Voxel AI-Powered Footfall Fraud: AI Analytics Providers Falsify In-Store Data to Defraud Brands of Growth Commissions

Who Gets Targeted

Victims are primarily decision-makers—often Marketing Directors or Founders—at SMEs, retail chains, and franchise brands eager for digital transformation. Their vulnerabilities are threefold: First, the 2026 'AI FOMO' creates anxiety that they will fall behind competitors. Second, they are susceptible to quantitative promises, such as '30% footfall increase,' while lacking expertise in computer vision and data precision metrics. Third, they tend to trust PPT case studies and live demos without independent third-party monitoring, often signing annual service contracts with growth commission clauses under the pressure of 'limited-time offer' sales tactics before conducting due diligence.

骗局怎么运作

• Step 1: Establishing a Technical Persona. Fraudulent companies register with 'AI' or 'Smart' in their names, claiming to have proprietary in-store vision analysis systems. They package their founding team as having backgrounds from major tech firms, create polished white papers and 'client case walls,' and leverage the 2026 retail AI hype to build professional credibility, while their actual systems are often just open-source wrappers or UI-only demos.
• Step 2: Low-Risk Bait Trials. They offer brands a three-month low-cost or free pilot, installing cameras or connecting to existing surveillance in a few stores. They claim to discuss commissions only after data is proven, using 'zero risk' and 'results-driven' rhetoric to lower decision barriers and discourage the brand from auditing data definitions.
• Step 3: Falsifying In-Store Data. During the trial, they use scripts to generate fake footfall curves or count repeat visits, employee entries, and passersby as 'valid' customers. By overlaying this with natural fluctuations during promotional periods, they produce impressive reports showing a 20-40% increase in footfall. The interface is designed to be professional and interactive, making the brand believe the data is real.
• Step 4: Inducing Growth Commission Contracts. At the end of the trial, they use the 'validated' data to push the brand into a 1-3 year growth commission agreement, stipulating a 5-15% service fee based on footfall or sales growth, often including high penalty clauses and exclusive data interface authorization, using urgency tactics like 'the next quarter's schedule is already full.'
• Step 5: Harvesting and Disappearing. Once the commission period begins, the brand notices no actual revenue growth. The provider blames poor store execution or statistical discrepancies while continuing to collect monthly fees. If the brand stops paying, the provider uses the penalty clauses to exert pressure, or simply dissolves the company entity after signing enough clients and restarts under a new brand name.

红旗信号（看到这些快跑）

• 🚩 The provider refuses to allow the brand to connect to an independent third-party monitoring system, insisting that all data must be viewed in their proprietary backend with no raw data export for cross-verification.
• 🚩 Footfall growth data is clearly disconnected from actual store revenue, receipt counts, and payment transaction logs; the reports look great, but the POS system shows no corresponding growth.
• 🚩 The brands listed on the 'case wall' cannot provide a verifiable contact person, and the 'top-tier chain' partnership is supported only by a single blurry screenshot of an agreement.
• 🚩 The contract requires exclusive data interface authorization and sets high penalty clauses, yet provides no compensation terms for the provider's own data inaccuracy.
• 🚩 The company avoids fixed service fees by claiming to be 'performance-based,' but is vague about the commission calculation base and footfall definitions, refusing to provide clear written definitions.
• 🚩 Live demos are impressive, but the provider refuses to conduct blind tests in stores designated by the brand, performing only in pre-arranged 'model stores'.

真实案例

• In 2026, CCTV's '315' report exposed the AI model data poisoning black market. A purely fictional product can reach the top of AI recommendation lists within hours for about 39.9 yuan, proving that data and content fraud in the AI chain is low-cost and hard to verify—the same logic applies to footfall data fraud.
• In June 2026, CCTV reported that police dismantled cross-border e-commerce AI scams. Victims were lured by 'easy money' claims in livestreams, starting with small deposits to see fake backend order data, then induced to invest more. The 'order flow' could never be withdrawn, mirroring the tactics used to forge operational data to sign commission contracts.
• In February 2024, Hong Kong police reported a deepfake video conference scam where fraudsters used AI face-swapping to impersonate a UK company executive, stealing approximately 200 million HKD. This highlights the lethality of AI forgery in B2B decision-making; relying solely on demos and materials provided by the other party is extremely risky.
• During the 2026 '618' shopping festival, 21st Century Business Herald reported on AI-driven fake reviews and 'brushing.' Merchants used AI tools to generate bulk fake reviews and simulate human interaction to create the illusion of high sales, far exceeding manual capabilities and confirming that AI-driven data fraud has permeated all aspects of retail marketing.

Official Stance

• In July 2026, the Cyberspace Administration of China announced the first phase of the 'Qinglang' campaign, focusing on AI data poisoning and the failure to label synthetic content, while opening a reporting channel for AI-related abuses.
• In March 2026, the CCTV '315' Gala and media outlets like 36Kr exposed the AI model poisoning black market, revealing the chain of low-cost AI output forgery and fake product information, warning enterprises to be wary of data fraud under the guise of AI.
• In June 2026, the China Consumer News reported on new forms of AI-driven consumer fraud, noting that fraud has evolved from image/text tampering to multi-sensory deception, urging companies and consumers to verify AI-generated promotional and data materials.
• In June 2026, 21st Century Business Herald's '618' report warned that AI-driven fake reviews have become a major e-commerce issue. As regulators and platforms crack down on fake transactions, companies must cross-verify any data they rely on.

How to Protect Yourself

• ✅ Before signing, require the provider to open raw data interfaces and introduce independent third-party monitoring or internal sampling audits. Insist on the three principles: data must be exportable, auditable, and cross-verifiable. Reject 'black box' reports.
• ✅ Insist on blind tests in stores randomly selected by the brand. During the test, keep the locations unknown to the provider until the last minute, and cross-check footfall data against independent metrics like POS transaction logs, receipt counts, and delivery order volumes.
• ✅ Clearly define the statistical metrics for footfall and growth in the contract. Include clauses for immediate termination and full refunds plus damages in the event of data fraud. Reject exclusive authorizations and asymmetric penalty clauses.
• ✅ Use the National Enterprise Credit Information Publicity System to check the provider's establishment date, equity structure, and litigation records. Check for administrative penalties and 'abnormal operation' listings. Verify every brand on their case wall by phone and request a contactable project manager.
• ✅ Use installment payments and set an observation period. Pay only fixed costs for the first phase without signing commission agreements. Observe for at least one full business cycle to confirm that revenue and footfall are growing in sync before discussing commission models.