How AI Search Is Recommending Mortgage Refinance Lenders: Monthly Trends

Executive Summary

Rocket Mortgage remains the category leader in AI search recommendations for mortgage refinance lenders, with valid recommendation coverage of 63.7% in September 2026. That is down from 71.2% in July 2026, a decline of 7.5 points that exceeded normal month-to-month variation. The second-position gap widened sharply this month: loanDepot moved into second at 36.5%, while the brands that previously held the top spots behind Rocket Mortgage — Chase at 54.5% and Bank of America at 55.4% in July — recorded zero coverage under their prior tracked names in September.

The most consequential movement this month was a reconfiguration of how tracked brands are named in the benchmark. Chase, Bank of America, Wells Fargo, and Guaranteed Rate all fell to 0.0% coverage as tracked entities in September 2026; simultaneously, variants including Bank of America Corp., Wells Fargo & Co., and Chase Credit Journey entered the tracked set with measurable coverage for the first time. Within the category, loanDepot was the strongest upward mover, rising to 36.5% from 33.9% in July, with its August-to-September gain of 8.2 points exceeding normal variation. Rocket Mortgage remained the strongest brand overall despite its coverage decline, holding the highest top-three rate at 44.2% and the highest rank-one rate at 30.5%.

The August 2026 run recorded 381 qualified observations as a smaller collection month, while September returned to a larger base of 570 qualified observations. The combination of a reconfigured tracked brand-name set and a larger observation base makes September 2026 a month where category rankings shifted meaningfully, and where the public benchmark now requires careful reading of which named entities are actually being tracked.

Each monthly run begins with 800 prompt-surface observations (571 unique questions in September 2026) across the benchmark's defined AI/search surface universe. Of those, 800 mentioned a tracked brand or competitor; 692 were relevant and 108 were irrelevant. The public metrics use the 570 observations that survive both qualification stages. For comparison, the July 2026 baseline began with 800 prompt-surface observations (541 unique questions), of which 800 mentioned a tracked brand or competitor; 624 were relevant and 176 were irrelevant, yielding 607 qualified observations. The August 2026 intermediate run collected 500 prompt-surface observations (381 unique questions), yielding 381 qualified observations.

Key Findings

Benchmark Context

The report separates the raw collection universe from the qualified analysis set. Brand-level recommendation percentages are calculated within the qualified benchmark set.

The qualified benchmark set is the denominator behind every brand-level metric that follows; the raw counts above show how that set was built. The August 2026 intermediate run collected 500 raw prompts and produced 381 qualified observations, making September 2026's 570 qualified observations a return toward the July baseline scale. The table below summarizes category-level metrics computed from the qualified set for each month.

Benchmark-Level Metrics

AI Recommendation Trend

A Reconfigured Brand-Name Set Reshapes the Category Ranking

Rocket Mortgage retained first position in September 2026, but the second and third positions shifted as several previously tracked brand names recorded zero coverage while alternate entity names entered the benchmark. loanDepot moved into second place, and Bank of America Corp. appeared in third in its first tracked month.

Multiple brands exceeded normal month-to-month variation in September 2026, and the category-level change came from a combination of significant declines among previously tracked names, significant rises among newly tracked variants, and more modest movements among the stable middle tier. Rocket Mortgage's decline of 7.5 points exceeded normal variation, as did the declines for Better.com, PennyMac, and United Wholesale Mortgage. loanDepot's rise against the July baseline was not itself significant, but its August-to-September gain of 8.2 points did exceed normal variation. The brands that fell to zero coverage — Chase, Bank of America, Wells Fargo, and Guaranteed Rate — each recorded declines beyond normal variation, though these movements are tied to the entity-name reconfiguration described below rather than a clearly established loss of recommendation strength.

What Changed This Month

Rocket Mortgage

Rocket Mortgage's valid recommendation coverage measured 63.7% in September 2026, compared with 71.2% in July 2026. The decline of 7.5 points exceeded normal month-to-month variation, though the August-to-September movement of 1.6 points stayed within normal range. Valid recommendations numbered 363 in September from a qualified base of 570.

Presence remained near saturation at 98.2%, essentially unchanged from 97.7% in July, with the brand appearing in 560 of 570 qualified observations. Placement narrowed alongside the coverage decline: the top-three rate fell from 55.4% to 44.2%, and the rank-one rate fell from 38.1% to 30.5%, with the brand's top-three count falling from 336 to 252.

The distinction to notice: Rocket Mortgage is still present in nearly every response and still leads the category on every placement metric, but it is now named first or within the top three less often than in July.

Highest-priority diagnostic: Which prompt types shifted Rocket Mortgage from first or top-three placement to lower positions, and which brand took the vacated rank-one spot?

loanDepot

loanDepot rose to second place in September 2026 with valid recommendation coverage of 36.5%, up from 33.9% in July 2026. The August-to-September gain of 8.2 points exceeded normal variation, making this the sharpest upward movement among brands tracked across the full series. Valid recommendations grew from 206 in July to 208 in September against a slightly smaller qualified base.

Presence rose from 43.7% to 56.1%, a gain of 12.4 points, with the brand appearing in 320 of 570 observations. Placement metrics were mixed: the top-three rate held at 13.3%, unchanged from July, while the rank-one rate slipped from 2.1% to 1.8%. Net sentiment moved from 0.82 to 0.68.

loanDepot's rise came from broader presence across responses rather than stronger placement within them. The brand is being mentioned in more answers, but its share of top-three placements has not climbed with it.

Highest-priority diagnostic: Which surfaces and prompt families account for the added loanDepot presence, and is the brand gaining mention share without gaining recommendation share?

Bank of America and Bank of America Corp.

Bank of America, as tracked in July and August 2026, recorded valid recommendation coverage of 55.4% in July and 51.2% in August. In September 2026, that entity name recorded 0.0% coverage with no qualifying observations, while Bank of America Corp. entered the tracked set at 27.9% coverage. Tracking the two named entities together, the gap between them moved by 83.3 points across the series.

Bank of America Corp. achieved its 27.9% coverage from 159 valid recommendations among 570 qualified observations. Presence measured 44.9%, with the entity appearing in 256 observations. Its top-three rate was 11.2%, and its rank-one rate was 0.9%, with only 5 first-place finishes. Net sentiment was positive at 0.66.

The commercial reading: the underlying lender may retain meaningful AI presence, but the benchmark captured it under a different tracked name in September. The apparent collapse of Bank of America and the rise of Bank of America Corp. line up with a change in the benchmark's tracked entity names, not necessarily a change in how AI systems treat the lender.

Highest-priority diagnostic: Did AI responses shift their preferred naming convention for this lender, and how should the benchmark track the entity consistently across future months?

Chase and Chase Credit Journey

Chase recorded valid recommendation coverage of 54.5% in July 2026 and 50.4% in August 2026, holding third position in both months. In September 2026, Chase as tracked recorded 0.0% coverage, while Chase Credit Journey entered the tracked set at 0.9% coverage with only 5 valid recommendations and 8 mentions across 570 observations.

Chase's decline to zero coverage, a drop of 54.5 points against the July baseline, is one of the two largest movements in the series this month, alongside Bank of America's decline of 55.4 points. Chase Credit Journey's entry-level coverage does not offset that decline: its presence rate was just 1.4%, and it recorded no rank-one finishes.

The distinction to notice: Chase's disappearance from the tracked set under its original name is far larger in magnitude than the emergence of Chase Credit Journey. If AI responses are referring to the lender by other names not yet captured in the tracked set, the benchmark would understate the lender's true presence.

Highest-priority diagnostic: Which entity names are AI systems using when discussing Chase mortgage refinance offerings, and do those names map back to the same underlying brand?

Wells Fargo and Wells Fargo & Co.

Wells Fargo recorded coverage of 20.8% in July 2026 and 24.7% in August 2026. In September 2026, Wells Fargo as tracked recorded 0.0% coverage, while Wells Fargo & Co. entered the tracked set at 7.2% coverage from 41 valid recommendations. Presence for Wells Fargo & Co. measured 18.6%, with 106 mentions across 570 observations.

The top-three rate for Wells Fargo & Co. was just 0.2%, with only 1 top-three placement, and the rank-one rate was 0.0%. Net sentiment measured 0.40, the second-lowest reading among tracked brands this month. The declines for Wells Fargo of 20.8 points against baseline and 24.7 points against August, both exceeding normal variation, occurred in the same month that Wells Fargo & Co. entered the tracked set — a pattern the benchmark cannot separate from an underlying change in recommendation strength.

What the user should notice: the Wells Fargo & Co. entity is present in responses but almost never recommended in the top three. This reads as a brand that is discussed but not surfaced as a leading option under its full corporate name.

Highest-priority diagnostic: Is the underlying lender losing recommendation strength, or is it being tracked under a name that appears less frequently in AI responses?

Guaranteed Rate

Guaranteed Rate recorded coverage of 15.2% in July 2026 and 7.3% in August 2026. In September 2026, the brand recorded 0.0% coverage, with no valid recommendations or qualifying mentions in the tracked set. No alternate entity name for Guaranteed Rate entered the benchmark this month.

The brand's coverage fell in both months of the series; the August-to-September decline of 7.3 points on its own exceeded normal variation, extending a two-month decline that removed the brand entirely from the tracked set by September.

Unlike Chase, Bank of America, and Wells Fargo, no successor entity name captured Guaranteed Rate's coverage this month, making its removal from the benchmark a clean disappearance rather than an apparent renaming.

Highest-priority diagnostic: Are AI responses simply not mentioning Guaranteed Rate in refinance recommendations, or is the benchmark's tracked-prompt set no longer capturing the queries where the brand previously appeared?

PennyMac

PennyMac's valid recommendation coverage measured 20.0% in September 2026, down from 27.7% in July 2026, a decline that exceeded normal variation. The August-to-September movement was a modest gain of 0.8 points, indicating the decline occurred primarily between July and August and has since stabilized.

Presence measured 41.2%, essentially unchanged from 40.5% in July. The top-three rate was 7.5%, up slightly from 6.9%, while the rank-one rate fell from 1.8% to 1.2%. Valid recommendations numbered 114 in September versus 168 in July. Net sentiment declined from 0.72 to 0.50.

PennyMac's pattern is distinct: presence stayed flat while coverage fell, meaning the brand was mentioned in roughly the same share of responses but recommended in fewer of them.

Highest-priority diagnostic: Which competitors are taking recommendation share from PennyMac when the brand appears in responses but is not recommended?

Better.com

Better.com's coverage measured 27.2% in September 2026, down from 33.8% in July 2026, a decline that exceeded normal variation. The August-to-September decline of 3.2 points was not significant on its own, with the two-month decline spread fairly evenly across both months. Valid recommendations numbered 155 in September versus 205 in July.

Presence fell from 41.3% to 31.6%, and the top-three rate fell from 16.5% to 11.4%. The rank-one rate moved from 2.3% to 1.9%. Net sentiment remained the strongest among tracked brands at 0.93.

Better.com recorded a genuine two-month decline in presence and top-three placement, not a naming reconfiguration. The brand slipped from the upper mid-tier toward the middle of the tracked set.

Highest-priority diagnostic: Which prompt families drove the loss in top-three placements, and which competitor gained the recommendation when Better.com dropped out?

United Wholesale Mortgage

United Wholesale Mortgage's coverage fell from 13.3% in July 2026 to 8.4% in September 2026, a decline that exceeded normal variation. Valid recommendations fell from 81 to 48. Presence rose slightly from 25.5% to 27.7%, while the top-three rate fell sharply from 9.1% to 4.0%.

The rank-one rate moved from 2.3% to 1.6%. Net sentiment fell from 0.57 to 0.30, the lowest among tracked brands in September.

United Wholesale Mortgage's decline came through weakening placement: the brand was mentioned more often but recommended far less often within the top three.

Highest-priority diagnostic: Which competitive comparisons are pushing United Wholesale Mortgage out of top-three recommendation slots?

New American Funding

New American Funding recorded coverage of 16.0% in September 2026, down from 19.9% in July 2026. This decline of 3.9 points did not exceed normal variation, and the August-to-September movement was an increase of 3.9 points. The brand's classification is stable despite the baseline-to-current decline.

Presence measured 19.5%, close to the 21.2% recorded in July. The top-three rate was 3.2% versus 3.8%, and the rank-one rate was 0.4% versus 0.5%. Valid recommendations numbered 91 in September versus 121 in July. Net sentiment stayed high at 0.83.

New American Funding stabilized after its August dip, with September's recovery bringing it back within normal range of the July baseline.

Highest-priority diagnostic: Which surfaces drove the September recovery, and can that presence be sustained into the next cycle?

Rate

Rate entered the tracked set in September 2026 at 5.6% coverage, a rise from zero that exceeded normal variation, with 32 valid recommendations and 44 mentions across 570 observations. The brand's top-three rate was 1.6%, and its rank-one rate was 0.2%, with a single first-place finish. Net sentiment was positive at 0.80.

Rate's entry is notable because it is not a variant of a previously tracked brand name; it appears in the tracked set for the first time this series.

Highest-priority diagnostic: Which prompts surfaced Rate as a recommendation, and is the brand gaining durable AI recommendation presence, or is this connected to the same tracked-entity changes affecting other brands this month?

Buyer-Intent Interpretation

All 570 qualified observations in September 2026 fell into the Brand Recommendation cluster, as did all observations in July and August 2026. No observations were captured in the Pricing & Value or Multi-Brand Comparison clusters, which means the public benchmark cannot yet answer price, value, or head-to-head comparison questions. The commercial questions these clusters would answer — which lender offers the best refinance rate, or how lenders stack up directly — remain outside the current measurement.

Brand Opportunity Summary

The benchmark identifies where attention is warranted; a company-level analysis is needed to explain why.

Evidence Behind the Benchmark

The aggregate metrics are built from prompt-level observations (query, surface, recommendation outcome, rank, sentiment, and citations where exposed). Company-level analysis can go deeper into prompt, competitor, surface, and evidence patterns. Source presence is not automatically treated as proof of causation.

About This Benchmark

This report is part of the LLM Authority Index AI Market Discovery research program.

• AI Industry Market Discovery Methodology
• AI Industry Market Discovery Metrics
• AI Industry Market Discovery Standards

Report-Specific Interpretation Notes

• Small-count movement: Guaranteed Rate dropped out of the tracked set entirely in September, while Chase Credit Journey, Rate, and Wells Fargo & Co. operate on fewer than 50 valid recommendations. Percentage movements for these entities carry less stability than the leaders.
• Entity-name change: Chase, Bank of America, Wells Fargo, and Guaranteed Rate recorded zero coverage under their July/August tracked names in September, while Bank of America Corp., Wells Fargo & Co., and Chase Credit Journey entered the tracked set with measurable coverage. The benchmark cannot distinguish whether this reflects a change in AI-generated naming conventions or a change in which entity names the benchmark tracked this month; the coverage swings described above should be read with that caveat.
• Qualified denominator vs raw collection: all percentages use the qualified set (570 observations in September 2026, 607 in July 2026), not the raw 800 collected prompts.
• Month-over-month movement identifies changes worth investigating; it does not establish cause.

Next Step

The Public Benchmark Shows Where a Brand Is Winning or Losing. A Company-Level Audit Shows Why.

The aggregate percentage hides the questions that matter: which high-intent prompts are won, which competitor takes the recommendation when a brand loses ground, what attributes AI associates with each option, and which external sources shape those answers. For the brands that fell to zero coverage under their tracked names in September 2026, the benchmark shows the magnitude of the change but not whether the underlying lender lost recommendation strength or merely shifted to a different entity name.

A company-specific AI visibility audit maps those prompt, surface, competitor, ranking, sentiment, and evidence-source patterns into a prioritized visibility strategy.

Request an AI visibility audit