Upstart’s Trustpilot Signal Just Flashed Red—But the Real Story’s Under the Hood

TL;DR: Upstart’s Trustpilot Index (UTPI) rolled over hard in early August. Hitting the Q3 review target now requires a late-quarter sprint that UTPI doesn’t support. Meanwhile, mix shift (Auto/HELOC/SDL), smaller average loan sizes, and lower revenue-per-loan complicate the read-through. Our conservative math lands Q3 revenue around $262.8M, but balance-sheet loan sales could add roughly $11.4M, putting results near (but just shy of) the $280M guide—with the caveat that UTPI undercounts channels that don’t generate Trustpilot reviews. [1–3, 6–10]

UTPI’s August Slide

UTPI retraced from 33.5 to 21.5 by Sept 3 (–35.8%). Momentum broke in August.

The 28-day UTPI fell from 33.5 to 21.5 by September 3—a 35.8% retrace from the recent high. Through the first 65 days of Q3, Upstart still posted 1,833 Trustpilot reviews (+13.5% vs the comparable Q2 period), but momentum broke in early August. The model’s Q3 estimate sits at 2,654 reviews, revised down from 2,931 in early July. [1]

Why it matters: UTPI is a solid proxy for personal-loan flow through upstart.com. When the proxy rolls over, you re-test your assumptions about origination volume and product mix. [1]

The Math: 30 Reviews a Day or Bust

There are 27 days left in Q3. To hit 2,654 reviews, Upstart needs 821 more—30.41/day, about +41% above the current UTPI of 21.5. If UTPI just cruises at today’s run-rate, Q3 likely lands near 2,413 reviews—basically flat with Q2’s 2,422. [1]

Context: Q2 revenue was $257M, while Q3 guidance is ~$280M. Management guided that ~$275M would come from fees and ~$5M from net interest income (NII)—i.e., Q3 depends on origination volume, not carry. [2–3]

Credit Backdrop: A Sticky Macro

Upstart’s own UMI (Upstart Macro Index) shows default pressure sitting in the mid-1.5x range for July and slightly higher month-over-month, signaling persistent macro friction for unsecured credit. [6] In parallel, U.S. consumer confidence dipped again in August and labor demand softened (JOLTS job openings fell), both of which argue for consumer caution and tighter capital. [7–8]

Translation: If borrowers wait for better rates and capital stays picky, personal-loan demand cools at the margin—and that shows up in UTPI. [6–8]

Why UTPI Can Mislead Right Now

Auto/HELOC/SDL now >10% of originations—channels that generate fewer Trustpilot reviews.

Auto, HELOC, and small-dollar loans (SDL) are no longer rounding errors—they’re meaningful (10%+) contributors and growing. Those products are often originated through banks/credit unions and dealer channels, which don’t generate Trustpilot reviews at anywhere near the rate of core personal loans. [4–5]

Result: UTPI undercounts originations as the product mix shifts, making the August drop look worse than the underlying business. Add to that: management has been moving R&D loans off the balance sheet throughout 2025 and aims to transition most of the funding for these products off-balance-sheet by year-end—incremental revenue that never touches UTPI. [10]

Déjà Vu: Flat Reviews ≠ Acceleration

We’ve seen a flat-to-sluggish UTPI stretch before. Back in Q4 ’24 → Q1 ’25, Trustpilot reviews were roughly flat while revenue didn’t accelerate. Today, loans on the books are larger than they were then, and the new products mix is bigger, which further weakens UTPI’s predictive power for total revenue. [2–3, 10]

The Unit-Economics Drag: Smaller Average Loan Sizes, Lower Rev/Loan

Avg. personal-loan size fell to $7,193 in Q2 ’25 (–23.1% from Q1 ’24).

Management commentary around Q2 flagged smaller average loan sizes (mid-$7Ks) and a double-digit sequential decline attributed to model changes and SDL mix. That dynamic pressures revenue-per-loan even when volumes hold. [9]

Big picture: More loans, less revenue per loan—unless take rates rise or the mix tilts back toward larger personal loans. [2–3, 9]

A Conservative Baseline for Q3

Chart showing 22% quarter-over-quarter decline in Upstart revenue per loan in Q2 2025.

If you assume:

• UTPI holds at ~21.5/day for the final 27 days (≈2,413 Q3 reviews total),

• Review-rate inches lower as mix skews to Auto/HELOC/SDL, and

• Revenue-per-loan slips further given the smaller average loan size,

…you land near $262.8M for Q3 revenue. That’s below the headline guide, but it doesn’t factor a key swing item. [1, 2–3, 9]

The Swing Item: Balance-Sheet Loan Sales

Management has been explicit: the goal is to transition most Auto/HELOC/SDL funding off the balance sheet by year-end 2025. [10] A reasonable (not guaranteed) Q3 scenario is that some portion of these R&D loans are sold during the quarter, adding incremental fee revenue that doesn’t show up in UTPI. Conservatively, that could add ~$11.4M—putting the quarter around $274M, near the $280M guide without requiring a UTPI rebound. (This is an illustrative sensitivity, not company guidance.) [2–3, 10]

The Rebellionaire Take

Don’t trade the Trustpilot headline; trade the mix math.

• Momentum clearly stalled in August (UTPI). [1]

• UTPI undercounts as Auto/HELOC/SDL scale through non-Trustpilot channels. [4–5]

• Average loan sizes are smaller and rev/loan is lower, consistent with SDL growth. [9]

• Baseline math: ~$262.8M; with likely loan-sale tailwind, ~$274M—close to guide but not a blowout. [2–3]

This quarter isn’t about “AI magic.” It’s about blocking and tackling—managing mix, unloading balance-sheet risk, and keeping revenue inside the strike zone while consumer credit cools. If the market trades the UTPI downtick without accounting for mix and loan-sale dynamics, there’s room for mispricing—in both directions. [1–3, 6–10]

---

Sources:

1. HFF UTPI Dataset — Upstart Trustpilot Index (UTPI) Review, Sept. 4, 2025. Internal analysis of Trustpilot reviews, 28-day UTPI, and Q3 regression estimates. (Halter Ferguson Financial / Henry Dierkes)

2. Upstart Q2 2025 Earnings Press Release / 8-K (Aug 5, 2025). Revenue: $257M; other quarterly metrics. Guidance for Q3: $280M total revenue. SEC

3. Upstart Q3 2025 Guidance Detail (fees vs NII). Company guidance indicates $275M fees / $5M NII for Q3; corroborated in IR materials and coverage. Q4hqInvesting.com

4. Auto & HELOC scale in Q2 2025. Company IR presentation (Q2 2025) and coverage highlight Auto ~$114M and HELOC ~$68M originations; new products surpass 10% of mix. ir.upstart.comInvesting.com

5. Auto channel growth context. AutoFinanceNews: Upstart auto originations up ~533% YoY to $114M in Q2 2025. Auto Finance News

6. UMI (Upstart Macro Index) – July 2025. Upstart’s official UMI site and activity log show mid-1.5x defaults vs baseline and a July uptick. Upstart+1

7. Consumer Confidence – August 2025. The Conference Board: Index fell to 97.4 in August (from 98.7 in July). The Conference Board

8. Labor Market Softening – JOLTS (Sept 3, 2025). Reuters recap of BLS data: job openings decreased, hiring tepid—evidence of cooling demand. Reuters

9. Average Loan Size Commentary. PYMNTS coverage of Q2 call: average loan size in mid-$7Ks, double-digit sequential decline attributed to model changes and SDL mix. PYMNTS.com

10. Balance-Sheet Strategy for R&D Loans. Post-earnings coverage noting management’s goal to transition most Auto/HELOC/SDL funding off-balance-sheet by end-2025 (aligns with IR commentary). Morningstar