CAPABILITIES / REPUTATION & PERCEPTION

Eight dimensions.Ninety metrics.One reputation you can measure.

The Sovereignty Infinium tracks 8 reputation dimensions and 90+ sub-metrics in real time, across 17+ languages, with source reliability at the cell level — so perception is a calibrated measurement, not a sentiment count.

8 DIMENSIONS

90+ SUB-METRICS

17+ LANGUAGES

ADMIRALTY-RATED

REAL-TIME TO ANNUAL

8 reputation dimensions

Source-tiered, cross-language, time-evolving.

01

Overall

02

Trust

03

Competence

04

Leadership

05

Safety & Security

06

Innovation

07

Cultural & Heritage

08

Economic & Opportunity

90+ sub-metrics

Source-tiered · cross-language · time-evolving

The Problem

Sentiment is not reputation. Share-of-voice is not perception.

Reputation is the most consequential perception an entity has, and the most poorly measured. Conventional tools track sentiment — positive, negative, neutral — and call that reputation. Sentiment is a count of words. Reputation is a multi-dimensional, source-weighted, cross-language, time-evolving, driver-attributed construct.

Sentiment = reputation

A high-volume negative moment looks like low reputation; a low-volume positive moment looks like high reputation

8 dimensions, 90+ sub-metrics, source-tiered weighting, volume-adjusted

Single-language coverage

Reputation is measured in English; the most consequential narratives are in regional languages

17+ languages at production quality, 50+ at digital listening, with cross-language comparison

No driver attribution

A reputation shift is observed but not explained

11 driver classes, with attribution from the multi-INT graph

Reputation is what your stakeholders believe about you, weighted by how much their belief matters. Sentiment is what the internet said. The two are not the same, and conflating them is the most expensive mistake in strategic communications.

The Capability

A measurement instrument, not a dashboard.

Reputation & Perception is the Sovereignty Infinium's calibrated, multi-dimensional, source-weighted, cross-language, time-evolving reputation construct. The construct has 8 dimensions, 90+ sub-metrics, 11 driver classes, a defined scoring formula with per-dimension weights.

The 8 reputation dimensions

1 / 8

Overall Reputation

Composite perception

Per source, per geo, per demographic

2 / 8

Trust

Trustworthiness perception

Government, institutions, media, brand

3 / 8

Competence

Capability perception

Service delivery, leadership, innovation

4 / 8

Leadership

Authority perception

Domestic, global, sector

5 / 8

Safety & Security

Safety perception

Travel, investment, residence

6 / 8

Innovation & Modernity

Forward-looking perception

Tech, society, economy

7 / 8

Cultural & Heritage

Cultural perception

Soft power, arts, tradition

8 / 8

Economic & Opportunity

Economic perception

Investment, job, growth, business

The 11 drivers of reputation

Every shift is attributed to a driver

Policy

Government action, legislation

Crisis response

How a crisis is handled

Communication

Strategic messaging, transparency

Leadership

Head of state, cabinet

Performance

Service delivery, GDP, security

Events

Sports, cultural, diplomatic

Incidents

Negative events, leaks

Disinfo / narrative

External narrative attacks

Media coverage

Tone, volume, framing

Influencer activity

KOL endorsement/criticism

Public mood

Underlying sentiment

The scoring formula

Reputation(t, dim, audience) =
  α * Sentiment(t-Δ, dim, audience)
  + β * Volume(t-Δ, dim, audience)
  + γ * SourceReliability(t-Δ, dim)
  + δ * TrustFactor(audience)
  + ε * Trend(t-Δ, dim)
  + ζ * ComparatorGap(dim, peer-set)

α + β + γ + δ + ε + ζ = 1.0
weights tuned per dimension/audience

The 90+ sub-metrics (representative)

Net Promoter Score equivalent

Trust score (0–100)

Sentiment (per dimension)

Share of voice (per topic, vs. peers)

Narrative share (per narrative)

Influencer endorsement rate

Media tone ratio (pos/neu/neg)

Source-by-source sentiment

Cross-source agreement (consensus)

Cross-language sentiment

Cross-cultural reception

Demographic breakdown

Trend (over time)

Comparator (vs. peer set)

The Mechanism

Source-tiered. Cross-language. Time-evolving. Driver-attributed.

Source-tiered weighting — the heart of calibration

Source Class

Default Weight

Government official

High

Wire services

High

Major international

High

Major national

High

Specialized trade

Medium-High

Local / regional

Medium

Independent / opposition

Medium

Social — verified

Medium-Low

Social — unverified

Low

Dark web

Very Low (per use)

HUMINT tip

Variable, per vetting

Citizen report

Variable, per vetting

3+ corroboration = upgrade. Single-source items are downgraded. Source reliability and information credibility are scored on the Admiralty scale; the reputation score carries the rating at the cell level.

Cross-language reception

Production-quality NLP

17+ languages

Digital-listening quality

50+ languages

Dialect awareness

Yes, per language

Code-switching detection

Yes, L1 ↔ English / L1 ↔ regional

Transliteration

Yes, script → Latin with phonological accuracy

Cross-lingual embeddings

LaBSE, mUSE, BGE-M3 for language-agnostic similarity

The 7 use cases the capability serves

Strategic communication prioritization

Where to communicate

Crisis response

Which dimension to defend

Brand strategy

Sector-level

Investment promotion

Per audience

Tourism / soft power

Per market

Diplomatic positioning

Per peer

Talent attraction

Per demographic

Reputation defense — the action side of measurement

Pre-bunk emerging narrativesDebunk active narrativesCounter-narrative deploymentInfluencer engagementStrategic communication offensiveMedia literacyCoalition buildingPublic engagement

AI + human fusion

Score 90+ sub-metrics per dimension

Translate 17+ languages in real time

Apply source-tiered weighting

Detect narrative share drift

Compute cross-source agreement

Attribute drivers via multi-INT graph

Decide which dimension to defend in a crisis

Author the perception-defense section of a brief

Approve a counter-narrative deployment

Counsel the principal on reputation trade-offs

Sign off on a reputation-grade intelligence product

Outputs

Daily pulse to annual report — calibrated and versioned.

The capability produces a defined cadence of outputs, each calibrated, each versioned, each with a defined audience.

Daily pulse

Per day, per audience, per dimension

Section chief, strategic comms

Weekly summary

Per week

Strategic comms lead, principal

Monthly deep-dive

Per month, per dimension

Section chief, ministry lead

Quarterly benchmark

Per quarter, vs. peer set

Cabinet, board

Annual report

Per year, year-on-year, long trend

Cabinet, board, public affairs

Crisis impact report

Per crisis

Section chief, principal

Comparator brief

Per request

Cabinet, board

Reputation quality metrics— the platform's own SLAs

8

Dimensions tracked

90+

Sub-metrics tracked

17+

Languages with production NLP

50+

Languages with digital listening

11+

Source classes in registry

200+

Source entries (tiered)

Multi

Update frequency (real-time → annual)

11

Driver classes, via multi-INT graph

Honest limits, surfaced to the consumer

Reputation is a construct. The platform’s formula is a defined construct with auditable inputs and weights. The construct is not ‘the truth.’ It is a measurement instrument.

Survey-based inputs depend on the survey’s methodology. The platform surfaces the methodology in the cell. Survey-based reputation is not the same as social-derived reputation.

Social-media-derived inputs are weighted by source class. An unverified social account contributes less than a wire service; the platform does not pretend otherwise.

Cross-cultural reception is a derived construct. The platform computes it from multi-source inputs; the construct is auditable, not assumed.

Some reputation dimensions are harder to measure than others. Cultural & Heritage is a more diffuse construct than Safety & Security. The platform surfaces confidence per dimension.

Anonymized Scenarios

Three reputation outcomes.

Scenario 01

Coordinated Reputational Attack Controlled Across 11 Languages

Situation

A sovereign client is the target of a coordinated reputational attack in the run-up to a major international summit. The attack is multilingual, multi-platform, and has narrative, financial, and cyber dimensions.

Challenge

The attack’s amplification network operates across 11 languages, with the most consequential narratives in 4 of them. Conventional monitoring catches the English-language strands; the regional-language strands go unmonitored for 36–72 hours.

Approach

The Sovereignty Infinium’s Reputation & Perception capability detected the attack in 17 languages, scored the per-dimension impact, and attributed the driver (disinfo / narrative) via the multi-INT graph. The platform’s narrative-INT derived layer identified the operator cluster and the coordination graph. The reputation defense playbook was activated, with pre-bunk messaging deployed in the 4 most consequential languages within 4 hours.

Outcome

Share-of-voice impact was limited to under 8% of baseline before the narrative trended globally. Recovery tracking continued for 14 days; the per-dimension recovery curve was within platform-typical bounds. The AAR was completed; the playbook was updated for the next cycle.

Lessons

Reputation attacks are multi-language problems. The platform’s 17+ language coverage + source-tiered weighting + driver attribution is the only architecture in which a coordinated 11-language attack can be controlled in 4 hours.

Scenario 02

Investment Promotion: Audience-Level Reputation Targeting

Situation

A sovereign wealth fund is preparing an investment-promotion campaign targeting three distinct audiences: institutional investors, family offices, and sovereign peers. Each audience has a different perception profile.

Challenge

Conventional promotion campaigns use a single narrative across all audiences. The result is a message that does not resonate with any of them. The campaign underperforms; the perception scores do not move.

Approach

The Sovereignty Infinium’s Reputation & Perception capability delivered an audience-level reputation brief for each of the three target audiences. The brief included per-dimension scores, per-dimension drivers, per-language reception, and a comparator gap to peer-set entities. The platform’s predictive foresight engine forecasted the per-dimension impact of three candidate campaign narratives.

Outcome

The campaign was launched with three audience-targeted narratives, each calibrated to the dimension scores the platform had identified as most consequential for that audience. The 90-day post-campaign measurement showed per-dimension improvements of 11–18% in the targeted dimensions, and a comparator gap improvement of 6–9% relative to peer set.

Lessons

Reputation is per-audience. A campaign that does not start with per-audience measurement cannot target per-audience. The platform’s 90+ sub-metrics and 7 use cases (one of which is investment promotion) are the architecture in which per-audience targeting is a designed outcome.

Scenario 03

Long-Term Reputation Recovery After a Major Incident

Situation

A multinational enterprise has experienced a major reputational incident (industrial accident, supply-chain failure, leadership scandal — anonymized). The conventional recovery pattern is 6–18 months of declining coverage and then a slow return to baseline.

Challenge

The conventional recovery is reactive. The enterprise does not know which dimensions to defend, which narratives to pre-bunk, which audiences to prioritize. The recovery takes longer than it needs to; the dimension-level damage persists.

Approach

The Sovereignty Infinium’s Reputation & Perception capability delivered a per-dimension, per-audience, per-language baseline measurement in the first 30 days post-incident. The platform identified the 3 dimensions most damaged, the 5 drivers most attributable, the 4 audiences most affected, and the 2 languages in which the damage was most acute. A 12-month recovery plan was built on the per-dimension, per-audience targets.

Outcome

The 12-month recovery plan was executed with monthly measurement and quarterly course-correction. Recovery-to-baseline was achieved in 8 months for the most damaged dimension (vs. 14-month industry baseline), 11 months for the second-most damaged (vs. 18-month baseline), and 14 months for the third (vs. 24-month baseline). The long-term reputation trend post-recovery was 4% above pre-incident baseline.

Lessons

Reputation recovery is a per-dimension, per-audience, per-language problem. A single-narrative, single-audience recovery is slower and less complete. The platform’s measurement instrument + the predictive foresight engine’s recovery forecast + the 7 use cases (one of which is brand strategy) are the architecture in which a faster, more complete recovery is a designed outcome.

How It Fits

Reputation reads the unified graph; the graph updates from reputation.

Reputation & Perception reads from the same multi-INT knowledge graph that every other capability writes to, and writes per-entity reputation updates back to the graph. A reputation score that is not grounded in the multi-INT graph is a sentiment count.

Multi-INT Fusion

Entity dossiers, narrative trends, source-tiered signals

Per-entity reputation updates, dimension-level driver changes

Predictive Foresight

Reputation trend forecasts, dimension-level trajectory

Forecasted reputation shifts, dimension-level driver changes

Real-Time Crisis Intelligence

Crisis-state events, narrative spikes

Reputation crisis events, recovery tracking

Threat Detection & Attribution

Threat-actor reputation, attribution confidence

Threat-actor reputation updates, reputational-attack attribution

Disinformation & Influence

Bot/CIB activity, narrative patterns, deepfake artifacts

Reputation impact of disinfo events, counter-narrative outcomes

Geopolitical Foresight

Country reputation, bilateral posture

Country-reputation updates, comparator gap

Cyber Threat Intelligence

Cyber-incident reputation impact

Cyber-incident reputation attribution

AI & LLM Perception

LLM-perception drift, hallucination evidence

LLM-perception reputation attribution

Cross-language example

A reputation shift in a regional-language market is detected via the platform’s 17+ language NLP. The multi-INT graph correlates the shift to a narrative spike (Narrative INT), a social-media amplification (SOCMINT), and a state-media posture change (OSINT). The reputation score is updated per language, per dimension, with source-tiered weighting. The threat-actor dossier is updated with the new evidence.

Anchored to the same entity IDs

Cross-temporal example

A reputation shift detected in real-time (sub-hour) is cross-referenced with the historical baseline, the trend trajectory, and the predictive forecast. The strategic communications team sees what is happening, what has been happening, and what is likely to happen — on one screen. The forecast is anchored to the same entity IDs that the alert uses.

Anchored to the same entity IDs
What It Does Not Do

Honest boundaries.

01

Reputation is a construct, not a measurement. The platform’s formula is a defined construct with auditable inputs and weights. The construct is not ‘the truth.’ It is a measurement instrument with a defined methodology.

02

Survey-based inputs depend on the survey’s methodology. The platform surfaces the methodology in the cell. Survey-based reputation is not the same as social-derived reputation.

03

Social-media-derived inputs are weighted by source class. An unverified social account contributes less than a wire service. The platform does not pretend otherwise.

04

Cross-cultural reception is a derived construct. The platform computes it from multi-source inputs; the construct is auditable, not assumed.

05

Some reputation dimensions are harder to measure than others. Cultural & Heritage is a more diffuse construct than Safety & Security. The platform surfaces confidence per dimension.

06

Multilingual coverage is not uniform. 17+ languages at production quality, 50+ at digital-listening quality. The platform does not pretend dialect coverage it does not have.

07

Reputation does not equal reality. A reputation shift may reflect a real-world event (incident, policy, performance) or a narrative attack. The platform’s driver-attribution layer separates the two, but the construct is still a perception, not a measurement of underlying reality.

08

The platform’s reputation defense is a measurement-and-recommendation capability, not an influence operation. The platform measures, attributes, and recommends. The decision to act, the legal framework under which action is taken, and the ethical review of the action are the client’s.

09

Some capabilities are subject to national export controls. Certain data sources, certain analytical methods, and certain deployable components may not be available in all jurisdictions.

When You're Ready

See reputation measured on your hardest audience.

Bring the audience you most need to understand. We will demonstrate Reputation & Perception against your specific audience — sovereign, institutional, public, regional — in a confidential setting, under your security protocols.

  • Response within 1 business day
  • Mutual NDA · no obligation
  • Under your security protocols

Or write to briefing@sovereignty.co.in

What you walk away with

Reputation, per dimension, per audience.

  • We measure 8 dimensions, 90+ sub-metrics on your audience
  • We attribute the drivers via the multi-INT graph
  • You see the per-language reception and cross-source agreement
  • We hand you a reputation brief, not a sentiment count
A measurement instrument, not a dashboard. Calibrated. Auditable. Multilingual.

Sovereignty Infinium is built for sovereign clients · All engagements operate under mutual non-disclosure · Some capabilities subject to national export controls

SOC 2 Type IIISO 27001GDPRFedRAMPFIPS 140-3Common Criteria EAL5+