Olivier Delangre

The through-line

Across Olivier Delangre's posts and interviews, one preoccupation never lets up: physical stores are data-blind compared to e-commerce, and closing that gap is both a technical and a mindset problem. In earlier interviews he explained the mechanics of this gap and how Amoobi's sensors fill it [15][17]; in more recent posts the framing has shifted toward industry-wide validation, pointing to Carrefour, Walmart, Leroy Merlin and others as proof that the "Connected Store" is no longer hype but an accelerating reality [10][13]. The constant underneath both phases is his insistence that AI and technology are only as good as the granular, real-world behavioral data feeding them [7][12].

On AI hype versus data

Delangre is consistently skeptical of AI conversation that isn't grounded in real inputs. At Connect Lille he noted that "the current conversation centers around AI. Everyone is talking about it, but the reality is always grounded in data" [12], and pushed the same audience on a concrete question: "How can you leverage AI for the physical store if you can't actually 'see' it?" [12]. Elsewhere he puts it more bluntly: "AI doesn't fix bad data, it just make it looks more convincing and shiny" [7], adding that retailers focused on physical stores need "quality data ... about your customers" as the precondition for anything else [7].

On the Connected Store

He treats the "Connected Store" as a real, trackable trend rather than a marketing term. Asked directly whether it "actually exist[s] in real life, or is it just industry hype," his answer was that "it's becoming a reality much faster than many think" [10], citing Carrefour's partnership with Vusion following Walmart's similar move as evidence that "physical operations are fully digitalized" [10]. He frames the payoff in category management terms: digitalizing the shelf "bridges the gap between e-commerce-style insights and brick-and-mortar reality" [10], and building that connected environment requires "granular, real-world data on how shoppers navigate your aisles" rather than "guesswork or buzzwords" [12].

On store layout and congestion

A recurring behavioral claim is that shoppers avoid friction. On the podcast circuit he coined the idea that "congestion kills conversion," arguing that excessive displays and clutter reduce sales because "shoppers naturally choose paths of least resistance and avoid congested areas" [16]. In his written posts he applies this operationally: "space creates movement" and shoppers "navigate toward open spaces where they feel comfortable moving," while something as ordinary as an employee restocking a display can create "an instant bottleneck" [9]. He also treats decision-tree-driven layout as something retailers should be deliberate about, asking whether "shoppers actually understand your store layout and how sub-categories are grouped together" [2], and observing in store visits that categories like pasta are "almost always organized as a brand block on the customer decision tree" [1].

On lost sales

Delangre frames Amoobi's purpose around sales that never happen rather than sales that do: "We track lost sales that didn't make it to the register" [6][17]. He describes the underlying system as functioning "similar to Google Analytics for brick-and-mortar locations," identifying where customers hesitate before recommending layout or adjacency changes that "typically result in 5-15% sales increases" [17].

From the stage

In podcast appearances he goes further into the technical description than his written posts do, explaining that Amoobi relies on "3D optical sensors placed in store ceilings to track customer movements and behavior" [15], and drawing the direct analogy to web analytics tools to explain the model to a general audience [17]. The "congestion kills conversion" phrase itself appears to originate from these spoken appearances rather than his LinkedIn writing [16].

Takeaways

• Treat AI claims with skepticism unless backed by granular physical-store data; "AI doesn't fix bad data, it just make it looks more convincing and shiny" [7].
• Watch for real-world Connected Store proof points, not just industry talk: Carrefour/Vusion and Walmart's computer-vision moves are cited as evidence the trend is materializing now [10].
• Audit in-store congestion actively: replenishment schedules and staff habits can create bottlenecks that shoppers avoid entirely, directly suppressing conversion [9][16].
• Measure "lost sales" rather than only completed transactions; layout and adjacency fixes informed by shopper hesitation data have produced 5-15% sales increases in Amoobi's tracked deployments [17].
• Check whether category layouts (e.g. brand-block merchandising) actually match how customers mentally navigate the decision tree, rather than assuming they do [1][2].