A Case Study in AI Reservations for a Multi-Location Restaurant Group in Amsterdam
In early 2026, a prominent multi-location restaurant group operating seven venues across Amsterdam faced a familiar problem: fragmented reservation systems, inconsistent guest data, and a booking abandonment rate that was quietly bleeding revenue. This case study of AI reservations across their Amsterdam portfolio documents what happened when they consolidated everything under a single AI-powered reservation engine and what the numbers looked like six months later.
The group, which operates under a shared ownership structure but with distinct culinary identities ranging from a canal-side brasserie to a plant-based fine dining concept in De Pijp, had been running three different reservation platforms simultaneously. Staff at each location handled phone bookings differently. Guest preferences logged at one venue were invisible to another. The result was a disjointed experience that contradicted the group's ambition to build loyalty across its portfolio.
The Problem: Seven Venues, Zero Shared Memory
Before the consolidation, each restaurant maintained its own booking workflow. Two locations used CoverManager, three relied on a legacy system with manual confirmations, and the remaining two accepted reservations only through phone and Instagram DMs. The general manager described the situation as "running seven separate businesses that happen to share a bank account."
On average, 27% of phone-based reservation attempts went unanswered during peak hours across the group's seven Amsterdam locations in Q4 2025.
That number came from an internal audit the group commissioned before making any technology decisions. Missed calls were not just an inconvenience. Each one represented a potential cover lost to a competitor or a third-party platform collecting commission on what should have been a direct booking.
Guest Profile Fragmentation
A regular who dined at the group's Indonesian-inspired restaurant in Oud-Zuid every Friday was treated as a first-time guest when she visited their wine bar in Jordaan. Allergy notes, seating preferences, and spending history existed in silos. The group estimated that roughly 35% of their cross-venue guests were repeat visitors who received no recognition outside their "home" restaurant.
Consolidation with an AI Reservation Engine
The group deployed LlamaChilly across all seven locations in January 2026. The AI reservation engine served as a single layer handling inbound bookings from web forms, WhatsApp, Instagram DMs, phone calls via voice AI, and walk-in waitlist management. Each venue retained its own branding and booking page, but behind the interface, every interaction fed into a unified guest profile database.
Staff training took less than a week per venue. The system's natural language processing handled Dutch, English, and French out of the box, which mattered for a city where nearly half of restaurant guests are international visitors. According to a 2026 Statista report on Dutch tourism, Amsterdam welcomed over 21 million overnight stays in 2025, a figure projected to grow by 6% in 2026.
How the AI Handled Booking Conflicts
One counterintuitive finding emerged within the first month. The group expected that AI would simply reduce missed calls. Instead, the largest measurable impact came from conflict resolution across venues. When a guest's preferred time slot at one location was full, LlamaChilly automatically suggested availability at a sister restaurant with a similar cuisine profile, complete with a map link and menu preview. This cross-venue recommendation accounted for 11% of all completed bookings in February 2026, a channel that had not existed before.
Results After Six Months: The Numbers
The group recorded a 43% reduction in missed bookings between January and June 2026 compared to the same period in 2025, measured as confirmed covers divided by total inbound booking attempts.
Here is what the data revealed across other key metrics. No-show rates dropped from 9.1% to 5.4% after the system introduced predictive confirmation sequences that adapted based on guest history and booking lead time. For more on how that scoring model works, the explanation in our article on predictive no-show modeling covers the technical side. Average party size increased by 0.3 guests per booking, partly because the AI prompted group size adjustments during the booking flow when it detected large-party intent in message language.
Unified Guest Profiles in Practice
By June 2026, the system had merged over 14,000 duplicate guest records into unified profiles. Staff at any venue could see a guest's full history across the group, including dietary restrictions, preferred seating, average spend, and visit frequency. The Indonesian restaurant's Friday regular was finally greeted by name at the wine bar. That kind of recognition, previously reserved for single-venue establishments with long memories, became standard across all seven locations.
The operational savings were also measurable. The group reduced front-of-house phone handling time by an estimated 22 hours per week across all venues. That time was reallocated to floor service. For groups considering similar moves, our coverage of scaling AI across multi-location restaurant groups outlines the infrastructure decisions that matter most.
What Surprised the Team
The general manager noted that the biggest surprise was not efficiency. It was guest behavior change. Repeat visit frequency across the portfolio increased by 18% in the first half of 2026. Guests who had previously been loyal to a single venue began exploring sister restaurants, apparently encouraged by the personalized cross-venue suggestions LlamaChilly embedded in post-dining follow-up messages.
18% increase in cross-venue repeat visits within six months, driven primarily by AI-generated post-dining recommendations.
This outcome challenges the common assumption that restaurant groups should keep their brands entirely separate to avoid internal competition. The data from this Amsterdam group suggests the opposite: when a shared AI layer connects the portfolio intelligently, the venues lift each other rather than cannibalize.
Where This Points for Amsterdam's Restaurant Groups in 2027
The hospitality industry in Amsterdam is entering a phase where guest expectations are shaped by the personalization standards set by e-commerce and streaming platforms. Multi-location restaurant groups that treat each venue as a data island will increasingly struggle to compete with operators who build connected guest experiences. The technology to do this exists today, as this group's results confirm. The question for 2027 is not whether AI reservation systems will become standard across restaurant portfolios but how quickly independent groups will adopt them before the competitive gap widens further.