Demo — AIVO Evidentia Ltd

Leading Global Bank

4 AI PLATFORMS · 20 RUNS · FOUR-TURN CONTROLLED SUBSTITUTION TEST

Overview

Platform Detail

Competitor Analysis

Remediation

Conversational Survival Rate

Final recommendation survival

First Failure Point

Replaced at attribute weighting

Replacement Rate

100%

20/20 runs · explicit displacement

Cross-Platform Consistency

100%

Identical across all 4 LLMs

Primary Threat

Competitor A

40% of all displacements

Conversational Funnel — Brand Survival

0% Survival

Present in 100% of initial responses

100%

PRIMARY

Mentioned but weakened — "solid but not best"

100%

WEAKENED

REPLACED

CONFIRMED

Cross-Platform Consistency

Systemic Failure

ChatGPT

5/5 REPLACED

→ Competitor A

Gemini

5/5 REPLACED

→ Competitor B

Perplexity

5/5 REPLACED

→ Competitor A / Competitor B

Grok

5/5 REPLACED

→ Competitor D / Competitor C

Universal Erosion Pattern (100% of runs)

Listed as option

Demoted in narrative

Explicitly replaced

Loss confirmed

Competitive Displacement Map

Concentrated Threat

Replacing Bank	Displacement Share	Primary Models	Type
Competitor A	40%	ChatGPTPerplexity	Direct
Competitor B	25%	GeminiPerplexity	Direct
Competitor D	15%	Grok	Direct
Competitor C	10%	Grok	Value-driven
Competitor E	10%	Mixed	Direct

Risk Classification & Failure Modes

Severe

Late-stage decision riskSevere

Narrative fragilityHigh

Model sensitivityLow

Competitive exposureConcentrated

Recovery potential (current)None Observed

Detected Failure Modes

Attribute misalignmentDETECTED

Decision-stage compressionDETECTED

Narrative demotionDETECTED

✓

Silent omissionCLEAR

✓

Model-specific biasCLEAR

ChatGPT

5 RUNS · 100% REPLACEMENT · PRIMARY REPLACER: Competitor A

C8-CG-DEM-01

→ Competitor A

C8-CG-DEM-02

→ Competitor A

C8-CG-DEM-03

→ Competitor A

C8-CG-DEM-04

→ Competitor A

C8-CG-DEM-05

→ Competitor A

Gemini

5 RUNS · 100% REPLACEMENT · PRIMARY REPLACER: COMPETITOR B

C8-GM-DEM-01

→ Competitor B

C8-GM-DEM-02

→ Competitor B

C8-GM-DEM-03

→ Competitor B

C8-GM-DEM-04

→ Competitor B

C8-GM-DEM-05

→ Competitor B

Perplexity

5 RUNS · 100% REPLACEMENT · PRIMARY REPLACER: COMPETITOR A / COMPETITOR B

C8-PX-DEM-01

→ Competitor A

C8-PX-DEM-02

→ Competitor A

C8-PX-DEM-03

→ Competitor B

C8-PX-DEM-04

→ Competitor A

C8-PX-DEM-05

→ Competitor B

Grok

5 RUNS · 100% REPLACEMENT · REPLACERS: VARIED

C8-GR-DEM-01

→ Competitor D

C8-GR-DEM-02

→ Competitor D

C8-GR-DEM-03

→ Competitor C

C8-GR-DEM-04

→ Competitor C

C8-GR-DEM-05

→ Competitor E

Competitor A

40%

ChatGPT (5/5 runs)Perplexity (3/5 runs)

Direct substitutionReplaces subject institution as "larger, more established" universal bank with stronger international presence

Competitor B

25%

Gemini (5/5 runs)Perplexity (2/5 runs)

Direct substitutionPositioned as stronger UK-focused retail banking alternative with better customer service narrative

Competitor D

15%

Grok (2/5 runs)

Direct substitutionMutual society positioning creates trust differential versus shareholder-owned banks

Competitor C

10%

Grok (2/5 runs)

Value-driven substitution

Competitor E

10%

Grok (1/5 runs)

Direct substitution

Key Intelligence

Replacement pressure is concentrated, not diffuse. Two competitors (Competitor A + Competitor B) account for 65% of all displacements. This creates a focused competitive counter-positioning opportunity — rather than defending against the entire market, remediation can target the specific narratives these competitors benefit from. Notably, Grok shows the most diverse replacement behaviour, introducing ethical/value-driven alternatives (Competitor C) that don't appear on other platforms.

Decision-Stage Positioning

Critical Priority · Immediate Action Required

The institution never survives to T2 as a viable decision outcome. The brand is consistently eliminated at the point where AI systems transition from listing options to making weighted recommendations. This requires fundamental repositioning of how AI systems evaluate the brand at the attribute-weighting stage.

Evidence basis: 100% explicit replacement at T2/T3 across all 20 runs. No model preserves the institution through the decision phase. Loss occurs at first "best bank" framing in every instance.

Attribute Reframing

Critical Priority · Immediate Action Required

Replacement is triggered by attribute weighting — the brand's key differentiators are either not being surfaced or not being valued by AI systems during comparison. The attributes that LLMs associate with this institution are insufficient to sustain competitive positioning when directly compared against Competitor A (international scale) or Competitor B (domestic focus, customer service).

Evidence basis: Attribute misalignment detected. Replacement consistently initiated at T2 where attribute-weighted comparison begins. The institution is reframed as "solid but not best" — suggesting the brand lacks a decisive attribute advantage.

Narrative Reinforcement

High Priority

Weakening occurs before replacement — all 20 runs show P→W transition at T1, meaning the AI narrative around this institution is eroded before competitors are introduced. The "solid but not best" framing at T1 sets up the replacement at T2. Strengthening the T1 narrative could delay or prevent the T2 failure.

Evidence basis: Uniform P→W→R erosion pattern across all runs. Narrative demotion detected. T1 weakening is the precursor to T2 replacement in 100% of cases.

Competitive Counter-Positioning

High Priority

The replacement set is narrow and stable — 65% of displacements come from just two competitors (Competitor A and Competitor B). This concentration creates a targeted counter-positioning opportunity. Rather than defending against the entire market, remediation should focus on the specific narratives these competitors benefit from within AI systems.

Evidence basis: Competitor A accounts for 40% and Competitor B 25% of all displacements. Direct substitution is the dominant pattern (90% of replacements). The narrow competitive set means counter-positioning can be precisely targeted.

Model-Specific Tuning

No model-specific tuning required Low Priority

Behaviour is identical across all four LLMs — there is no model divergence to exploit. Remediation efforts should be model-agnostic.

Ready to remediate?

The AIVO Evidentia team will develop a detailed implementation plan with specific tactical recommendations, timeline, content strategy, and expected impact modelling.