer context

Opportunity — Control motion
No Control opportunity on record
Customer health
Healthy
Support escalation
None open
CSM commercial hold
None
Routing
Not routed

Derived assessments

Technical validation
Emerging

Setup milestones only. Nothing production-grade has been connected or executed.

Commercial validation
None

Nobody at the account has engaged Control pricing, business-case, legal or procurement material.

Signal confidence
High

Nothing is degrading the reading: first-party product evidence and no unresolved data-quality conditions. Confidence measures the absence of known distortions, not the strength of the evidence.

Product momentum
Active

Decision trace

The same chain runs for every scenario. Reading it top to bottom is reading the reason for the action.

  1. Event

    Day 1 — Control trial started from inside Observe

  2. Evidence

    product telemetry, observed

  3. Entity state

    Control: trialing · buying group emerging

  4. Assessment

    Technical emerging · commercial none · readiness product interest

  5. Constraint

    None

  6. Permission to act

    Allowed

  7. Action

    Marketing / Growth — Hold in Control nurture and keep building account context. Do not occupy a seller with this yet.

The chain, in general

The trace on the left is this account at this step. The chain below is the shape every account runs through.

  1. 01

    Event

    Something happens. A form, a login, a policy execution, a support ticket, an opt-out.

  2. 02

    Evidence

    The event is recorded with a source and a basis: observed, or inferred with a confidence.

  3. 03

    Entity state

    Person, account, product relationship, buying group, opportunity and customer context each move on their own axis.

  4. 04

    Assessment

    Technical validation, commercial validation, buying-group coverage, signal confidence and readiness are derived separately.

  5. 05

    Constraint

    Hard blocks, existing-owner precedence and warnings are evaluated against what the system wants to do.

  6. 06

    Eligibility

    Permission to act is resolved. It is a different question from readiness and can move the opposite way.

  7. 07

    Orchestration

    An action, an owner, participating teams, and the reason the engine chose them.

VantageGrid’s legacy qualification rule

VantageGrid scores marketing engagements and routes at 60 points. This is that company’s own rule, run on the same events — not a claim that marketing engagement or lead scoring is invalid in general.

Legacy rule

0 / 60

Below threshold — hold in nurture

Scored at zero by this rule

  • Product Telemetry × 1

Orchestration model

Product interestAllowed

Hold in Control nurture and keep building account context. Do not occupy a seller with this yet.

Divergence

Neither model acts, for different reasons. The legacy rule is below threshold at 0 points; the orchestration model reads product interest.

Where automation stops

This lab makes no model call. Every value on this page is produced by deterministic rules over a fixed fixture, which is why the same scenario at the same step is identical every time. The boundary below is what the architecture would delegate if a model were wired in.

No model call is made by this page, and no value on it was generated by one.

Deterministic automation

Rules, not intelligence. Running today.

  • Routing from explicit rules
  • SLA timers and breach detection
  • Suppression and hard-block enforcement
  • State transitions with defined preconditions

Assistive, low consequence

A model could do these. A wrong answer is cheap and visible.

  • Summarise account activity for an owner
  • Normalise a job title into a tentative role category
  • Identify which buying-group seats are empty
  • Explain why a deterministic decision was reached

Human approval required

A wrong answer here costs an account, a contract or a compliance position.

  • Personalised outbound execution
  • Material lifecycle or status override under ambiguity
  • Overriding a governance veto
  • High-consequence exception handling

What this model does not infer

  • Purchase intent from third-party intent alone. Intent aims a motion; it does not start one.
  • A verified role from a repeated inference, however high the vendor's own confidence climbs.
  • Technical validation from marketing behaviour, or commercial validation from telemetry. The inputs are structurally separate, not filtered.
  • A resolved identity from name similarity. Reconciliation is an action someone takes, and the lab shows the state before and after it.
  • Anything about a real company. There is no data source behind this page.

Design tradeoffs

  • Deterministic fixtures over a live model

    Every reviewer sees identical output, and every rule is inspectable. The cost is that the lab cannot demonstrate handling genuinely novel input — it demonstrates the decision architecture, not a model's judgement.

  • Coarse levels over continuous scores

    Validation is none, emerging or meaningful rather than 0-100. Coarse levels force the rule that separates them to be stated. A score would hide the same rule behind a number that looks more precise than it is.

  • Scoped blocks over a global kill switch

    A hard block forbids a class of action, not the account. It costs more to model, and it is the difference between suppressing outreach and destroying the intelligence that would have justified it later.

  • One account-level motion, not a portfolio view

    This is an inspection lab for a single account's decision chain. The executive question — where is commercial velocity breaking down across the business — needs aggregation this version deliberately does not build.

Known limits of this version

  • Twelve fixed scenarios. There is no free-form input, no way to author an account, and no persistence.
  • The legacy comparison is VantageGrid's own scoring rule, invented for this lab. It is not a claim about marketing qualification in general.
  • Momentum, confidence and coverage use small integer thresholds chosen to make the mechanics legible, not calibrated against any real dataset.
  • The clock advances only when an event fires. Trial momentum decays at the next recorded event rather than continuously, which is what makes a step reproducible; a production implementation would run it on a timer.
  • Account fit is recorded and displayed but no rule reads it. It is context here, not an input to routing.
  • Identity reconciliation is triggered by an event in the fixture. There is no matching service behind it.

VantageGrid, its two products, every account, every person and every event in this lab are invented. Nothing here is a real customer, a real outcome or a real performance claim.

Synthetic scenario dataVantageGrid is a fictional company. No real customer data is used anywhere on this page.