Table of Contents

1. Key Highlights:
2. Introduction
3. SmartSales and Athena: Two AI initiatives with distinct operational aims
4. Why intake-to-listing velocity matters for a consignment business
5. SmartSales and the growth playbook: aligning marketing, sales and retail
6. Physical retail playing a strategic role beyond customer acquisition
7. Demand is holding up at the high end — why that matters
8. Financial signals: growth, margin improvement and operational leverage
9. How AI alters unit economics and the labor mix
10. Authenticity at scale: technical and human safeguards
11. Risks and limits: what could derail the strategy
12. Competition and the broader resale ecosystem
13. Implementation blueprint: how Athena could work in practice
14. Governance, auditing and transparency: maintaining trust in AI decisions
15. What to watch as the transformation continues in 2026
16. Balancing automation with human expertise: a hybrid model as the prudent path
17. Strategic outlook: path to becoming the definitive authority in luxury resale
18. FAQ

Key Highlights:

• The RealReal positions artificial intelligence as a multi-year operational advantage, centering two initiatives—SmartSales (seller targeting) and Athena (faster, more accurate listing and authentication)—to expand high-value supply and compress time from intake to live listing.
• Physical retail and expert-led intake are driving materially higher average selling prices, while buyer demand for premium items shows resilience; Q4 2025 GMV rose 22% YoY and full-year GMV reached $2.13 billion with positive adjusted EBITDA each quarter.
• The combination of AI-driven targeting, improved throughput, and a coordinated growth playbook links supply quality, pricing power, and margins; success depends on balancing automation with expert oversight and managing fraud, bias and regulatory risks.

Introduction

Earnings messages normally revolve around the quarter’s numbers. The RealReal used its latest results to signal something broader: a strategy that treats artificial intelligence not as a feature but as the scaffolding of a redesigned luxury resale operation. The company’s CEO, Rati Levesque, highlighted two named initiatives—SmartSales and Athena—that aim to tackle the twin constraints that have plagued consignment marketplaces for years: sourcing steady flows of high-quality inventory and moving items through authentication and listing fast enough to capture demand.

Those operational fixes matter because they alter the company’s product mix, unit economics and competitive position. The RealReal isn’t merely chasing volume; the strategy is to shift toward stickier, higher-ticket consignors and categories—watches, handbags, jewelry—while using digital models and in-store expertise to maintain authenticity and lift average selling prices. The firm’s Q4 and full-year 2025 results show early financial validation of that play: accelerating gross merchandise volume, revenue growth and sustained adjusted EBITDA profitability. The coming year will test whether AI can become a durable moat or simply another productivity tool in a crowded resale market.

The following analysis unpacks the technical and commercial mechanics of SmartSales and Athena, explains why physical retail matters more than many expect, evaluates the financial implications, and identifies the operational and regulatory risks that will determine whether this approach scales.

SmartSales and Athena: Two AI initiatives with distinct operational aims

The RealReal has labeled its AI roadmap with two names for a reason: they address different points along the supply chain. SmartSales is a frontline intelligence layer for the company’s customer-facing teams; Athena is an operational engine for intake, authentication and description.

SmartSales: seller identification and prioritization

• Purpose: Increase conversion of potential consignors and steer outreach toward "mid to high-value" sellers who are more likely to consign premium items.
• Functionality: Predictive models ingest CRM history, engagement signals, purchase and browsing data, past consignments and perhaps external signals (social activity, local affluence indices) to rank leads. The output is not a one-size outreach list but contextualized suggestions for luxury managers—flagging which clients deserve phone calls, in-store appointments or targeted offers.
• Expected impact: Lower customer acquisition cost for high-value sellers, higher average sold price per consignor, and a better return on sales and marketing spend.

Athena: compressing intake-to-listing while improving accuracy

• Purpose: Reduce the time and human labor required to move items from receipt to a live listing without increasing authentication errors.
• Functionality: Computer vision to classify brand and item type; natural language processing to extract descriptors from seller-supplied text; automated condition assessment models trained on historical graded examples; pricing models that propose list prices and estimate likely sell-through and time-to-sale; confidence scores that route uncertain items to specialized human experts (gemologists, watchmakers, or senior authenticators).
• Expected impact: Faster inventory turnover, higher throughput per authenticator, ability to meet seasonal demand spikes, and a more predictable supply pipeline.

These initiatives are complementary. SmartSales funnels better inventory into intake; Athena converts that intake into market-ready listings faster. The combination improves both the quality and velocity of supply, which together affect pricing power and conversion on the demand side.

Why intake-to-listing velocity matters for a consignment business

A marketplace’s ability to monetize an item depends on three factors: the item’s intrinsic value, the speed with which the marketplace can make that item available to buyers, and the platform’s credibility in guaranteeing authenticity. For consignment models, authentication and listing are often the slowest, most expensive parts of the revenue chain.

Slow intake translates into three economic penalties:

• Capital tie-up: Inventory locked in warehouses or processing centers carries storage and handling costs while producing no revenue.
• Missed market windows: Demand for seasonal or scarce items is time-sensitive. An item that lands weeks after peak demand will attract lower bids.
• Seller dissatisfaction: Long lag times reduce consignor retention and encourage sellers to use alternatives with faster payouts.

Athena promises to compress that lag. A faster path to listing increases the number of items live at any time without proportionally increasing headcount. When throughput rises, average daily GMV can scale faster than employee growth, improving operating leverage.

Practical example: imagine a boutique that consigns a collection of vintage watches during a high-demand quarter. If intake and authentication take 30 days, many of those pieces might miss peak buyer interest or competitive bidding windows. If Athena can reduce that to 7–10 days while keeping authentication error rates low, the same set of watches will likely fetch higher prices and clear sooner, improving take-rate economics and consignor satisfaction.

SmartSales and the growth playbook: aligning marketing, sales and retail

Levesque described a "growth playbook" that knits together marketing, sales and retail. SmartSales sits at the center of that orchestration by enabling sharper segmentation and resource allocation. The playbook has several moving parts:

• Data-driven lead scoring: Not all consignors are equal. Historical sale prices, category preferences, and engagement signals help prioritize outreach. Models can identify lapsed high-value buyers who own luxury items now ripe for consignment.
• Channel optimization: Different outreach tactics work best for distinct seller segments. High-touch sellers—those likely to consign jewelry or watches—benefit from in-person appointments and expert evaluation. Mid-value sellers may convert through targeted email or local drop-off offers.
• Incentive alignment: Commission structures, referral incentives and store-level KPIs can be tuned to reward employees who bring in high-quality inventory, not just raw consignor numbers.
• Creative marketing: Content that juxtaposes new retail prices against resale savings, demonstrations of the authentication process, and testimonials from sellers who received premium prices help prime prospective consignors.

SmartSales adds two advantages: it reduces guesswork and it scales human expertise by telling staff where to spend their time. A luxury manager with finite appointment slots will see higher ROI when AI flags consignors who own items with a high expected selling price.

Real-world parallel: High-end real estate agents use lead-scoring systems to prioritize homeowners with the highest likelihood to list; similar logic applies here, but with a narrower focus on items that command top resale prices due to brand, condition, and rarity.

Physical retail playing a strategic role beyond customer acquisition

The RealReal’s bricks-and-mortar presence contributes more than foot traffic. The company reports that roughly 25% of new consignors flow through retail locations and that items sourced in-store average five to seven times higher selling prices than other channels. That magnitude demands scrutiny.

Why retail yields higher-ticket consignments

• Expert assistance: In-store gemologists and watchmakers can authenticate and appraise on the spot, building seller confidence and enabling quicker consignor decisions.
• Curation and trust: A physical environment lets the seller interact with brand-trained staff, touch paper and leather, and get nuanced condition assessments.
• Local market effects: Certain stores in affluent neighborhoods naturally draw consignors with luxury inventories, producing geographic clustering of high-value items.

Strategic implications

• Higher per-unit margins: High-ticket items carry outsized revenue even if they represent a minority of unit volume.
• Lower acquisition friction: Seeing a specialist in person reduces seller hesitation compared to remote submissions.
• Learning feed for AI: Expert appraisal outcomes from retail intake can provide labeled training data to improve Athena’s classification and condition models.

Retail is not free capital. Maintaining stores, staffing experts, and logistics adds cost. The trade-off is deliberate: invest in a physical presence where it generates a multiplier effect on the quality of intake and the velocity of conversion.

Comparable examples: Luxury resale boutiques often host appraisal events to source valuable consignments; high-end watch marketplaces employ watchmakers to inspect movements and provenance. Those examples reflect the same dynamic: physical expertise reduces uncertainty and unlocks higher ticket-size consignments.

Demand is holding up at the high end — why that matters

Levesque noted that buyers spending $5,000 or more rose 20% year over year. That uptick signals demand resilience in premium categories, with implications for the platform’s pricing curve and inventory mix.

Drivers of resilient premium demand

• Value-minded luxury buyers: Shoppers who want luxury brands but prioritize value increasingly see authenticated resale as the efficient path to higher-tier items.
• Rarity and provenance: Collectible watches and limited-edition handbags appreciate or hold value over time, attracting investors and enthusiasts.
• Supply constraints in primary markets: Long wait lists, retail allocation policies and price increases in primary channels nudge buyers to secondary markets.

The RealReal’s positioning — an intersection of luxury and value — makes it the kind of marketplace buyers turn to when they want premium goods without full retail price. If the platform continues to attract more high-ticket buyers, it can sustain price floors for sought-after brands and further incentivize high-value consignors.

Economic feedback loop: Better supply attracts higher-spending buyers, who in turn generate stronger realized prices that justify pursuing more high-value consignors. SmartSales accelerates the loop by identifying and securing those consignors; Athena ensures the items hit the market before demand fades.

Financial signals: growth, margin improvement and operational leverage

The company’s Q4 2025 results and full-year numbers provide concrete evidence that the strategic shift is producing measurable outcomes.

Key figures (as reported)

• Q4 2025: Gross merchandise value (GMV) up 22% year over year; revenue up 18%; adjusted EBITDA of $22 million.
• Full-year 2025: GMV grew 16% to $2.13 billion; revenue rose 15% to $693 million; adjusted EBITDA positive in every quarter of the year for the first time.

Interpreting those metrics

• GMV growth outpacing revenue growth suggests either stable or slightly compressed take rate, or a shift in category mix where higher-ticket items lift GMV but may take longer to convert into recognized revenue.
• Positive adjusted EBITDA across all quarters signals operational discipline and improved operating leverage. For marketplaces reliant on human experts and logistics, sustaining adjusted EBITDA reveals successful margin management—likely due to a combination of better mix, cost controls, and early AI-driven productivity gains.
• The Q4 adjusted EBITDA of $22 million is significant because fourth quarters often include holiday-driven inventory turnover; maintaining profitability during a high-cost period indicates durable margin progress.

Beyond headline numbers, investors should watch unit economics:

• Take rate (platform revenue divided by GMV): Changes indicate pricing power and mix effects.
• Customer acquisition cost (CAC) for high-value consignors: SmartSales aims to lower CAC for top-tier sellers; if realized CAC falls, LTV/CAC ratios improve.
• Days-to-list and days-to-sale: Shorter intervals increase throughput and reduce inventory holding costs.
• Authentication accuracy and return rates: These affect trust, buyer retention, and fraud-related losses.

The interplay of these metrics will determine whether AI investments deliver sustained margin expansion or merely one-time efficiency gains.

How AI alters unit economics and the labor mix

The promise of AI is not limited to cost-cutting. It changes where and how labor is deployed while shifting capital allocation.

Throughput and productivity

• Authentication throughput rises when AI handles routine classification and pre-screening. Human authenticators concentrate on high-risk or high-value items, increasing effective output per expert.
• Writing descriptions and generating standard listing copy becomes largely automated. Human editors focus on nuanced storytelling for flagship items or curated collections.

Labor composition

• Demand for specialists shifts: fewer staff needed for routine tasks; more needed for quality control, model oversight, and exceptions handling.
• Up-skilling: Staff will require training to interpret AI confidence scores and to make judgment calls on routed items.
• Workforce reductions are not the sole outcome; redeployment toward seller relationships and in-store experience can offset some automation-driven headcount declines.

Cost structure

• Fixed costs for model development and infrastructure increase, while variable labor costs per item fall. The result is higher operating leverage once the AI infrastructure carries sufficient volume.
• Capital expenditures may rise in the near term for image capture setups, lab-quality equipment in select stores, and model validation pipelines.

Return on AI investment depends on adoption scale. Small batch benefits are visible, but large returns require consistently higher throughput and lower marginal authentication costs across thousands of items per month.

Authenticity at scale: technical and human safeguards

Authentication is the trust anchor of any luxury resale marketplace. AI can accelerate processes, but it can also introduce new failure modes. The RealReal’s strategy of routing low-confidence items to human experts acknowledges that hybrid systems will be necessary for a long time.

Technical tools

• Computer vision models trained on millions of labeled images can recognize brand signatures, stitching patterns, hallmarks, serial numbers and hardware details.
• Metadata extraction from seller-submitted images and receipts helps corroborate provenance.
• Sensor-enabled image capture (multi-angle photos, macro shots) improves model fidelity.

Human safeguards

• Specialist checks: gemologists and watchmakers perform mechanical and material analyses unavailable to pure-vision models.
• Random audits and blind re-checks: These maintain quality control and populate training datasets to correct model drift.
• Dispute resolution workflows: Clear policies for returns, refund thresholds, and loss-sharing with consignors reduce legal friction.

Fraud evolves. Sophisticated counterfeiters adapt to recognition systems. Continuous model retraining, adversarial testing and a layered authentication approach remain necessary.

Example: sneaker marketplaces have found that high-resolution macro images and material testing reduce false negatives. The RealReal can apply comparable techniques for jewelry and watches, but those require equipment and human expertise that cannot be scaled purely with software.

Risks and limits: what could derail the strategy

Several risks could undermine the AI-driven transformation, each requiring different mitigation measures.

Accuracy and false positives

• Over-reliance on automated models could lead to authenticated counterfeits slipping through, damaging brand trust and creating liability.
• Conversely, overly conservative models may flag too many items for human review, undermining throughput gains.

Model bias and data quality

• Training data skewed toward popular brands or styles can reduce accuracy on rare, vintage or region-specific items.
• Models might misclassify items from underrepresented markets, creating geographic or demographic bias in acceptance and pricing.

Operational scaling

• Living systems require continuous retraining, labeling and validation. If the cost of maintaining model accuracy is high, the theoretical cost savings may not materialize.
• Peak season surges may still require temporary human scaling; AI does not eliminate the need for physical infrastructure.

Regulatory and legal exposure

• Authentication mistakes create fraud risk that could escalate into class-action suits or regulatory scrutiny.
• Data privacy rules govern how customer data feeds into SmartSales. Misuse or opaque profiling could invite legal challenges.

Competitive responses

• Rivals may accelerate their own AI or in-store expert programs. Luxury brands could expand certified pre-owned operations, narrowing the available high-value supply.

Market cyclicality

• Premium demand could weaken in a broader economic downturn, compressing realized prices and reducing the incentive for sellers to consign high-value goods.

Mitigation strategies

• Layered verification combining AI with specialist audits.
• Conservative risk thresholds for automated authentication, especially for items above CRITICAL_PRICE_LEVELS.
• Transparent seller and buyer protections, including robust return policies and insurance for high-value shipments.
• Continuous investment in data labeling and a monitored production environment for models.

Competition and the broader resale ecosystem

The RealReal operates in a populated ecosystem: pure marketplaces, niche platforms, auction houses and increasingly, luxury brands’ own resale programs. Each competitor offers different trade-offs in authenticity guarantees, speed, fees and brand alignment.

Marketplace rivals

• StockX emphasizes authenticated marketplaces for sneakers, watches and streetwear with centralized authentication centers, leaning on scale and standardized testing.
• eBay has invested in authentication programs for sneakers and watches and benefits from an enormous active buyer base and flexible seller options.
• Vestiaire Collective focuses on curated luxury consignments and has built a strong presence particularly in Europe through specialized curation and shipping logistics.

Luxury brand programs

• Several luxury houses have signaled or launched certified pre-owned initiatives, which can siphon premium consignors—especially those loyal to brand channels.
• Certified programs often include brand-backed refurbishing and provenance, offering an alternative to third-party resale.

The RealReal’s differentiator is a mix of curated service, physical expertise, and now AI-driven operational scale. Its success will depend on converting that differentiation into measurable advantages—lower acquisition costs for high-value consignors, better sell-through rates, and a reputation for reliable authenticity.

Competitive response scenarios

• If rivals match Athena-like automation, differentiation will shift to brand partnerships, unique consignment pipelines (e.g., estate sales, auctions) and exclusive in-store events.
• If The RealReal establishes a reliable moat with AI-enabled throughput and superior data, it could capture share in high-value categories that competitors find harder to scale profitably.

Implementation blueprint: how Athena could work in practice

A practical implementation of Athena resembles a pipeline of automated steps with human interventions at defined thresholds. Below is a hypothetical flow that reflects real-world engineering and operational best practices.

1. Intake capture
   • Standardized image capture protocol (multiple angles, macro, serial numbers) for in-store submissions; seller app prompts for required shots for remote submissions.
   • Automatic ingestion of seller-supplied receipts, provenance documents and purchase histories.
2. Pre-processing
   • Image enhancement to correct lighting and remove artifacts.
   • Optical character recognition (OCR) on receipts and authentication tags.
3. Brand and category classification
   • Computer vision model assigns probable brand, model, and category with a confidence score.
   • Uncommon brands or vintage items flagged for human review.
4. Condition assessment
   • Models estimate wear level, scrapes, missing hardware, and likely need for refurbishment.
   • Condition grade proposed (e.g., new, excellent, good, fair).
5. Pricing recommendation
   • Dynamic pricing model suggests list price and a range for expected sale time, integrating recent completed sales, seasonality and platform demand signals.
   • Suggests reserve/starting price for auctions when appropriate.
6. Authentication confidence routing
   • Items with confidence below threshold routed to human specialists.
   • High-confidence items proceed to automated listing creation, with a "verified by AI" badge and human audit sampling.
7. Listing generation
   • Auto-generated title, bullet points, condition notes, and pricing tags using templates tuned for brand and category.
   • SEO-optimized copy reflecting keywords buyers use (brand, model, year, materials).
8. Post-list checks and monitoring
   • Track seller returns and buyer disputes to feed back into model training.
   • Monitor model drift using a holdout human audit set.
9. Continuous learning
   • Maintain a closed-loop labeling system where human-reviewed items augment training data.
   • Active learning prioritizes labeling of items where model confidence is low or outcomes vary significantly.

This blueprint reduces manual labor on standard items while preserving expert oversight for edge cases. The setup requires investment in image capture standards, model retraining pipelines, and a robust human-in-the-loop workflow for quality control.

Governance, auditing and transparency: maintaining trust in AI decisions

Implementing AI at scale requires governance scaffolding. The RealReal will need to address explainability, auditability and customer-facing transparency.

Explainability

• Provide human-friendly explanations for pricing suggestions and authenticity decisions—e.g., “Authentication flagged due to serial number mismatch” or “Price based on recent comparable sales for [brand/model/year].”
• Internally, maintain decision logs linking model inputs to outputs for later review.

Audit trails

• Keep immutable records of images, metadata, and model versions applied to each item.
• Use independent audits and third-party reviews for high-value categories to validate systems periodically.

Transparency with users

• Clearly describe when AI assisted in authentication and what safeguards exist.
• Offer recourse: disputes, returns and reimbursement policies should be explicit and actionable.

Model governance

• Assign a cross-functional committee (AI ethics, legal, product and operations) to review model changes and high-risk deployments.
• Implement monitoring thresholds for false positives, false negatives and dispute rates.

Good governance reduces legal exposure, improves public trust and helps transform AI from a black box into a predictable tool.

What to watch as the transformation continues in 2026

Several milestones and metrics will indicate whether the AI-led strategy is gaining traction or encountering limits.

Operational metrics

• Average time from intake to live listing (target reductions and actual figures).
• Items authenticated per authenticator per day (productivity gains).
• Percentage of listings generated end-to-end by Athena without human editing.
• Fraction of new consignors sourced via retail and the associated average selling price for those items.

Financial and marketplace metrics

• GMV growth and take-rate trends: Is GMV rising faster than revenue or vice versa?
• Adjusted EBITDA margins over multiple quarters: sustained improvement will validate operating leverage.
• Repeat buyer rates and buyer cohort LTV: crucial for revenue stability.
• CAC for high-value consignors and LTV/CAC ratios.

Quality and trust indicators

• Authentication dispute rate and return frequency for authenticated items.
• Number and severity of any counterfeit incidents and the company’s average resolution time.
• Third-party audit results for authentication accuracy.

Strategic milestones

• Broader deployment of SmartSales across markets and integration with retail staff incentives.
• Expansion of Athena’s capabilities to new categories such as high-end jewelry or rare watches.
• Partnerships or integrations with brands or certification bodies to enhance provenance signals.

Investor focus will center on whether AI investments translate into durable competitive advantages: better supply without proportional cost increases, improved realized prices, and higher buyer lifetime values.

Balancing automation with human expertise: a hybrid model as the prudent path

The RealReal’s public statements already reflect a hybrid approach—use AI to improve scale and speed while relying on human expertise for high-value and high-risk decisions. That balance is not merely pragmatic; it’s essential for credibility.

Principles for the hybrid model

• Threshold-based automation: establish confidence thresholds for fully automated processing, partial automation with human oversight, and mandatory human review.
• Continuous reallocation of human capital: as routine tasks become automated, redeploy staff to relationship management, fraud investigations, and quality checks.
• Investment in tooling: provide human experts with AI-assisted interfaces that surface relevant model outputs, comparable sales, provenance indicators and magnified image views.

The hybrid philosophy avoids both extremes: the inefficiency of all-human operations and the reputational peril of all-automated authentication that cannot convincingly explain its decisions.

Real-world illustration: a watchmaker uses an AI-derived list of serial number anomalies and wear patterns as a preliminary triage; the expert then confirms mechanical condition and provenance. This collaboration raises throughput while preserving the trust that buyers and sellers expect.

Strategic outlook: path to becoming the definitive authority in luxury resale

Levesque framed The RealReal’s work as more than incremental improvements. The goal is to establish the company as a definitive authority in luxury resale by combining better inventory sourcing, faster listing and credible authentication. Achieving that requires simultaneous execution on several fronts:

• Continue investing in AI to improve seller targeting (SmartSales) and intake efficiency (Athena).
• Expand the data feed from retail experts to further train and validate models.
• Maintain explicit human oversight for edge cases and maintain transparent customer protections.
• Track unit economics closely, ensuring AI investments improve margins and LTV metrics.
• Prepare for competitive and regulatory pressures with robust governance and public communications.

If the company sustains GMV growth while preserving or improving profitability, and if AI demonstrably increases throughput without compromising authenticity, The RealReal can redefine scale economics in segments of the secondary luxury market that historically depended on human judgment.

FAQ

Q: What are SmartSales and Athena, and how do they differ? A: SmartSales is a predictive-sales tool that scores and prioritizes consignor leads, helping frontline staff focus on sellers likely to offer higher-value items. Athena is an operational AI platform that accelerates intake, classifies items, recommends pricing, and routes low-confidence cases to human specialists. SmartSales targets supply acquisition; Athena focuses on throughput and listing accuracy.

Q: How will AI affect authenticity and trust? A: AI improves the speed and consistency of preliminary authentication by recognizing brand markers, stitching, serial numbers and wear patterns. However, AI alone cannot replace specialized human expertise for rare or complex items. A hybrid system—automated triage with mandatory human review for flagged or high-value items—maintains trust while increasing efficiency.

Q: Will tech automation lead to job losses? A: Automation will reduce the need for human labor on repetitive tasks, but it also creates demand for higher-skilled roles: model trainers, data scientists, senior authenticators and seller relationship managers. The net employment impact depends on how the company redeploys staff from routine tasks to value-added roles.

Q: Why does physical retail matter if AI is scaling online processes? A: Stores provide direct access to high-value consignors, offer expert-led appraisals that reduce uncertainty, and generate labeled data that improves AI models. Items sourced in-store typically command much higher selling prices because of on-the-spot authentication and targeted selling strategies.

Q: What financial metrics should investors watch to judge success? A: Key indicators include GMV growth, take rate, adjusted EBITDA margins, days-to-list, items authenticated per authenticator per day, CAC for high-value consignors, buyer repeat rates, and authentication dispute rates. Improvements in these metrics will validate AI-driven operational benefits.

Q: What are the main risks to the AI-driven approach? A: Primary risks include authentication errors (false negatives/positives), model bias and data quality issues, cost of continuous model maintenance, regulatory or legal exposure from authentication failures, and competitive pressure from rivals or luxury brands building their own certified resale channels.

Q: How quickly will AI impact the business materially? A: Some benefits—like more efficient lead prioritization—can show results in months. Throughput and margin improvements from Athena will appear as the system processes larger volumes and confidence thresholds are optimized; expect more visible structural changes within 12–24 months as models mature and labeled data accumulate.

Q: Could other resale platforms copy this strategy? A: Competitors can adopt similar AI tools, but The RealReal’s advantage lies in combining curated retail, a large authenticated dataset, specialist expertise and cross-functional coordination. The durability of the advantage depends on how quickly rivals can replicate both the data depth and the human-in-the-loop processes.

Q: What would a failure look like? A: A failure would manifest as rising authentication disputes, reputational damage from counterfeit incidents, stagnating supply because SmartSales misprioritizes consignors, or escalating costs to maintain model accuracy that negate throughput gains. Sustained margin erosion would signal strategic misalignment.

Q: What makes this strategy defensible over the long term? A: Defense rests on data, expertise and execution. A platform that accumulates high-quality labeled examples of authenticated items gains a training advantage. Pair that with in-store experts feeding validated examples and a governance model that preserves trust, and the company builds a data moat that is costly and time-consuming for competitors to replicate.

---

The RealReal’s latest earnings message is not a promise of instant transformation. It is a roadmap: invest in AI to change how supply is found and processed, lean into retail expertise where it pays off, and measure success through improved GMV, better margins and sustained customer trust. Execution will determine whether the company moves from a successful reseller to the definitive authority in luxury resale.