Nerdy delivered another quarter where cost control beat growth. The tutoring platform reported a third-quarter loss of $1.01 per share, beating the consensus estimate of a $1.50 loss by 32.7% and improving 32.7% from a year ago. Revenue of $33.19 million fell 10.3% year over year and came in 0.9% below consensus, though it was within management's $32.0 million to $35.0 million range. Fourth-quarter guidance came in below estimates. That is the central tension. Nerdy keeps shrinking its losses faster than expected, but the revenue base is now shrinking faster too, and the product transformation meant to restart growth missed the back-to-school season.

The quality of the beat deserves scrutiny. Adjusted EBITDA was a loss of $5.7 million, ahead of the guided range of negative $9.0 million to negative $6.0 million and $4.5 million better than a year ago. Margin expanded by more than 1,000 basis points. Lower sales staffing, lower AI costs and tight G&A management drove the outperformance, and headcount is now down 44% year over year versus 34% at the second quarter. Non-GAAP gross margin rose 160 basis points to 64.5%. However, that gain came mainly from lower amortization after Nerdy wrote off capitalized software, including a $1.6 million abandonment charge this quarter that pulled GAAP gross margin down to 59.7%. Part of the margin improvement is therefore an accounting byproduct of abandoned investment, not better unit economics. Expert costs actually rose as sessions per member increased. The GAAP net loss did narrow, to $12.8 million from $18.7 million.

The operating trend beneath the headline weakened. Revenue went from 2% growth in the first quarter to a 4% decline in the second and a 10% decline in the third. Part of the drop is the planned wind-down of the Varsity Tutors for Schools business. Still, Consumer revenue, which makes up 93% of the total, fell 7%. Active Members dropped 7% to 31.8 thousand, worse than the 5% decline at the second quarter. That ends four straight quarters of narrowing declines that management had highlighted on the prior call. ARPM slipped 2% to $366. Retention improved across cohorts, but new-client acquisition fell. Management attributed the shortfall to the self-service registration and checkout flow and other launches running more than three months behind plan. The call also revealed that sales headcount was deliberately under-hired on the expectation that the new platform would carry more of the load. The cost savings that helped EBITDA also limited acquisition capacity when the product arrived late.

Guidance carries the damage forward. Fourth-quarter revenue is guided to $37.0 million to $39.0 million, with adjusted EBITDA around breakeven. Combined with year-to-date results, that implies full-year revenue of roughly $162 million to $164 million. That is below the $168 million to $175 million range set last quarter, which had already been cut from $180 million to $190 million. Implied full-year EBITDA of about negative $5.6 million also falls short of the prior negative $4 million to breakeven range. Notably, the earlier goal of returning to positive member growth by the end of 2026 was not repeated. Management's tone moved from confident earlier in the year to acknowledging execution shortfalls.

The catalysts are real but unproven:

• **Messaging-first student experience:** a redesign with more than 100 agentic actions.
• **Self-service checkout:** aimed at lowering customer acquisition cost.
• **Smart Actions:** an in-session AI assistant for live tutoring.
• **K-5 experience:** a dedicated offering for elementary students.
• **Stripe:** a switch from PayPal as payment provider.

AI spending fell to $1.5 million from $2.0 million in the second quarter. Management expects the year-over-year gap in new-client acquisition to narrow as the remaining features ship, but that is an expectation, not a result.

The balance sheet also deserves attention. Free cash flow was negative $5.9 million versus negative $4.1 million a year ago, though management said it improved by about $4.5 million excluding lost VT4S prepayments and exit costs. Nerdy drew the remaining $10 million of its term loan, bringing borrowings to $30 million. The year-end cash outlook rose to $35 million to $37 million, but net of the added debt it is lower than before, at roughly $5 million to $7 million. Management says existing liquidity will fund the company through free-cash-flow breakeven, so execution now has a thinner margin for error.

The market had largely priced in disappointment before the report. Shares entered earnings at $6.23, down 48.1% since opening at $12.00 after the August report. That is about 50% below the inter-earnings high of $12.33 and 16% above the 52-week low of $5.37 set on October 29. Investor sentiment remained negative but improved, rising from -0.29 to -0.16. Earnings Whispers trend signals are uniformly negative, consistent with the steady erosion in price and sequential growth.

The bottom line is that Nerdy proved again it can cut costs, but not yet that it can grow. A narrower loss and a breakeven EBITDA target matter for a company with net cash in the single-digit millions. However, a third straight outlook reduction, re-accelerating member losses and a dropped growth target leave bears with legitimate ammunition. The fourth-quarter rollout of self-service and agentic features now has to show up in new-member acquisition, not just in management's commentary.