{"slug": "the-fans-who-chatted-for-ten-minutes-to-pick-a-seat-at-spa", "title": "The fans who chatted for ten minutes to pick a seat at Spa", "summary": "Spa Grand Prix deployed Jack, a voice and text AI concierge built by Napster, on its 2026 Belgian Grand Prix ticket shop, where Napster reports the agent handled 54,647 conversations in 21 languages over a 62-day sales window and lifted checkout clickthrough by 70%. Napster's case study, published 14 September 2026, anticipates €539,000 in extra revenue, with more than 1,000 sessions running past ten minutes and about 2,000 users viewing the 360-degree seat preview. Napster chief product officer Patrick Conway said Jack's answers were checked against client-approved material, with Napster putting the pre-launch non-hallucination rate at about 99%, though the case study provides no baseline or comparison group.", "body_md": "7 min read\n\n[Spa Grand Prix](https://www.spagrandprix.com/en/) put a Belgian Grand Prix AI concierge called Jack on its ticket shop. Napster, which built it, says Jack handled 54,647 conversations in 21 languages and lifted checkout clickthrough by 70%.\n\n## **TL;DR**\n\n- Spa Grand Prix added Jack, a voice and text AI sales agent, to its Belgian Grand Prix ticket site, where it showed seats, matched budgets and added extras to the basket.\n- Napster says Jack handled 54,647 conversations in 21 languages over a 62 day sales window, and more than 1,000 sessions ran past ten minutes.\n- Napster reports 70% higher checkout clickthrough and anticipates €539,000 in extra revenue.\n- The public case gives no baseline or comparison group, so the next step is a clean holdout measured on paid orders.\n\n## Ten minutes to choose a seat\n\nMore than 1,000 times during the sales window for the 2026 Belgian Grand Prix, a fan stayed in a conversation with an AI for over ten minutes. They were working out where to sit.\n\nIt makes sense once you look at what a ticket for Spa-Francorchamps really involves. A buyer has to pick a grandstand, picture the view, stay inside a budget and then start on parking, camping or hospitality, and each choice leans on the last. If the buyer and the ticket desk don’t share a language, the whole thing gets harder.\n\n[Spa Grand Prix](https://www.spagrandprix.com/en/), the race organiser, put an answer in the middle of its shop. Jack, the Belgian Grand Prix AI concierge built by Napster, could talk or type with a fan, ask about budget and preferences, show a 360 degree view from a seat and drop the choice into the basket. It could then suggest parking, meals, camping or hospitality to go with it.\n\nFor the organiser, that made Jack a sales assistant as much as a help desk. [Napster’s case study](https://www.napster.com/case-studies/turning-fans-into-buyers-at-the-formula-1-belgian-grand-prix), published on 14 September 2026, says checkout clickthrough rose 70% after Jack launched. The number deserves attention, and it comes with one big gap, which we’ll get to.\n\n## A Belgian Grand Prix AI concierge in 21 languages\n\nA Belgian race sells to a whole continent, and a continent comes with different languages, payment habits and travel plans. Jack worked as a multilingual AI concierge across 21 languages. A fan could switch language in the middle of a conversation while the concierge kept hold of what they had already chosen, so one sales flow served every language.\n\nThe build moved fast. Napster says it took about one week. In an [aiblLIVE practitioner account](https://aiblmedia.com/downloads/ai-concierge-belgian-grand-prix/) released after the event in London on 20 October 2026, Napster chief product officer [Patrick Conway](https://ie.linkedin.com/in/patrick-conway-2a7b4058) described a training setup where Jack’s answers were checked against material the client had approved. Napster puts the non-hallucination rate at about 99% before launch. The size of the test set and the failure threshold haven’t been published, which makes them good questions for the next round.\n\nThe lesson travels well for European software and event companies. Instead of building a separate sales flow for each market, a team can build one governed layer and let it speak whatever language the buyer brings.\n\n## The cart is the point\n\nMost chatbots stop at answering questions. As an AI ticket sales agent, Jack sat close enough to the seat inventory and the basket to turn an answer into an action, and that shortens the road between “I’m not sure” and a click on checkout. The lesson is to give the system one clear commercial job.\n\nThat job also explains the metric. Spa Grand Prix tracked checkout clickthrough, a step far closer to revenue than sessions, message counts or satisfaction scores. Over the 62 days, Napster reports 54,647 conversations. About 2,000 people used the 360 degree seat view, and more than 1,000 sessions ran past ten minutes. Napster also estimates that the automation saved almost 4,000 staff hours.\n\nThe add-ons follow the same logic. Parking and camping passes make sense once Jack knows enough about a fan’s trip, so the order of the conversation matters as much as the offer itself.\n\n## What a 70% lift is a lift in\n\nThe 70% describes checkout clickthrough, the moment a fan moves from the basket towards payment. The €539,000 is Napster’s anticipated extra revenue. Both come from a live race sale, and both leave a question open. Napster hasn’t shared the baseline, the order completion rate, average order value, refunds or how traffic was split, so nobody outside can yet say how much of the lift came from Jack and how much from demand, ticket availability or other changes on the site.\n\nThat’s a normal gap for a first public case. The useful question for AI checkout conversion is what a clickthrough lift becomes once payment completes. Clickthrough can rise while paid orders stay flat if poorly matched buyers reach the cart, or if sold out seats show up later. The published evidence is enough to justify a controlled test of the Napster AI agent. It isn’t yet enough to book €539,000 as proven extra revenue.\n\n## The experiment at a glance\n\n| Element | What is documented | \n|---|---|\n| Growth constraint | A complex ticket catalogue serving buyers across many languages | \n| Hypothesis | A voice and text sales agent connected to seat views and the basket could move more visitors towards checkout | \n| Intervention | Jack answered questions, matched preferences and budget, showed seats, filled baskets and offered add-ons | \n| Audience and duration | 54,647 conversations across 21 languages over 62 days, as reported by Napster | \n| Primary reported result | 70% higher checkout clickthrough | \n| Commercial estimate | €539,000 in anticipated additional revenue | \n| Still to be shown | Baseline, comparison group, traffic split and how revenue was attributed | \n\n## The next test\n\nThe best next experiment for the Belgian Grand Prix AI concierge is a small one. Show Jack to a random share of eligible visitors, keep a clean group that never sees it, and decide in advance that paid orders are the main outcome. Keep checkout clickthrough as a diagnostic. Report results by language, device and new versus returning visitor, with minimum sample sizes set before the test starts.\n\nIf the paid order lift holds, test how much Jack should be allowed to do. Give one version answers only, a second seat and ticket recommendations, and a third the power to add parking or camping to the basket. That would show whether the value comes from faster answers, better recommendations or direct action in the cart.\n\nFans stayed with the Belgian Grand Prix AI concierge for ten minutes at a stretch. The question for the next race is whether those conversations turn into paid tickets at a higher rate than before.\n\n## Questions answered\n\n## **What did the Belgian Grand Prix AI concierge do?**\n\nIt answered ticket questions by voice or text in 21 languages, showed seat views, matched preferences and budget, added selections to a basket and offered parking, meals, camping or hospitality.\n\n## **Did the AI agent increase revenue?**\n\nNapster estimates €539,000 in additional revenue and reports 70% higher checkout clickthrough. The public material doesn’t yet include a baseline, comparison group or attribution method, so the revenue figure is Napster’s estimate and a controlled test would confirm it.\n\n## **What can another growth team copy?**\n\nGive an agent one clear commercial task, connect it to the smallest set of catalogue and basket actions it needs, check it against approved answers before launch and measure paid orders against a clean holdout.\n\n## Sources and disclosure\n\nThis story draws on [Napster’s 14 September 2026 case study](https://www.napster.com/case-studies/turning-fans-into-buyers-at-the-formula-1-belgian-grand-prix) and the [aiblLIVE case account](https://aiblmedia.com/downloads/ai-concierge-belgian-grand-prix/) released after the 20 October 2026 event. Napster built Jack and published the performance figures, and aiblLIVE notes that figures in its case library come from the speakers. Spa Grand Prix commercial director [Vanessa Maes](https://be.linkedin.com/in/vanessa-maes-54611112) is named in the case material.\n\n## Related reading\n\n- [Radisson’s four-part framework for off-platform conversions](https://industrycontents.com/radisson-chatgpt-conversion/)\n- [The AI chatbot that turned roadside questions into sales](https://industrycontents.com/atlas24h-ai-sales-agent-revenue/)\n\n**Growth Signal Index** [Browse the full archive of AI growth experiments](https://industrycontents.com/ic-lab/growth-signal-index/).", "url": "https://wpnews.pro/news/the-fans-who-chatted-for-ten-minutes-to-pick-a-seat-at-spa", "canonical_source": "https://industrycontents.com/belgian-grand-prix-ai-concierge/?utm_source=rss&utm_medium=feed", "published_at": "2026-10-25 12:09:25+00:00", "updated_at": "2026-10-25 12:13:23.326357+00:00", "lang": "en", "topics": ["ai-agents", "ai-products", "generative-ai", "artificial-intelligence"], "entities": ["Spa Grand Prix", "Napster", "Jack", "Belgian Grand Prix", "Patrick Conway", "Spa-Francorchamps"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/the-fans-who-chatted-for-ten-minutes-to-pick-a-seat-at-spa", "markdown": "https://wpnews.pro/news/the-fans-who-chatted-for-ten-minutes-to-pick-a-seat-at-spa.md", "text": "https://wpnews.pro/news/the-fans-who-chatted-for-ten-minutes-to-pick-a-seat-at-spa.txt", "jsonld": "https://wpnews.pro/news/the-fans-who-chatted-for-ten-minutes-to-pick-a-seat-at-spa.jsonld"}}