A Rare Canadian Global Success Story
Lululemon is one of the very few Canadian companies that has achieved sustained global success for nearly three decades. This is not an easy feat. Canada is known for its relatively high tax burden, high operating costs, and a market that is often less efficient compared with major global countries. Against this backdrop, building a premium global consumer brand is exceptionally challenging. That is why Lululemon stands out and naturally draws my attention as an internal auditor.
Founder Insight Through Observation
Lululemon founder Chip Wilson once shared that he could roughly assess a store’s profit and loss simply by walking into it. He would observe factors such as the number of staff on duty, store lighting, product pricing, customer engagement, queue length at the cashier, and overall store energy. Without opening a financial report, he claimed he could sense whether a store was performing well or struggling. This perspective highlights the power of operational awareness and real time observation.
Walk Around Management and the Role of Internal Audit
This approach closely resembles what is commonly known as walk around management. Walk around management refers to leaders spending time on the ground, observing operations directly, talking to staff, and understanding issues firsthand rather than relying solely on reports and dashboards.
From an internal audit perspective, walk around management is highly relevant. Internal auditors are not only control testers but also observers of how controls operate in reality. Store visits, warehouse walkthroughs, and face to face discussions often reveal risks, inefficiencies, or control gaps that documents alone cannot capture. Internal audit adds value by translating these observations into structured findings, root cause analysis, and practical recommendations.
An Experiment: Internal Audit vs Artificial Intelligence in Store Visits
This raises an interesting question. Can artificial intelligence support or even replace parts of a traditional internal audit store visit. As an experiment, one could take several photos of a retail store and ask an AI tool to analyze them. The AI might comment on product display, cleanliness, lighting, staff presence, customer flow, or potential safety issues. This offers a different angle compared to a human auditor walking through the store.
Comparing Human Judgment and AI Observation
A human internal auditor brings professional judgment, business context, and experience into the observation. An AI model brings speed, consistency, and the ability to process visual data at scale. Each has strengths and limitations. This experiment itself already highlights how technology may augment audit work rather than replace it entirely.
Photo 1 The referenced image was generated using Gemini for privacy reasons. I visited the Richmond Centre store on 23 December 2024.
IA: Two operational issues were observed. First, sales associates were not consistently identifiable due to the lack of a standard uniform. Some staff carried crossbody bags, while others only wore badges, making it difficult for customers to distinguish staff from shoppers during peak hours. This may negatively affect customer service and accountability.
Second, despite adequate staffing levels, no sales associate approached me for assistance for over ten minutes. Staff were observed engaging in casual conversations rather than proactively supporting customers. Possible causes include a high proportion of temporary staff, weak linkage between sales performance and incentives, or limited supervisory presence during the holiday period. Standard service procedures, such as positioning junior staff at store entrances and proactive engagement after a reasonable browsing period, did not appear to be followed.
AI: Image analysis focuses on identifying deviations from established retail standards, particularly in three areas: inventory and sell-through risk, visual merchandising compliance, and safety and loss prevention. AI can quickly highlight surface-level anomalies, but professional judgment is required to assess root causes and control effectiveness.
Photo 2
IA: Although no internal visual merchandising guideline was available, the combined photos highlight several concerns. Different product categories such as jackets, down jackets, T-shirts, jeans, and bags were displayed on the same rack. This created a cluttered appearance and made it difficult for customers to locate items efficiently. As a premium brand, Lululemon is not positioned as a small boutique or off-price retailer. Coordinated outfits should be presented on mannequins rather than mixing unrelated items on a single rack. The limited use of full-body mannequins further reduced customers’ ability to visualize product combinations.
Additionally, merchandise was not well maintained or neatly arranged. Given that this was after Black Friday and before the Christmas peak, customer traffic was relatively low, suggesting the issue was not due to operational pressure. This points to either insufficient frontline supervision or inadequate training of temporary staff.
A further concern relates to product labeling. The display tags, particularly for items such as the Flow Y Bra, were too small and did not allow customers to quickly identify product information and pricing. Effective merchandising should enable customers to understand key product details within a few seconds. Competitors such as Uniqlo demonstrate stronger practices through clear product categorization, frequent tidying, and effective use of mannequins to showcase styling.
AI: While broader risk categories were referenced, the relevant observations aligned with non-standard signage that may affect brand integrity and inconsistent mannequin styling that could confuse product positioning and customer understanding.
Photo 3 The photos were taken at the Burnaby Metrotown and Langley Willowbrook stores during the first week of January.
IA: Beyond the previously noted visual merchandising issues, the most notable concern is the lack of footwear displayed on wall racks, resulting in visible empty space and a poor in-store presentation. Initially, after observing a similar issue at the Richmond Centre store, I suspected a replenishment problem, potentially caused by Minimum Order Quantity or Reorder Point thresholds being set too high. This could prevent automatic replenishment if both display and backroom stock levels fall below ordering requirements.
However, after visiting two additional stores and observing the same condition, it appears this is unlikely to be a replenishment execution issue. A more plausible explanation relates to product strategy. As brands such as Nike and Lululemon expand their SKU range into categories outside their core strengths, such as footwear, volumes per SKU tend to be low. This reduces purchasing leverage with major suppliers, compresses gross margins, and may lead management to scale back or rationalize the footwear assortment. As a result, available shoe SKUs may be insufficient to support proper in-store display requirements.
It is possible that management has not fully considered the impact of SKU reduction on store presentation standards. From a frontline operations perspective, store managers should avoid leaving empty racks visible to customers. Even repeating the same shoe models on display would be preferable unless the company invests additional capital in redesigned fixtures. Using slower-moving inventory to maintain a full display may be a practical interim solution.
AI: Limited insights were generated for the photos. One relevant observation highlighted that excessive repetition of the same item, such as multiple identical products facing outward, may indicate stale inventory being used to mask a lack of assortment.
Why Structured Audit Methodology Still Matters
It is important to note that such experiments are only random tests. A more appropriate and professional approach is still a structured internal audit methodology. This includes a well designed store audit checklist, clearly defined audit objectives, and appropriate weighting across key areas such as inventory management, cashier controls, merchandising standards, and staff compliance.
By scoring stores consistently and monitoring results over time, internal audit can provide management with meaningful insights into store performance, control effectiveness, and emerging risks. Observation is powerful, but discipline, structure, and professional judgment are what turn observation into assurance.
Conclusion
A decade ago, I worked on a in-house tax compliance project that used OCR technology to identify sensitive information from scanned documents, which was already an early form of applied AI. While today’s large language model based AI is far more advanced, it is still not particularly effective for retail store operation audits from an internal control perspective, as image based models often lack the domain training needed to identify control issues or root causes from photos alone. However, AI remains highly valuable for deeper risk research and analysis, and when used correctly, it can meaningfully support auditors’ professional judgment. I will explore this further in a future blog post.
Comments
Post a Comment