The North Face
Conversational shopping assistant that personalizes the product recommendation through dialogue (use case, weather, activity)
As early as 2015, The North Face tested XPS, a conversational shopping assistant on IBM Watson that recommended products by use case and weather; the pilot reached 60% click-through to recommendations and 40% more time on site.
Objective
Reproduce online the advice of a specialist salesperson: ask a few questions about the intended use and surface the right products, rather than leaving the customer to filter a catalog alone.
The deployment
In 2015-2016, The North Face put XPS (Expert Personal Shopper) online, a conversational shopping assistant built by the agency Fluid and powered by IBM Watson. The customer describes what they are looking for in natural language, for example a jacket for a hike at a given place and season; the system interprets the request, estimates the conditions (temperature, wind), and ranks the products accordingly. Launched in beta in December 2015 after a one-month pilot, the tool was at first limited to jackets. On the pilot, the average session lasted two minutes and the click-through rate to recommendations reached 60%. According to later accounts, customers who used the assistant spent 40% more time on the site and 75% said they would use it again. In November 2016, IBM acquired Fluid's XPS unit. The original setup no longer appears in service today; the pattern itself has become widespread with LLM-based conversational assistants.
Results Proof C
Official launch release (pilot figures: 60% click-through, 2-minute sessions) and consistent third-party analyst accounts for time on site and intent to reuse. No financial result; an old setup whose liveness is no longer confirmed.
How it works
Documented architectureThe stack in detail
- plateforme IBM Watson Cognitive technology (NLP) that interpreted the request in natural language and estimated the context (temperature, wind) for the place and season.
- outil Fluid Expert Personal Shopper (XPS) Conversational shopping assistant designed by the agency Fluid on Watson; the XPS unit was acquired by IBM in November 2016.
- infra Site e-commerce thenorthface.com Integration point for the assistant, at first limited to the jackets category.
How it runs, concretely
For ops teams-
1Collect the request Customer
The customer states their need in natural language (activity, place, season).
-
2Interpret and contextualize AI / conversational assistant
NLP extracts the intent and estimates the conditions (temperature, wind) for the place and period.
-
3Rank the products AI / conversational assistant
Items are ordered by fit to the stated need and context, then presented to the customer.
-
4Measure engagement E-commerce team
Time on site, clicks to recommendations, and intent to reuse serve to judge the tool.
The request expressed by the customer in natural language, crossed with context data (place, season, weather). Without a reliable mapping between intent and product attributes, the assistant recommends off target.
How your customers perceive this type of use
Sourced studiesLe paradoxe est documente des deux cotes : 71% des consommateurs attendent des interactions personnalisees et 76% sont frustres quand elles manquent (McKinsey, 2021), mais 75% declarent ne pas acheter aupres d'organisations auxquelles ils ne confient pas leurs donnees (Cisco, 2024). La « creepy line » est localisee : messages recus quelques secondes apres une recherche et suivi de localisation sont les pratiques qui mettent le plus mal a l'aise (Periscope by McKinsey, 2019).
Acceptance conditions
- La confiance dans le traitement des donnees precede l'achat : 75% ne achetent pas sans elle (Cisco 2024)
- Un cadre legal protecteur rassure : 59% des consommateurs disent que des lois fortes sur la vie privee les rendent plus a l'aise pour partager des informations dans des applications IA (Cisco 2024)
- La personnalisation elle-meme est attendue quand elle est consentie : environ la moitie des consommateurs (US 55%, UK 52%) disent s'inscrire souvent ou parfois a des services personnalises (Periscope by McKinsey 2019)
Red lines
- Le message declenche quelques secondes apres une recherche ou un achat : deuxieme ou troisieme cause de malaise selon les pays (Periscope by McKinsey 2019)
- Le suivi de localisation percu comme de la surveillance : 40% de malaise en Allemagne et au Royaume-Uni (Periscope by McKinsey 2019)
- Le mesusage des donnees personnelles par l'IA, devenu la premiere inquietude des consommateurs, a 53% et en hausse (Qualtrics 2025)
Sources: McKinsey & Company 2021 · Periscope by McKinsey 2019 · Cisco 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- A catalog with usable use-case attributes (activity, season, conditions)
- A mapping between customer intents and product attributes
- An external context source if relevant (weather, location)
Org prerequisites
- A framed conversational component (guardrails, transparency)
- An e-commerce team to maintain the intent-to-product mapping
Possible stack
- An LLM-based conversational assistant (Azure OpenAI, Google, or equivalent) connected to the catalog
- A contextual recommendation engine behind the assistant
The plan, step by step
- Step 1Pick a category with a high need for advice and map the catalog's use-case attributes (activity, season, conditions).Deliverable: Mapping of customer intent to product attributes.
- Step 2Connect a conversational assistant (on an LLM today) to the category catalog, with guardrails and transparency about its AI nature.Deliverable: Assistant in pre-production on the pilot category.
- Step 3Open in beta on part of the traffic and measure click-through to recommendations, time on site, and conversion against the classic filtered path.Deliverable: Controlled-test results.
- Step 4Iterate on the intent-to-product mapping from real conversations and decide on extending to other categories.Deliverable: Review and extension roadmap.
First step: Target a category with a high need for advice (outdoor gear, tech), connect a conversational assistant to the catalog, and measure click-through to recos and conversion against the classic filtered path.
The same pattern, proven in other industries
Function of Beauty
personalized product formulation by algorithm from a quiz
Henkel
genAI content supply chain for one-to-one personalization at scale
Hims & Hers
AI-driven personalized treatment routing on intake data (MedMatch)
Sources
- S1 The North Face, IBM and Fluid Launch New Interactive Shopping Experience using Artificial Intelligence (AI) - PR Newswire Interested party archive pending
- S2 Smarter e-shopping with The North Face and Watson - Harvard Business School (Technology and Operations Management) Secondary archive pending
- S3 IBM buys Expert Personal Shopper from Fluid to build out Watson's conversation skills - TechCrunch Established press archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.