How AI Search Is Recommending Certificates of Deposit

Certificate of deposits are becoming a rate-comparison market inside AI search. Consumers are not only asking which bank is trustworthy. They are asking which institution has the best CD rates right now, which bank is best for a CD account, which provider has the best 9-month CD, and where a large deposit can earn the highest current APY.

The 2026 LLM Authority Index public benchmark shows a fragmented recommendation environment. Marcus by Goldman Sachs is highly visible across CD-rate prompts, but it does not consistently control the top slot. AI systems also surface Synchrony Bank, Ally Bank, Capital One, Popular Direct, Bread Savings, Abound Credit Union, and E*TRADE in rate-maximization and best-CD-rate contexts.

Methodology

1. Market studied: Certificate of deposits, CD rates, online CD accounts, best CD-rate prompts, short-term CDs, 9-month CDs, high-yield deposit accounts, and adjacent savings-rate comparison prompts.
2. Brands/entities included: The supplied Marcus by Goldman Sachs structured dataset tracks Marcus by Goldman Sachs against Ally Bank, Barclays, Bread Savings, Capital One, CIT Bank, Discover, Popular Direct, Quontic Bank, SoFi, and Synchrony Bank. The structured company packet narrows the recurring CD competitor universe to Marcus by Goldman Sachs, Ally Bank, Barclays, Bread Savings, Capital One, CIT Bank, Popular Direct, Quontic Bank, and Synchrony Bank.
3. Data collection date/window: May 2026. The structured dataset was loaded on May 20, 2026 and reports the benchmark month as 2026-05.
4. AI platforms tested: The public benchmark is based on AI discovery observations across the tracked AI surfaces. The structured dataset includes AI observations from major AI/search-assistant environments represented in the Marcus packet.
5. Number of prompts tested: The public CD benchmark reports 18 CD-related prompt observations and approximately 69,948 modeled monthly searches in CD prompts. The uploaded Marcus dataset also contains a broader online banking / deposit-account packet, so this report uses the public CD benchmark for CD-specific scope and treats the broader structured file as supporting company-index evidence.
6. Prompt categories: CD-rate discovery, best-bank-for-CD prompts, short-term CD rates, 9-month CD rates, high-deposit CD-rate prompts, CD comparison prompts, and adjacent online banking / savings-account prompts. A QA note: the structured company packet contains stale “Medical Alert Systems” cluster labels, so this report normalizes those fields to the actual banking, savings, and CD prompt context rather than treating the stale labels as publishable taxonomy.
7. Definition of a mention: A bank or financial institution counted as mentioned when it appeared in an AI answer as a detected entity, regardless of whether it was recommended.
8. Definition of a valid recommendation: A valid recommendation required positive, shortlist-quality recommendation framing. Neutral rate references, factual APY mentions, product examples, or broad comparison anchors were not treated as full recommendation credit unless the dataset marked them as valid recommendations.
9. Ranking/scoring metrics used: Raw mention presence, valid recommendation coverage, recommended Top 3 rate, recommended Rank 1 rate, average recommended rank, positive/neutral/negative visibility, net sentiment score, citation/source patterns, and modeled monthly captured recommendation value. The structured methodology states that only positive valid recommendations receive rank credit, and only positive valid Top 3 recommendations receive modeled captured recommendation value.
10. Limitations: This is a point-in-time benchmark. CD rates change frequently, and AI outputs vary by platform, retrieval state, prompt wording, source freshness, geography, and date. Modeled monthly captured recommendation value is a benchmark estimate, not revenue, deposits, account openings, or attributable financial performance.

Key findings

1. Marcus is highly visible, but not uncontested. The public benchmark states that Marcus appeared in 15 of 18 CD-related prompt observations, but the category signal is fragmentation: Marcus is often included as a trusted national-bank option rather than treated as the automatic highest-rate leader.

2. Marcus has meaningful structured recommendation-stage strength. In the structured company packet, Marcus by Goldman Sachs shows an 8.0% recommended Top 3 rate, 4.15% Rank 1 rate, 23.2% positive visibility rate, 0.624 net sentiment score, 1.7284 average recommended rank, and 166,865.9404 modeled monthly captured recommendation value.

3. Ally Bank is the strongest modeled competitor in the broader structured packet. The structured competitor index identifies Ally Bank as the winner in the broad discovery and evaluation clusters, with a modeled captured value of 248,671.0386 in the main discovery cluster, ahead of Marcus at 128,737.504 in that cluster.

4. Marcus performs better in decision/pricing-style deposit prompts than in broad online-banking discovery. The structured packet shows Marcus with 37,977.103 modeled monthly captured recommendation value in the decision-stage pricing cluster, where the competitor-index view identifies Marcus as the winner against several competitors.

5. The citation layer is heavily finance-editorial. The public benchmark identifies Forbes, WSJ, Bankrate, NerdWallet, The Motley Fool, Kiplinger, and rate-comparison sites as the dominant citation environments. The structured observations also repeatedly cite Forbes, WSJ, Bankrate, NerdWallet, and similar financial editorial or comparison sources in banking, savings, and CD-rate answers.

What changed in the market

CD shopping has always been rate-sensitive. Buyers compare APY, term length, minimum deposit, early withdrawal penalties, FDIC or NCUA insurance, institution trust, and whether the account is easy to open online.

AI search compresses that comparison process. A buyer can now ask:

“Who has the best CD rates right now?” “What is the best CD rate for $100,000 today?” “Which bank has the best CD rates?” “What are the best 9-month CD rates?” “What is the best bank to open a CD account?”

These are decision-stage questions. The buyer is not just learning what a CD is. They are comparing institutions, rates, term structures, and trust signals.

That makes CDs different from many brand-led categories. In CDs, AI systems appear to reward whichever institution is easiest to validate through current rate pages, finance-publisher rankings, and “best CD rates today” source environments.

What the benchmark found

The public benchmark identifies Marcus by Goldman Sachs as highly present across CD prompts, but not dominant in every ranked answer.

Marcus benefits from brand trust, Goldman Sachs association, simple online account positioning, and recurring inclusion in CD and savings-rate comparisons. In raw structured observations, Marcus is often framed as a simple, reliable, trusted, or strong-rate option rather than always the highest-yield institution.

Synchrony Bank appears strongly in short-term and 9-month CD-rate contexts. In one raw observation for best 9-month CD rates, Synchrony ranked first, while Marcus appeared later in the shortlist.

Ally Bank is a recurring broad online-banking and deposit-account competitor. In the structured competitor view, Ally is the strongest modeled competitor in broad discovery and comparison-style clusters.

Capital One appears in broader “best bank” and hybrid online/branch narratives. It is less purely CD-rate-led than some challengers, but it benefits from brand familiarity, trust, and broader banking context.

Popular Direct, Bread Savings, Abound Credit Union, E*TRADE, and other rate-led competitors surface when the prompt emphasizes highest current yield. This is the category’s key dynamic: AI systems may include Marcus as a trusted option, then elevate a smaller or more rate-aggressive institution when the user asks for maximum APY.

Why visibility is not enough

CDs are one of the clearest examples of the difference between being visible and being the best recommendation.

Marcus appears often. But in rate-led prompts, AI systems do not appear to reward presence alone. They reward current-rate validation.

That matters because “trusted but not highest” can be a weaker position in a CD-rate market. If the user asks for the best CD rate right now, an AI system may include Marcus as a credible national option while ranking another institution higher because a finance publisher or rate tracker shows a stronger APY.

The structured data reinforces this distinction. Marcus has meaningful positive visibility and modeled recommendation value, but competitors still capture larger modeled value in several broad discovery and comparison environments. The Marcus company packet reports 166,865.9404 modeled monthly captured recommendation value against 440,741.6482 modeled competitor captured recommendation value.

For CD issuers and banks, the strategic question is not only:

Are we mentioned?

It is:

Are we recommended? Are we in the Top 3? Are we ranked first? Are we framed as highest-rate, safest, easiest, or most trusted? Are rate-comparison sources current enough to support our position? Are AI systems using stale or incomplete APY evidence when ranking competitors above us?

The citation layer

The citation layer is especially important in CDs because the product is rate-sensitive and time-sensitive. AI systems need current, trustworthy evidence to explain why one bank belongs above another.

The public benchmark identifies the key citation environments as finance publishers and rate-comparison sources, including Forbes, WSJ, Bankrate, NerdWallet, The Motley Fool, Kiplinger, and similar financial editorial infrastructure.

The structured dataset supports that pattern. Raw observations cite Forbes, WSJ, Bankrate, NerdWallet, and related banking-comparison sources across online banking, savings, and CD-rate prompts.

That creates a practical citation architecture problem for banks.

Owned rate pages matter, but they are not enough. AI systems also synthesize third-party rankings, editorial explainers, daily rate updates, “best CD rates today” pages, and comparison tables. If those sources do not include a bank, show outdated rates, or frame the institution as safe but not top-yielding, the bank may appear without winning the prompt.

For Marcus, the citation challenge is not basic legitimacy. Marcus has that. The challenge is rate-led displacement: competitors can outrank Marcus when AI systems find stronger current APY evidence elsewhere.

What brands need to fix

Banks and CD issuers need to manage AI search as a dynamic recommendation environment, not just a brand-awareness or SEO channel.

The first fix is rate-source freshness. CD-rate pages, third-party listings, and editorial comparisons must reflect current APYs, term options, minimums, and availability.

The second fix is prompt-specific positioning. “Best CD rates,” “best 9-month CDs,” “best bank for CDs,” and “best CD for $100,000” do not all reward the same institution. Brands need to know which prompts they win and which prompts send demand to competitors.

The third fix is Top 3 and Rank 1 tracking. In a rate-comparison market, appearing fifth is materially different from appearing first. Marcus is visible, but the benchmark’s warning sign is partial displacement.

The fourth fix is trust plus yield framing. Trusted brand positioning helps, but CD buyers often prioritize APY. Banks need sources that connect trust, safety, simplicity, and rate competitiveness.

The fifth fix is citation architecture. Banks need a public evidence layer across official rate pages, editorial rate trackers, comparison sites, finance publishers, and account-opening resources that gives AI systems current, consistent, and ranking-ready information.

How CiteWorks Studio helps

1. Map AI recommendation visibility. Track prompts, platforms, company presence, valid recommendations, Top 3 and Rank 1 performance, framing, and citation sources.
2. Identify the sources shaping AI answers. Find the editorial, review, forum, government, directory, owned, and search-visible sources that influence brand framing.
3. Build the citation architecture plan. Strengthen the public evidence layer so AI systems have more accurate, consistent, and persuasive source material to synthesize.

Commercial takeaway

Certificate of deposits are becoming AI-mediated rate shortlists.

Marcus by Goldman Sachs is highly present and strongly trusted, but the public benchmark shows that it does not consistently control the top slot. In CD-rate prompts, AI systems elevate whichever institutions are easiest to validate as current, competitive, and relevant to the requested term.

That creates opportunity and risk. Marcus can win when the prompt rewards trust, simplicity, and strong national-bank positioning. It can be displaced when the prompt rewards the highest current APY, short-term rate specials, credit-union offers, or rate-table recency.

For banks, the next competitive advantage will not come from brand awareness alone. It will come from keeping the public evidence layer current, consistent, and strong enough for AI systems to rank the institution in the buyer’s shortlist when rate-sensitive prompts are asked.

From Market Insight to Brand Action

Want to know how AI systems are recommending your bank, CD product, or savings account?

Request an AI Visibility Audit from CiteWorks Studio to see where your brand appears, where competitors outrank you, which prompts carry the most commercial risk, and which sources are shaping AI-generated banking recommendations.