How AI Search Is Recommending Student Loan Refinance: Monthly Trends

Executive Summary

SoFi remains the category leader for student loan refinance recommendations. Valid recommendation coverage for SoFi held stable across the two-month series, easing to 59.1% of qualified observations in September 2026 from 61.1% in the August 2026 baseline. Earnest remained in second place, at 56.3% in September compared with 60.0% at baseline, putting the gap between the top two brands at 2.8 points.

Across the ten tracked brands, PNC Bank was the only brand whose movement this month is flagged by the benchmark as a significant change: its valid recommendation coverage fell from 3.8% at baseline to 1.5% in September 2026, driven mainly by a decline in raw presence. Citizens, Earnest, ELFI, Laurel Road, LendKey, MEFA, RISLA, SoFi, and Splash Financial all moved within the normal range of month-to-month variation on this metric.

Within that normal range, Citizens posted the largest gain, up 2.7 points to a 28.6% coverage rate in September 2026, and ELFI recorded the largest percentage decline, down 5.8 points to 32.5%. These movements sit inside expected variation for the category, but they mark where the distribution of recommendation credit is loosening in the middle of the field.

Each monthly run begins with 800 prompt-surface observations (570 unique questions in September 2026, versus 562 in August 2026) across the benchmark's defined AI/search surface universe. Of those, 800 mentioned a tracked brand or competitor in both months; 594 were relevant and 206 were irrelevant in September 2026, compared with 588 relevant and 212 irrelevant in August 2026. The public metrics use the 455 qualified observations in September 2026 and 447 in August 2026 that survive both qualification stages.

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.

These figures set the qualified base against which every brand-level percentage in this report is calculated.

Benchmark-Level Metrics

The qualified observation count rose by 8 month over month, a small expansion of the analysis set. The share of answers that took a recommendation-shaped form declined, as did the share that produced a valid recommendation shortlist, indicating that September's prompt mix produced more factual or pricing-oriented responses and fewer direct recommendation lists.

AI Recommendation Trend

Coverage by Brand

SoFi held the top position across the two-month series, with Earnest in second place throughout. PNC Bank was the only brand whose move this month is classified as significant; the remaining brands, including Citizens' gain and ELFI's decline, moved within the normal range of month-to-month variation for their respective observation counts.

What Changed This Month

SoFi: Leader Holds, but the Gap Narrows

SoFi's coverage of 59.1% in September 2026 is down 2.0 points from 61.1% in August 2026. The brand remains first in the category, but the cushion over second place has compressed to 2.8 points, down from 1.1 points at baseline, as Earnest's decline was steeper at 3.7 points.

SoFi's raw presence actually rose slightly, from 94.2% to 94.7% of observations, meaning the brand appears in nearly all AI-generated answers. The decline came in how often it was recommended: top-three placement eased from 49.7% to 46.2%, and rank-one appearances fell from 27.5% to 25.7%, or 123 down to 117 observations. Sentiment held steady at 0.7.

The distinction for SoFi is that it remains the most visible brand in the category by a wide margin, visible in 431 of 455 observations in September 2026. The coverage decline reflects a shift in recommendation placement, not a loss of presence.

Highest-priority diagnostic: Which prompt types are no longer placing SoFi at rank one, and which brand is taking those top slots?

Earnest: Placement Eases While Presence Holds

Earnest's coverage fell 3.7 points to 56.3% in September 2026, from 60.0% in August 2026. The brand's raw presence was essentially flat at 92.3%, down from 92.4%, indicating the decline was not driven by disappearance from answers.

The sharper story is in top-three placement, which fell 6.7 points from 47.6% to 40.9%. Rank-one appearances dropped from 17.2% to 13.0%, or from 77 down to 59 observations. This suggests Earnest is still present in AI answers but is being recommended lower in the list or less prominently.

Earnest's average recommended rank slipped from 2.17 to 2.29, reinforcing that the brand is appearing further down in recommendation sets. Sentiment held at 0.7, so the change is about placement, not perception.

Highest-priority diagnostic: Which surfaces are demoting Earnest from top-three positions, and what is the competing evidence source?

Citizens: The Only Gain Among Tracked Lenders

Citizens rose 2.7 points to 28.6% coverage in September 2026, from 25.9% in August 2026, making it the sole brand to gain meaningful ground. Raw presence was essentially flat at 43.1%, down slightly from 43.6%, meaning the gain came from converting existing presence into stronger recommendations.

Top-three placement eased from 9.2% to 7.5%, but rank-one appearances ticked up from 1.3% to 1.5%, or from 6 to 7 observations. The coverage gain was driven by a broader expansion of valid recommendations, from 116 to 130 observations, even as top-three placement softened.

The gap analysis shows Citizens narrowing its deficit to ELFI from 12.4 points to 3.9 points across the series, and to Earnest from 34.1 points to 27.7 points. Citizens is the quiet riser in a category where most brands lost ground.

Highest-priority diagnostic: Which prompts are now returning Citizens as a valid recommendation, and what evidence sources are supporting those answers?

PNC Bank: Significant Decline in a Small-Count Window

PNC Bank's coverage fell 2.3 points to 1.5% in September 2026, from 3.8% in August 2026, a movement the benchmark flags as significant for this brand's typical range. In absolute terms, valid recommendations dropped from 17 to 7 observations, a small but meaningful contraction.

The decline was driven by raw presence, which fell from 7.6% to 3.3%, from 34 to 15 observations. The brand also recorded its first negative sentiment observation, a single instance that pulled its net sentiment score from 0.6 to 0.4. Top-three and rank-one rates were unchanged at 0.9% and 0.0%, respectively.

PNC Bank operates in a small-count window, so shifts of a handful of observations register as large percentage movements. The presence decline of 19 observations accounts for nearly all of the coverage change, while sentiment and placement were secondary factors.

Highest-priority diagnostic: Which prompt categories stopped mentioning PNC Bank, and which competitor absorbed those mentions?

ELFI: Largest Raw Decline Without Significance

ELFI posted the largest percentage decline of any tracked brand, down 5.8 points to 32.5% in September 2026 from 38.3% in August 2026, a movement that remained within the normal range for this brand this month. Valid recommendations fell from 171 to 148 observations.

Raw presence declined from 54.1% to 50.5%, accounting for most of the coverage change. Top-three placement was relatively steady at 20.0%, down from 20.8%, and rank-one rate was unchanged at 0.9%. The brand's average recommended rank held close to steady, moving from 3.51 to 3.46.

The gap analysis shows ELFI's lead over MEFA narrowing from 37.4 points to 30.7 points, and Citizens closing the gap to ELFI substantially. ELFI remains in third place but its cushion over fourth has compressed sharply.

Highest-priority diagnostic: Which surfaces or prompt types are driving ELFI's presence loss, and what evidence sources are no longer citing the brand?

Buyer-Intent Interpretation

In September 2026, all 455 qualified observations fell into the brand recommendation class. No observations were qualified into pricing and value or multi-brand comparison clusters. The public benchmark therefore measures which lenders AI systems recommend for student loan refinance, but it cannot yet speak to how those systems handle pricing transparency, cost trade-offs, or direct head-to-head comparisons between lenders. Those commercial questions require a company-level analysis that can isolate pricing and comparison prompts specifically.

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 capturing the query, the AI surface that generated the answer, whether a recommendation occurred, the rank of any recommended brand, the sentiment of the response, and citations where exposed. Company-level analysis can go deeper into prompt, competitor, surface, and evidence patterns to understand why AI systems recommend some lenders over others. Source presence is not automatically treated as proof of causation.

About This Benchmark

This report is part of CiteWorks Studio's AI Industry Market Discovery research program.

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

Report-Specific Interpretation Notes

• Several brands in this vertical operate with small observation counts, so percentage movements can appear large even when the absolute number of recommendations changed by only a handful.
• The qualified denominator (455 observations in September 2026) is smaller than the raw collection universe (800 prompts), because only prompts that were relevant and delivered a usable AI answer enter the public metrics.
• Month-over-month movement identifies changes worth investigating; it does not by itself establish the cause of those changes. Directional analysis requires prompt-level review.

Next Step

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

Beneath the aggregate percentages sit the questions that drive commercial outcomes: which high-intent prompts a brand wins, which competitor takes the recommendation when a brand loses, what attributes AI systems associate with each lender, and which external sources shape those answers. A lender can see from this benchmark that its coverage moved, but the public percentages cannot identify the prompts, competitors, or sources causing the result.

A company-specific AI visibility audit maps those prompt, surface, competitor, ranking, sentiment, and evidence-source patterns into a prioritized visibility strategy. It answers the "why" behind the movement and identifies the specific levers that can improve a brand's standing in AI-driven recommendation.

Request an AI visibility audit