How Insight 360 works

This is a transparent, deterministic demo. The same message and scenario always produce the same score, because the rules below are applied in the open - no hidden model, no unexplained magic.

What this demo is

Insight 360 simulates a complete stakeholder-intelligence loop: it listens to synthetic social and media signals, lets you draft a message, stress-tests that draft against synthetic audience segments, and routes the result through a human governance workflow.

The demo is intentionally simplified. It shows the shape of the product - not a live production deployment. Every score, recommendation and persona reaction is generated from the rules on this page.

The workflow: LISTEN → TEST → DECIDE → LEARN

  1. LISTEN

    Home, social listening and media monitoring show what the organisation is hearing: topics, sentiment, alerts and emerging issues.

  2. TEST

    Content Analysis (the Message Lab) runs the drafted message through the rule engine. Personas show where different stakeholders disagree.

  3. DECIDE

    The Decision Brief consolidates the score, risk dimensions and a recommended rewrite. Governance requires analyst, manager and policy sign-off.

  4. LEARN

    Monitoring compares predicted response with observed response, so the next message starts from a smarter baseline.

Signals the engine checks

The engine scans the text for ten boolean signals. Each signal has a weight and a direction. Positive signals improve scores; negative signals reduce them.

Positive signals

  • Names a concrete date or deadlineweight 10
  • States who qualifiesweight 12
  • Gives a concrete call to actionweight 11
  • Names an application or contact channelweight 9
  • Lists required supporting documentsweight 8
  • Uses inclusive, supportive wordingweight 7
  • Reads clearly with short, specific sentencesweight 8
  • Long enough to carry the essentials (30–160 words)weight 6

Negative signals

  • Leans on pressure wordse.g. urgent, immediately, last chance, must act
  • Contains blame-oriented languagee.g. failure, blame, lazy, negligent

The score is not produced by a large language model. It is a deterministic weighted sum of these signals, capped between 5 and 95 to avoid false certainty.

The seven score dimensions

Six dimensions are scored directly from the signal set. The seventh, Risk, is derived from the average of the other six. Each dimension produces a short explanation based on whether it is below 50, 50–75, or above 75.

Clarity

Are the practical details concrete and easy to act on?

Trust

Does the message give verifiable specifics sceptical readers look for?

Relevance

Can audiences see themselves in the message?

Inclusivity

Does it address offline and rural audiences?

Emotional impact

Is the tone encouraging rather than alarmist or administrative?

Shareability

Is there quotable, forwardable practical detail?

Risk

Residual risk derived from the average of the other dimensions.

Predicted reception score is the average of Clarity, Trust, Relevance, Inclusivity, Emotional Impact and Shareability. The context-adjusted score adds a scenario modifier that reflects the historical difficulty of the topic in the demo's synthetic listening data.

How the recommendation is formed

The brief uses the context-adjusted score to choose one of three recommendation statuses:

  • Approve and proceed - score 75 or above. The message carries the practical detail audiences ask for.
  • Refine and improve - score 40 to 74. The message lands with some segments but leaves practical elements unaddressed.
  • Rethink and reframe - score below 40. The message is likely to generate more questions than action.

The recommended rewrite is not generated on the fly. It is the pre-written improved version that ships with each scenario, so you can see the difference between a vague draft and a concrete draft under the same rules.

Demo scenarios

The scenario switcher in the top bar loads one of three pre-built fictional scenarios. Switching resets the current draft, analysis and rewrite to that scenario's default state.

Graduate Skills Access Programme

National skills drive. Context lowers scores until clarity around eligibility and channels is added.

Community Safety Information Campaign

Sensitive safety topic. Vague or urgent wording is penalised more strongly.

Public Service Access Update

Operational service update. Queue frustration makes concrete hours and online alternatives valuable.

Governance & audit

The Review page is a stepped governance pipeline: analyst recommendation → manager review → policy check. Each step can approve, return to the previous step with a reason, or leave a note. Every action is written to an append-only timestamped audit log.

Nothing in the demo publishes automatically. The final approve button is a simulation of the human sign-off that would be required before a real message goes out.

Data & privacy

All listening data, personas, comments and scenarios are fictional. The demo stores its state in your browser's local storage so you can reload without losing your draft, but nothing is sent to a backend or used to train models.

Start exploring

The best way to see the rules in action is the guided demo tour. It walks through listening, the Message Lab, the Decision Brief, personas, governance and monitoring.

Sign in to startGo to the app