American Express AI Market Strategy Report - Prepaid Cards

Answer Capsule

American Express holds a visible but under-recommended position in the Prepaid Cards category, with 13.00% valid recommendation coverage in September 2026 against a 23.61% raw mention presence rate. The benchmark shows the brand appearing in AI answers at roughly twice the rate it is actually recommended, which signals a recommendation conversion gap rather than a discovery problem. Its clearest strength is a stable positive framing profile and a meaningful rank-one rate on Google AI Mode, while its clearest weakness is a two-month decline across presence, top-three placement, and top placement. The clearest opportunity is converting existing mentions into shortlist and first-position recommendations on the surfaces where the brand already appears.

Who This Report Is For

This report is for American Express brand, growth, and digital strategy leaders who need to understand how the prepaid card portfolio is being represented in AI-generated recommendations, and for category analysts tracking recommendation-stage visibility in consumer financial products.

Report Card

Executive Summary

American Express is present in AI-generated prepaid card answers but is not converting that presence into recommendations at the rate its visibility would suggest. The September 2026 benchmark recorded 149 mentions across 631 qualified observations, a raw mention presence rate of 23.61%, while valid recommendation coverage stood at 13.00%. That gap means the brand is being referenced in roughly one of every four qualified answers but recommended in only about one of every eight.

The framing profile is positive. Of the 149 mentions, 84 were classified positive, 59 neutral, and 6 negative, producing a net sentiment score of 0.5235. There is no evidence of a reputational or framing problem in the public answer layer. The issue is placement and selection, not perception.

The strongest cluster signal sits in the single qualified buyer-intent class the benchmark currently measures, Brand Recommendation. All 631 qualified observations fell into that class in September 2026, and American Express recorded 82 valid recommendations, 40 top-three placements, and 10 rank-one placements within it. The brand is being considered, but it is rarely the default answer.

The weakest signal is the direction of travel. American Express declined to 13.00% valid recommendation coverage in September 2026 from 19.00% in July 2026, a 6.00-point drop. The largest single-month move came between August and September, down 5.20 points. Presence fell 6.20 points to 23.61%, top-three rate dropped 6.40 points to 6.34%, and rank-one rate fell 3.70 points to 1.58%. The decline is broad-based across every stage of the recommendation funnel.

The strongest platform signal is Google AI Mode, where the brand recorded 29 valid recommendations, an 18.01% valid recommendation coverage rate, and an 8.70% top-three rate across 161 observations. Google AI Overviews followed with 27 valid recommendations and 16.27% coverage. The weakest platform signal is Copilot, where the brand recorded 5 valid recommendations, a 7.25% coverage rate, and a single top-three placement across 69 observations.

The clearest gap is competitive displacement. Walmart MoneyCard holds 45.48% valid recommendation coverage and Bluebird by American Express holds 40.10%, both roughly three times the American Express rate. PayPal Prepaid, Chime, NetSpend, and Green Dot all sit between 21% and 26% coverage, meaning American Express has fallen below the entire mid-tier cluster it previously competed within.

What American Express Is Winning

Questions This Section Answers

• Where is American Express converting AI visibility into actual recommendations?
• What does the positive sentiment and framing profile mean for the brand's AI position?

The brand's clearest win is framing quality. A net sentiment score of 0.5235 across 149 mentions, with 84 positive and only 6 negative classifications, indicates that when AI systems do discuss American Express in the prepaid card context, the framing is favorable. This is a durable asset because it means the correction work is about selection and placement rather than reputation repair.

The second win is Google AI Mode performance. Across 161 observations on that surface, American Express recorded 29 valid recommendations, an 18.01% valid recommendation coverage rate, and an 8.70% top-three rate. That coverage rate is meaningfully above the brand's 13.00% overall figure, which suggests the brand's owned and indexed content is performing better in Google's AI-generated answer layer than in standalone assistant surfaces.

The third win is a narrow but real first-position pocket. The brand recorded 10 rank-one placements overall, including 3 on Google AI Mode, 3 on Google AI Overviews, 2 on Gemini, 1 on ChatGPT, and 1 on Perplexity. Rank-one placement is the hardest recommendation outcome to earn, and the fact that it occurs across five of the six tracked surfaces indicates the brand can win the default answer when the surrounding evidence supports it.

These wins are real but limited. The brand does not hold a leading cluster, does not lead any platform, and does not hold a top-three position in the category standings. The wins describe a brand with a functioning baseline, not a brand with recommendation power.

Where American Express Has the Clearest AI Visibility Gaps

Questions This Section Answers

• Why does American Express appear in AI answers without being recommended?
• Where is American Express losing shortlist and first-position placement to competitors?
• Which platform surfaces show the widest recommendation gap for American Express?

The primary gap is recommendation conversion. American Express appears in 23.61% of qualified observations but receives a valid recommendation in only 13.00%. That 10.61-point spread is the widest presence-to-recommendation gap among the upper half of the tracked set, and it means the brand is frequently used as context, comparison anchor, or background reference rather than as a shortlisted option.

The second gap is top-three displacement. The brand's top-three rate fell to 6.34% in September 2026 from 12.70% in July 2026. Walmart MoneyCard holds a 37.24% top-three rate and Bluebird by American Express holds a 36.45% top-three rate. Both brands are occupying shortlist positions at roughly six times the American Express rate. In practical terms, when an AI system assembles a three-option prepaid card shortlist, American Express is absent from it in more than nine of every ten qualified answers.

The third gap is first-position loss. The rank-one rate fell to 1.58% in September 2026 from 5.30% in July 2026. Bluebird by American Express holds a 29.64% rank-one rate, which is roughly eighteen times the American Express figure. This is the most consequential gap because rank-one placement is where the default answer is formed, and the benchmark shows that position is being consolidated by a small number of brands.

The fourth gap is platform inconsistency. Copilot recorded only 5 valid recommendations and a single top-three placement for American Express across 69 observations. Perplexity recorded 7 valid recommendations and a 9.59% coverage rate. Both surfaces are materially below the brand's Google AI Mode performance, which suggests the brand's retrievable evidence layer is better aligned with Google's answer generation than with assistant-style surfaces.

The fifth gap is the trend itself. Four brands moved significantly downward versus baseline in this benchmark: American Express, Bluebird by American Express, Chime, and NetSpend. American Express recorded the sharpest single-month decline in September, down 5.20 points. A two-month decline across presence, top-three rate, and rank-one rate is not normal month-to-month variation, and it indicates that whatever evidence the brand was previously drawing on has weakened or been displaced.

Biggest Opportunity

Questions This Section Answers

• Which AI surfaces offer the best chance to close the presence-to-recommendation gap?
• What evidence types would make American Express mentions recommendation-eligible?
• Why does the lack of pricing and comparison prompts work in American Express's favor?

The single biggest opportunity is closing the presence-to-recommendation gap on Google AI Mode and Google AI Overviews, where the brand already has its strongest retrievable footprint. American Express records 23.61% presence but only 13.00% recommendation coverage, and its Google AI Mode coverage rate of 18.01% shows the brand can convert above its overall average when the answer layer has enough structured, attributable evidence to work with.

The path is not to increase mentions. The brand is already mentioned at more than double its recommendation rate. The path is to make the existing mentions recommendation-eligible by strengthening the specific evidence types that AI systems appear to use when assembling a shortlist: clear product positioning, comparison-ready attribute language, and third-party source coverage that names the brand in a recommendation context rather than a background context.

The benchmark cannot currently answer pricing or head-to-head comparison questions, because no qualified observations fell into the Pricing & Value or Multi-Brand Comparison classes in July, August, or September 2026. That is itself a signal. The category's qualified prompt set is concentrated in general brand recommendation questions, which means the shortlist decision is being made on brand-level recommendation signals rather than on fee or feature comparison. For a brand with strong framing and weak placement, that is the most favorable possible structure, because it means the correction work is about recommendation evidence rather than price competitiveness.

Competitive Landscape

Questions This Section Answers

• Where does American Express rank by recommendation strength compared to other prepaid card brands?
• Which competitors are converting presence into shortlist and first-position placements at the highest rates?

Walmart MoneyCard and Bluebird by American Express hold recommendation-stage strength in the Prepaid Cards category, with both brands converting presence into shortlist and first-position placements at rates far above the rest of the field. American Express sits in seventh place by valid recommendation coverage, below the mid-tier cluster formed by PayPal Prepaid, Chime, NetSpend, and Green Dot.

Average recommended rank covers rank-eligible recommendations only.

American Express holds the lowest top-three rate and the lowest rank-one rate of any brand in the upper half of the table, and its average recommended rank of 3.443 is the weakest among brands with meaningful recommendation volume. The table shows a brand that is recommended less often and placed lower than every competitor with comparable or greater presence.

Prompt Evidence

Questions This Section Answers

• On which prompts does American Express perform strongest and weakest in AI recommendations?
• Where did negative framing appear for American Express in AI-generated answers?

Google AI Mode / Brand Recommendation Prompt: "prepaid cards" Result: American Express recorded 29 valid recommendations and an 18.01% valid recommendation coverage rate on this surface, its strongest platform result, though top-three placement occurred in only 8.70% of observations.

Copilot / Brand Recommendation Prompt: "prepaid debit cards" Result: American Express recorded 5 valid recommendations and a single top-three placement across 69 observations, a 7.25% coverage rate and the brand's weakest platform result.

ChatGPT / Brand Recommendation Prompt: "prepaid visa card" Result: American Express recorded 6 valid recommendations, an 8.00% coverage rate, and 3 negative classifications across 75 observations, the only surface where negative framing appeared at a measurable rate.

Google AI Overviews / Brand Recommendation Prompt: "bank accounts" Result: American Express recorded 27 valid recommendations and a 16.27% valid recommendation coverage rate across 166 observations, with 3 rank-one placements and no negative classifications.

What CiteWorks Studio Would Do Next

Phase 1: AI Market Discovery Audit Map every qualified prompt where American Express is mentioned but not recommended, and identify which competitor takes the recommendation slot in each case.

Phase 2: Recommendation Readiness Plan Prioritize the Google AI Mode and Google AI Overviews surfaces where the brand already converts above its average, and define the specific evidence types needed to move mentions into shortlist positions.

Phase 3: Owned Answer Layer Buildout Restructure prepaid card product pages and comparison content so that recommendation-eligible attributes, eligibility language, and category positioning are extractable in the form AI systems appear to use when assembling shortlists.

Phase 4: Citation / Authority Layer Development Strengthen third-party source coverage that names American Express in a recommendation context, since the benchmark shows the brand is currently referenced more often as background than as a shortlisted option.

Phase 5: Monthly AI Visibility and Recommendation Tracking Track presence, valid recommendation coverage, top-three rate, rank-one rate, and net sentiment monthly across all six surfaces to confirm whether the two-month decline has stabilized and whether shortlist placement is recovering.

Why This Matters

Questions This Section Answers

• How are AI-generated recommendations shaping the prepaid card buyer shortlist?
• What makes closing the presence-to-recommendation gap a targeted correction rather than a brand rebuild?

AI-generated recommendations are forming the buyer shortlist before a consumer ever visits a comparison page. In the Prepaid Cards category, the benchmark shows that shortlist is being assembled from brand-level recommendation signals, not from fee or feature comparisons, because no qualified observations fell into the Pricing & Value or Multi-Brand Comparison classes across the three measured months. That means the brands that appear in the top three positions of an AI answer are capturing the decision moment, and the brands that appear as background references are not.

American Express has the harder half of the problem already solved. The brand is visible, it is framed positively, and it appears across all six tracked surfaces. What it does not have is recommendation conversion. Closing a 10.61-point presence-to-recommendation gap is a targeted correction of the prompt, page, and citation layers, not a rebuild of brand perception. The benchmark identifies where the gap is and how large it is. The next step is identifying which specific prompts, pages, and sources are producing mentions without producing recommendations.

Core Metrics

Sentiment Score

Sentiment Score = (positive mentions × 1 + neutral mentions × 0 + negative mentions × -1) / total mentions

For American Express in September 2026: (84 × 1 + 59 × 0 + 6 × -1) / 149 = 78 / 149 = 0.5235.

This matters because unclassified mention counts are misleading. A brand that appears in 149 answers sounds healthy until the mentions are separated into positive recommendations, neutral references, cautionary mentions, and competitor-displaced comparisons. American Express's 149 mentions break down into 84 positive, 59 neutral, and 6 negative, which means more than a third of the brand's visibility is neutral reference rather than active recommendation.

Share of voice is a diagnostic metric, not a business KPI. Knowing that American Express appears in 23.61% of qualified answers does not tell you whether the brand is being chosen. The 13.00% valid recommendation coverage does. The 10.61-point gap between those two numbers is the actual story, and it is only visible once mentions and recommendations are counted separately.

A positive recommendation, a neutral reference, a cautionary mention, and a competitor-displaced mention are not equal. Counting all mentions as wins is bad measurement. Classified sentiment is required before interpreting AI visibility, because a brand with 149 mentions and a 0.52 sentiment score is in a fundamentally different position than a brand with 149 mentions and a 0.90 sentiment score, even though the raw visibility number is identical.

Sentiment by Platform

Methodology

1. This report is a benchmark-based analysis of American Express within the Prepaid Cards category, drawing on the LLM Authority Index AI Market Discovery Index for September 2026 and the associated company-level metrics aggregation.
2. The reporting window covers September 2026, with July 2026 as the baseline month and August 2026 as an intermediate measurement.
3. Six AI and search surfaces were tracked: ChatGPT, Copilot, Gemini, Perplexity, Google AI Overviews, and Google AI Mode. All six recorded at least one qualified observation.
4. Each monthly run began with 800 prompt-surface observations. September 2026 produced 542 unique questions, 758 relevant observations, 42 irrelevant observations, and 631 qualified benchmark observations after both qualification stages.
5. The competitor universe contains 10 tracked brands: American Express, Bluebird by American Express, Brink's Money Prepaid, Chime, Green Dot, Movo, NetSpend, PayPal Prepaid, Varo Bank, and Walmart MoneyCard.
6. Three public high-intent clusters were defined: Brand Recommendation, Pricing & Value, and Multi-Brand Comparison. In September 2026, all 631 qualified observations fell into the Brand Recommendation class. No observations qualified for the Pricing & Value or Multi-Brand Comparison classes in July, August, or September 2026.
7. Stage 0 extraction retains the query, AI or search surface, answer, brand outcome, recommendation placement, sentiment classification, and, where exposed, citations or attributable evidence sources. Source presence is treated as evidence about the information environment and is not treated as proof of causation.
8. A mention is counted when a tracked brand appears anywhere in a qualified AI answer, regardless of whether it is recommended. Raw mention presence rate is the share of qualified observations containing at least one mention of the brand.
9. A valid recommendation is counted when the dataset explicitly marks the brand as a positive valid recommendation with a rank position between 1 and 10. Neutral references, cautionary mentions, comparison anchors, and listed-only appearances are not counted as valid recommendations.
10. Brand-level percentages use the 631 qualified observations as the public denominator, not the raw 800-prompt collection universe.
11. Varo Bank appears as a separately tracked brand name beginning in August 2026 following a transition from the legacy Varo label. The two names should not be combined when reading month-over-month movement.
12. Movement between months identifies changes worth investigating but does not establish cause. This is directional benchmark analysis, not a controlled experiment. Small observation counts for brands such as Movo and Brink's Money Prepaid make their percentages sensitive to small changes.

See Where Your Brand Stands in AI Recommendations

The public benchmark shows where American Express is visible and where it is not being chosen. A company-level AI visibility audit maps the specific prompts, surfaces, competitors, and evidence sources behind that gap, and turns the category-level pattern into a prioritized plan for recommendation-stage visibility.