American Express AI Search Visibility: Recommendation Coverage Declined Across Five of Six Platforms

Research status: Exploratory longitudinal research. American Express's AI recommendation momentum has not been validated as a predictor of cardmember spending, new accounts, billed business, merchant activity, revenue, earnings, analyst revisions, valuation, or stock returns.

Observation window: July through September 2026

Ticker: AXP

Public parent: American Express Company

Tracked entities: American Express and American Express Co.

Exposure type: Direct/core brand with duplicate entity variants merged

AI platform families: ChatGPT, Gemini, Google AI Mode, Google AI Overviews, Microsoft Copilot, and Perplexity

Current methodology version: V0

Answer Capsule

American Express recorded a directional decline in AI recommendation coverage from July to September 2026 in the initial LLM Authority Index public-company panel.

Across 131 matched prompt-platform cells, American Express recommendation coverage declined from 35.9% in July to 27.5% in September 2026, a change of -8.4 percentage points.

The exploratory 95% interval ranged from approximately -15.7 to -1.1 percentage points, remaining below zero. Five of six measured AI platform families declined:

• ChatGPT: -27.27 pp
• Gemini: -10.00 pp
• Google AI Mode: -6.45 pp
• Google AI Overviews: -7.55 pp
• Microsoft Copilot: +12.50 pp
• Perplexity: -30.00 pp

American Express therefore meets the V0 Negative AI divergence candidate rules and receives a Medium AI-measurement confidence classification.

That classification refers only to the measured AI-side movement. It does not mean that American Express card spending, new-account growth, merchant activity, revenue, earnings, or stock performance will weaken.

The most important counterpoint is average recommendation rank. Among responses where American Express was recommended, average rank improved materially from approximately 3.88 to 2.59 even as recommendation coverage declined. Simple presence also declined, from 78.6% to 71.0%, a change of -7.6 percentage points.

The resulting measurement pattern is specific:

1. American Express appeared in fewer eligible matched responses;
2. it was recommended in a smaller share of those responses; and
3. when it was recommended, its average position improved substantially.

That is exactly why AI recommendations, mentions, citations, presence, and rank should remain separate metrics.

American Express's official investor materials describe the company as a global payments and premium lifestyle brand serving consumers, small and medium-sized businesses, large corporations, and merchants. The V0 signal therefore maps directly to the American Express brand, but it should still not be interpreted as a complete representation of every product, geography, customer segment, merchant relationship, or economic driver inside the company.

The prospective question remains the one defined by the AI Commercial Momentum Hypothesis: do persistent changes in unbranded AI recommendation behavior contain incremental information about later commercial outcomes after accounting for information already available when the signal was measured?

The signal construction is documented in How We Measure AI Commercial Momentum, while the complete initial public-company panel is preserved in Initial Findings From 25 Public Companies.

American Express AI Recommendation Momentum at a Glance

Additional V0 signal properties:

American Express sits outside the narrower bank sector slice because its economics and consumer proposition differ materially from deposit-led banks. It is nevertheless useful to compare its AI-side movement with financial-services peers such as Citi, Bank of America, and Goldman Sachs / Marcus.

Questions This Section Answers

• How much did American Express's AI recommendation coverage decline?
• Was the decline broad across AI platforms?
• Why can recommendation coverage decline while average rank improves?

How Much Did American Express's AI Recommendation Coverage Decline?

American Express recommendation coverage declined 8.4 percentage points, from 35.9% in July to 27.5% in September 2026.

The magnitude exceeds the current -5 percentage-point threshold used as one component of the V0 negative-candidate framework. More importantly, the exploratory interval remains below zero, from approximately -15.7 to -1.1 points.

Under the current V0 rules, that makes the measured recommendation-coverage change directionally negative.

This does not mean American Express is financially weaker. It means the matched AI responses contained fewer American Express recommendations in September than in July under the current extraction and classification rules.

The distinction is central to the project. The AI Visibility Market Divergence framework is a future validation concept, not a license to interpret every negative AI-side observation as negative equity information.

Was the Decline Broad Across AI Platforms?

Yes, the direction was broad but not universal.

Five of the six measured platform families declined:

Microsoft Copilot was the only improving platform.

This broad platform participation strengthens the interpretation that the aggregate movement was not created by one isolated engine. At the same time, the positive Copilot result matters. The current framework does not assume that platform agreement must be perfect.

That distinction is developed in Does Cross-Platform AI Visibility Matter?, which treats platform breadth as a separate signal-quality dimension rather than a substitute for the aggregate estimate.

American Express therefore shows a negative aggregate result with substantial, but not complete, cross-platform portability.

Why Can Recommendation Coverage Decline While Average Rank Improves?

Because coverage and rank measure different things.

Recommendation coverage asks: in what share of eligible matched AI responses was American Express classified as a recommendation?

Average rank asks: when American Express was recommended, where did it tend to appear among the recommended entities?

American Express was recommended less often overall, but when it was recommended, its average position improved from 3.88 to 2.59.

That combination is analytically important. A company can lose breadth across prompts while gaining prominence inside the narrower set of responses where it still appears as a recommendation.

The same separation applies to presence. American Express presence declined 7.6 points, broadly consistent with the coverage decline, but presence is still not the same as recommendation status.

The recommendation-share versus market-share framework will eventually ask whether these AI-side gaps have measurable commercial meaning. The current American Express observation alone cannot answer that question.

What the Platform Pattern Suggests About Measurement

American Express is a useful contrast with the fully portable negative cases elsewhere in the initial panel.

Bank of America and Goldman Sachs / Marcus declined on all six measured platform families. American Express declined on five of six.

That difference matters because the current research does not assume every AI platform is measuring the same commercial surface. Platform-specific retrieval systems, training data, ranking logic, query interpretation, freshness, source selection, and recommendation behavior can create materially different company representations.

American Express therefore provides a broad negative AI-side observation, but not a perfectly portable one.

The correct research question is not whether five-platform agreement proves a future financial outcome. It is whether multi-platform persistence later proves more informative than isolated platform movement when tested prospectively.

Why This Matters for Investor Research

The potential investor relevance comes from timing and incremental information, not from the direction of this one observation.

American Express reported Q2 2026 results on July 24, 2026, during the observation window. Those results were already public information and must be included in future baseline models.

American Express's Q3 2026 earnings conference call is scheduled for October 23, 2026, after the September AI signal freeze. That creates a clean prospective checkpoint for the research archive.

The correct future test is therefore not:

Did the stock go down after AI recommendation coverage fell?

The stronger test is:

After controlling for Q2 results, market information, financial expectations, and other variables already known by September, did the frozen AI-side measurement contain incremental information about later commercial or financial outcomes?

Potential downstream variables could include:

• billed business growth;
• cardmember spending growth;
• new card acquisitions;
• loans and cardmember receivables;
• merchant activity;
• branded search demand;
• website or app traffic where comparable data is available;
• analyst estimate revisions after the freeze date;
• later reported revenue and earnings outcomes.

None of those relationships has yet been established.

What This Does Not Mean

The current American Express observation does not mean:

• American Express is losing market share;
• cardmember spending is weakening;
• customer acquisition is declining;
• merchant economics are deteriorating;
• revenue or earnings will decline;
• analyst estimates will move lower;
• the stock is overvalued, undervalued, a buy, a sell, or a hold;
• AI recommendation coverage causes financial outcomes.

The V0 Negative AI divergence candidate label is a research classification for unusual negative AI-side movement under predefined thresholds. It is not an investment recommendation or valuation judgment.

Methodology

The American Express result follows the V0 methodology documented in How We Measure AI Commercial Momentum.

Matched-panel construction

The point estimate uses 131 matched prompt-platform cells and 110 normalized prompt clusters where the same underlying prompt-platform combinations can be compared between the base period and September.

Recommendation coverage

For each eligible matched response, the system determines whether American Express was classified as a recommended company or brand under the extraction rules. Coverage is the share of eligible matched cells containing that recommendation.

Entity normalization

The source corpus contained the variants American Express and American Express Co. The V0 public-company row merges those duplicate entity variants into one American Express Company exposure.

Presence versus recommendation

Presence records whether the entity appeared in the response. Recommendation coverage records whether the entity was classified as a recommendation. These are separate variables.

Rank

Average rank is calculated only within the responses where the company was identified as recommended. An improving rank does not offset or reverse a decline in recommendation coverage.

Exploratory interval

The exploratory 95% interval is based on variation across normalized prompt-level means and a normal 1.96 interval. It is an uncertainty estimate for the AI-side recommendation change, not a causal confidence interval and not a forecast interval for future financial performance.

Sensitivity checks

American Express shows a no-dedupe sensitivity difference of approximately -0.69 percentage points and a capture-average sensitivity difference of approximately -0.69 points. Both remain within the current High-confidence sensitivity threshold, but the row remains Medium confidence because it has fewer than 200 matched cells.

V0 classification

American Express qualifies as a Negative AI divergence candidate because:

1. recommendation coverage declined by at least 5 percentage points;
2. the exploratory interval remains below zero; and
3. at least four platform families worsened.

Again, this classification applies to the AI-side measurement only.

Limitations

Several limitations matter for American Express.

Prompt-population limitation. The measured prompt set is a structured research sample, not every possible way consumers, businesses, or merchants might ask AI systems about cards, payments, travel, rewards, financial services, or purchasing decisions.

Time-window limitation. July through September is a short observation window. A persistent trend requires additional months.

Platform limitation. Six platform families are measured, but their products, models, retrieval systems, and source behavior can change.

Entity limitation. The merged American Express variants improve consistency, but entity normalization does not guarantee perfect capture of every product, card, subsidiary, or geography.

Business-model limitation. American Express has multiple economic drivers. AI recommendation behavior for the consumer-facing brand should not automatically be mapped to billed business, merchant discount revenue, lending, travel services, corporate payments, or every customer segment.

Outcome-validation limitation. No causal or predictive relationship between this AI-side observation and later commercial or financial outcomes has yet been established.

What We Will Test Next

The American Express observation is now frozen as a dated pre-outcome record.

Future validation should compare the September signal with later data while preserving strict time order. The initial tests should include:

1. whether the decline persists into later monthly AI observations;
2. whether multi-platform breadth persists or reverses;
3. whether branded search or digital demand changes after the signal freeze;
4. whether later customer-acquisition or spending metrics move in the same direction;
5. whether analyst revenue or earnings estimates change after the freeze;
6. whether adding the frozen AI variables improves out-of-sample models beyond financial and market baselines.

A valid future result can confirm, weaken, or falsify the AI Commercial Momentum Hypothesis. The dated article should not be rewritten later to make the original signal appear more accurate than it was.

Related LLM Authority Index Research

• Can AI Search Signal Future Revenue Growth? The AI Commercial Momentum Hypothesis
• Initial Findings From 25 Public Companies
• How We Measure AI Commercial Momentum
• AI Investor Signal Tracker
• AI Recommendations vs. Mentions vs. Citations
• AI Visibility Market Divergence
• AI Recommendation Share vs. Market Share
• Citi AI Search Visibility
• Bank of America AI Search Visibility
• Goldman Sachs and Marcus AI Search Visibility

External Timing and Company Context

American Express investor relations describes the company as a global payments and premium lifestyle brand serving consumers, small and medium-sized businesses, large corporations, and merchants.

American Express reported second-quarter 2026 financial results on July 24, 2026. Those results were public during the observation window and should be treated as baseline information in later predictive tests.

The company currently schedules its Q3 2026 earnings conference call for October 23, 2026, after the September signal freeze. That date provides a prospective checkpoint for the research archive.

Sources:

• https://ir.americanexpress.com/investor-relations/default.aspx
• https://ir.americanexpress.com/events/event-details/2026/Q2-2026-American-Express-Earnings-Conference-Call/default.aspx
• https://ir.americanexpress.com/events/event-details/2026/Q3-2026-American-Express-Earnings-Conference-Call/default.aspx