PennyMac AI Visibility Market Strategy Report - Mortgage Refinance Lenders

Answer Capsule

PennyMac holds meaningful AI presence in mortgage refinance recommendations but converts that presence into top-three placements far less often than its visibility suggests. In October 2026, the brand appeared in 37.43% of qualified AI responses yet earned a valid recommendation in only 26.47% of them, and a top-three placement in just 7.94%. Rocket Mortgage, by contrast, converted 98.11% presence into 79.21% valid recommendation coverage and a 69.94% top-three rate. PennyMac's clearest opportunity sits in closing the gap between being mentioned and being chosen, particularly on platforms where it is discussed as context rather than surfaced as a leading option.

Who This Report Is For

This report is written for PennyMac's marketing, brand, and growth leadership, and for category analysts tracking how mortgage refinance lenders are recommended inside AI-generated answers.

Report Card

Executive Summary

Questions This Section Answers

• How wide is the gap between PennyMac's AI presence rate and its rate of being recommended?
• Which platforms produced PennyMac's strongest and weakest recommendation signals?

PennyMac enters October 2026 with a visibility profile that looks stronger than its recommendation profile. The brand was mentioned in 198 of 529 qualified observations, a raw mention presence rate of 37.43%, which places it fifth in the tracked set. But only 140 of those observations produced a valid recommendation, a coverage rate of 26.47%, and only 42 produced a top-three placement, a rate of 7.94%. The gap between presence and placement is the defining feature of PennyMac's position in this benchmark.

The brand's sentiment profile is healthy. PennyMac recorded 141 positive mentions, 57 neutral mentions, and zero negative mentions, producing a net sentiment score of 0.7121. There is no negative framing problem in the data. The issue is not how PennyMac is described when it appears, but how often it appears in a position that influences a buyer's shortlist.

PennyMac's strongest platform signal comes from AI Overviews, where it recorded a 41.27% valid recommendation coverage rate and a 12.70% top-three rate across 126 observations. That is the only platform where PennyMac's recommendation conversion approaches the category leaders. On ChatGPT, by contrast, PennyMac recorded a 22.73% valid recommendation coverage rate and a 12.12% top-three rate, while loanDepot posted a 46.97% coverage rate and a 31.82% top-three rate on the same platform.

The weakest platform signal is Copilot, where PennyMac recorded a 9.76% valid recommendation coverage rate and a 1.22% top-three rate across 82 observations. On Copilot, Better.com recorded an 89.02% coverage rate and a 78.05% top-three rate, and loanDepot recorded an 87.80% coverage rate and a 58.54% top-three rate. PennyMac is effectively absent from the recommendation layer on that platform.

Against the July 2026 baseline, PennyMac's coverage moved from 27.7% to 26.5%, a decline of 1.2 points that fell within normal month-to-month variation. The September-to-October movement of 6.5 points was a significant gain, indicating the brand recovered after an August dip. The brand is stable against baseline, not declining, but also not gaining ground on the leaders.

The clearest structural gap is the absence of qualified observations in the Pricing & Value and Multi-Brand Comparison clusters. All 529 qualified observations in October 2026 fell into the Brand Recommendation cluster. This means the benchmark cannot yet show how PennyMac performs when buyers ask AI systems to compare lenders directly or to surface pricing information, which are the prompt types closest to a refinance decision.

What PennyMac Is Winning

Questions This Section Answers

• Where does PennyMac actually perform well in this benchmark?
• What does PennyMac's sentiment profile look like across mortgage refinance answers?

PennyMac's strongest evidence-backed win is its sentiment profile. The brand recorded zero negative mentions across 198 total mentions, with 141 positive and 57 neutral. A net sentiment score of 0.7121 places it mid-pack among tracked brands but confirms that AI systems are not framing PennyMac negatively when they discuss it.

The brand's second win is its AI Overviews performance. Across 126 observations on that platform, PennyMac recorded a 41.27% valid recommendation coverage rate and a 12.70% top-three rate. That top-three rate is the second-highest PennyMac recorded on any platform and represents the brand's most reliable recommendation pocket.

PennyMac's third win is its recovery trajectory. The September-to-October movement of 6.5 points in valid recommendation coverage exceeded normal month-to-month variation, indicating the brand regained ground after an August dip. The brand's top-three rate also improved modestly against the July baseline, moving from 6.9% to 7.9%.

These are real but narrow wins. PennyMac does not hold a dominant position on any platform, and its overall recommendation conversion rate of 26.47% sits well below the category leader's 79.21%.

Where PennyMac Has the Clearest AI Visibility Gaps

Questions This Section Answers

• Why does a presence rate of 37.43% convert into a top-three rate of only 7.94%?
• Which competitors are converting similar mortgage refinance presence into far more top-three placements?

PennyMac's most significant gap is recommendation conversion. The brand appeared in 198 qualified observations but earned a valid recommendation in only 140 of them. That means 58 mentions, roughly 29% of PennyMac's total presence, were reference-only appearances where the brand was discussed but not recommended. Rocket Mortgage, by comparison, converted 519 mentions into 419 valid recommendations, a conversion pattern that reflects near-total presence-to-recommendation alignment.

The second gap is top-three placement. PennyMac's 7.94% top-three rate places it fifth in the tracked set, behind Rocket Mortgage at 69.94%, loanDepot at 23.82%, Better.com at 19.09%, and Bank of America Corp. at 18.71%. The brand is being mentioned at a rate comparable to Better.com, which recorded a 28.73% presence rate, but Better.com converts that presence into top-three placements at more than twice PennyMac's rate.

The third gap is platform concentration. PennyMac's recommendation strength is heavily concentrated on AI Overviews. On Copilot, the brand recorded a 9.76% coverage rate and a 1.22% top-three rate, while Better.com and loanDepot both exceeded 58% top-three rates on the same platform. On ChatGPT, PennyMac recorded a 22.73% coverage rate and a 12.12% top-three rate, while loanDepot recorded a 46.97% coverage rate and a 31.82% top-three rate. The brand is present on these platforms but is not being surfaced as a leading option.

The fourth gap is cluster coverage. All 529 qualified observations fell into the Brand Recommendation cluster. PennyMac has no measured performance in Pricing & Value or Multi-Brand Comparison prompts, which means the benchmark cannot show whether the brand wins or loses when buyers ask AI systems to compare refinance options directly.

Biggest Opportunity

Questions This Section Answers

• Which platforms offer PennyMac the clearest path from being mentioned to being recommended?
• What is causing PennyMac to be cited as context on ChatGPT and Copilot instead of shortlisted?

PennyMac's biggest opportunity is closing the recommendation conversion gap on ChatGPT and Copilot, where the brand is present but rarely surfaced as a top-three option. On ChatGPT, PennyMac recorded 27 mentions across 66 observations but only 8 top-three placements. On Copilot, the brand recorded 13 mentions across 82 observations but only 1 top-three placement. These are the platforms where the gap between presence and placement is widest, and where competitors like loanDepot and Better.com are converting similar or lower presence rates into substantially higher recommendation rates.

The path from reference to recommendation on these platforms runs through the prompt types that drive comparison and selection behavior. PennyMac is being discussed, but the AI systems are not placing it in the shortlist. That pattern suggests the brand's public evidence layer is sufficient to generate mentions but not sufficient to generate recommendation credit. The fix is targeted: strengthen the citation and source footprint that AI systems use when forming recommendation lists, particularly on the platforms where PennyMac's conversion rate lags furthest behind its presence rate.

Competitive Landscape

Questions This Section Answers

• How does PennyMac's top-three rate and rank-one rate compare to Rocket Mortgage and other mortgage refinance lenders?
• What does PennyMac's average recommended rank of 3.81 say about where it lands in AI shortlists?

Rocket Mortgage holds dominant recommendation-stage strength in the mortgage refinance category, with loanDepot and Better.com operating as the strongest challengers. PennyMac sits in the middle of the tracked set, visible but under-recommended relative to its presence rate.

Average recommended rank covers rank-eligible recommendations only.

PennyMac's 7.94% top-three rate places it sixth in the tracked set, behind five competitors and ahead of four. The brand's 1.70% rank-one rate is the fifth-lowest in the set, and its average recommended rank of 3.81 indicates that when PennyMac does receive rank credit, it typically lands in the middle of the recommendation list rather than at the top. The gap between PennyMac's 37.43% presence rate and its 7.94% top-three rate is the widest conversion gap among the top six brands.

AI Response Inconsistency Alerts

Questions This Section Answers

• What conflicting claims did AI platforms make about which lender is the largest FHA lender?
• Which platforms and sources produced different answers about PennyMac's FHA lending position?

Two critical and high-severity factual inconsistencies were detected for PennyMac across four AI platforms. These conflicts involve competing claims about which lender holds the largest FHA lending position by volume, and they appear across ChatGPT, Copilot, Gemini, and AI Overviews.

The first conflict is a critical-severity factual inconsistency about which company is the largest FHA lender by overall volume. When asked "Who is the largest FHA lender?", AI Overviews stated that PennyMac is currently ranked as the largest FHA lender in the United States by overall volume and number of loans issued, citing Scotsman Guide's 2024 Top FHA Lenders ranking, Bankrate's FHA lender guide, and a Yahoo Finance article on best FHA loan lenders. Gemini, responding to the same question, stated that the largest FHA lender overall is United Wholesale Mortgage, which it said consistently leads the market in total FHA loan volume and FHA purchase loans, citing The Truth About Mortgage and Mortgage Research Center. The flagged sources show the Scotsman Guide ranking listing PennyMac first with a volume figure, while The Truth About Mortgage and Mortgage Research Center both list United Wholesale Mortgage in the top position.

The second conflict is a high-severity factual inconsistency about which company is the largest FHA lender by volume. When asked "Who is the best FHA lender?", Copilot stated that Freedom Mortgage is the number one FHA lender by volume, citing Mortgage Info, CNBC Select, and Forbes Advisor. ChatGPT, responding to the same question, stated that PennyMac was the largest FHA lender by 2025 volume in Bankrate's data, citing Bankrate's FHA mortgage lender page, CNBC Select, and a CalHFA approved lenders page. The flagged source shows Mortgage Info listing Freedom Mortgage as the top FHA volume lender with the most FHA loans originated.

These conflicts matter because they show that AI platforms are not aligned on PennyMac's competitive position in the FHA lending segment. When different platforms give different answers to the same question about market leadership, buyers receive inconsistent information about which lender holds the top position. For PennyMac, the inconsistency cuts both ways: one platform positions the brand as the volume leader, while another attributes that position to a competitor.

Prompt Evidence

AI Overviews / Brand Recommendation Prompt: "Who is the largest FHA lender?" Result: AI Overviews named PennyMac as the largest FHA lender by overall volume, citing Scotsman Guide, while Gemini named United Wholesale Mortgage for the same question.

Copilot / Brand Recommendation Prompt: "Who is the best FHA lender?" Result: Copilot named Freedom Mortgage as the top FHA lender by volume, while ChatGPT attributed the largest FHA lending position to PennyMac based on Bankrate data.

ChatGPT / Brand Recommendation Prompt: "Who is the best mortgage lender right now?" Result: PennyMac appeared in the response but was not surfaced as a top-three recommendation, consistent with the brand's 12.12% top-three rate on ChatGPT.

AI Overviews / Brand Recommendation Prompt: "Who is the best lender for mortgages?" Result: PennyMac received a valid recommendation in this prompt type, contributing to the brand's 41.27% coverage rate on AI Overviews.

What CiteWorks Studio Would Do Next

Phase 1: AI Visibility Market Discovery Audit Map PennyMac's prompt-level performance across all six platforms to identify exactly which question types produce mentions without recommendations, and which competitors are capturing the top-three slots PennyMac is missing.

Phase 2: Recommendation Readiness Plan Prioritize the ChatGPT and Copilot gaps, where PennyMac's presence-to-recommendation conversion is weakest, and define the specific prompt clusters where the brand can realistically reach top-three placement.

Phase 3: Owned Answer Layer Buildout Strengthen PennyMac's owned content so that AI systems have clear, retrievable answers to the comparison and selection questions where the brand is currently mentioned but not recommended.

Phase 4: Citation / Authority Layer Development Build the third-party source footprint that AI systems cite when forming recommendation lists, with particular focus on the review and comparison sites that dominate the top cited domains in this category.

Phase 5: Monthly AI Visibility and Recommendation Tracking Track PennyMac's coverage, top-three rate, and rank-one rate month over month to measure whether the conversion gap is closing and whether the brand is gaining ground on the leaders.

Why This Matters

AI presence alone is not enough. PennyMac appears in more than a third of qualified AI responses about mortgage refinance, but it earns a top-three recommendation in fewer than one in twelve. That gap represents buyers who see the brand mentioned but do not see it positioned as a leading option. In a category where Rocket Mortgage is named first in nearly half of all qualified observations, being mentioned without being recommended is a measurable commercial disadvantage.

The next move is targeted correction of the prompt, page, and citation layers that determine whether PennyMac is surfaced as a recommendation or merely referenced as context. The benchmark shows where the brand is losing ground. A company-level audit shows which prompts, which competitors, and which sources are driving that outcome.

Core Metrics

Sentiment Score

Questions This Section Answers

• How many of PennyMac's AI mentions are positive, neutral, or negative?
• Why are neutral mentions counted differently from positive recommendations in this benchmark?

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

For PennyMac in October 2026, the calculation is (141 × 1 + 57 × 0 + 0 × -1) / 198, which produces a score of 0.7121.

This matters because unclassified mention counts are misleading. A brand that appears in 198 responses sounds impressive until you separate those appearances into positive recommendations, neutral references, and cautionary mentions. PennyMac's 198 mentions break down into 141 positive and 57 neutral, with no negative framing. That is a healthy profile, but it does not tell you whether the brand is being recommended or merely described.

Share of voice is a diagnostic metric, not a business KPI. A positive recommendation, a neutral reference, a cautionary mention, and a competitor-displaced mention are not equal. PennyMac's 57 neutral mentions are appearances where the brand was discussed without being positioned as a recommendation. Those mentions contribute to presence rate but not to recommendation coverage. Counting all mentions as wins is bad measurement. Classified sentiment is required before interpreting AI visibility, because the difference between being mentioned and being recommended is the difference between awareness and consideration.

Sentiment by Platform

Methodology

1. This report is a benchmark-based analysis of PennyMac's AI recommendation visibility in the mortgage refinance lending category. It is not a client implementation case study and does not reflect any CiteWorks Studio engagement with PennyMac.
2. The reporting window is October 2026, with comparison data from July 2026, August 2026, and September 2026 where available.
3. Six AI platforms were tracked: ChatGPT, Copilot, Gemini, Perplexity, AI Overviews, and AI Mode. All six registered at least one qualified observation in October 2026.
4. The October 2026 benchmark analyzed 529 qualified observations from a raw collection of 800 prompt-surface observations. The July 2026 baseline analyzed 607 qualified observations from the same raw collection size.
5. The competitor universe includes ten tracked brands: Rocket Mortgage, loanDepot, Bank of America Corp., Better.com, PennyMac, United Wholesale Mortgage, New American Funding, Wells Fargo & Co., Rate, and Chase Credit Journey.
6. All 529 qualified observations in October 2026 fell into the Brand Recommendation cluster. No qualified observations were captured in the Pricing & Value or Multi-Brand Comparison clusters.
7. Stage 0 extraction retained the query, AI platform, answer, brand outcome, recommendation placement, sentiment, and, where exposed, citations or attributable evidence sources for each observation.
8. A mention is defined as any appearance of the brand in an AI response, regardless of whether the brand was recommended. A valid recommendation is defined as an observation where the brand appears in a recommendation shortlist with positive sentiment and rank eligibility.
9. Top-three rate measures the share of observations where the brand appears among the top three recommended options. Rank-one rate measures the share of observations where the brand is the first or primary recommendation. Average recommended rank covers rank-eligible recommendations only.
10. The benchmark applies a qualification step to the raw prompt-surface collection before brand-level metrics are calculated. All percentages use the qualified set of 529 observations as the denominator, not the raw collection of 800.
11. Entity-name tracking changed between July 2026 and October 2026 for several brands. Chase, Bank of America, Wells Fargo, and Guaranteed Rate recorded zero coverage under their July tracked names in September and October, while Bank of America Corp., Wells Fargo & Co., and Chase Credit Journey entered the tracked set. This report uses the October 2026 tracked names throughout.
12. Directional analysis identifies changes worth investigating. It does not establish cause. Source presence in AI answers is evidence about the information environment, not proof that a source caused a recommendation.

See How AI Is Recommending Your Brand

The public benchmark shows where PennyMac stands in AI-generated mortgage refinance recommendations. A company-level AI visibility audit maps the prompt, platform, competitor, and citation patterns behind those numbers into a prioritized strategy for closing the recommendation gap.