Peter Nyssen has been selling bulbs to UK gardeners since 1958. In 2025, their customers left over 4,500 Feefo reviews. For most of that time, 97% of those customers – the satisfied ones – heard nothing back. Feefo's AI Replies changed that: the team now responds to 75% of reviews, reaching thousands of customers who'd previously gone unacknowledged.
Key takeaways
- Reply rate increased from approximately 50% to 75%, with the small team now responding to four- and five-star reviews as well as complaints
- An extra 2,558 customers a year – around 53 a week – now receive a reply that previously went unanswered
- AI Replies enabled Stephen to delegate the response process to his customer service team without sacrificing tone consistency or quality
- Responding to satisfied customers – not just dissatisfied ones – is now a core part of how Peter Nyssen builds loyalty with a community of repeat gardeners
The challenge
Peter Nyssen isn't a typical retailer. Bulbs are a seasonal product, and the relationship between purchase and result can span months. A customer who orders snowdrop bulbs in autumn won't see them flower until February. Someone who orders alliums might be waiting until June.
That seasonality creates two very different kinds of review. The first, a service review, arrives within weeks of an order shipping – typically focused on delivery and condition. The second – the product feedback request – comes months later, after the bulbs have flowered, when many customers have long forgotten the specifics of what they ordered or where they bought it.
Managing those two review streams, and responding meaningfully to each, took time Peter Nyssen's team didn't always have. Each response needed to be thoughtfully put together.
97% of Peter Nyssen's service reviews are four or five stars. Yet before AI Replies, almost none of those customers received a response. The team's attention was concentrated entirely on the small minority of unhappy reviewers – which meant the people most likely to come back, recommend the business, and leave another review the following season were being unintentionally ignored.
As Stephen Teeuw, who oversees marketing, explains, every difficult reply had to balance three things: making the customer feel heard, addressing any legitimate recompense, and giving them a reason to trust the business again. This meant he handled every one- to three-star review response personally. It wasn't a task he was prepared to delegate.
The solution
Before December 2025, the team responded only to reviews of three stars and below. With AI Replies providing a structured, well-calibrated starting point every time, it became practical to respond to every review – one star through to five.
The process isn't automated. Priscilla, Peter Nyssen's customer service manager, reads each review, checks the AI-generated response, and amends it where needed. For lower-star reviews, she may regenerate the response once or twice to ensure it addresses everything the customer raised – not just the most obvious point of complaint. For four- and five-star reviews, the AI response typically requires only minor adjustment before sending. Across a batch of 50 reviews, that process – reading, checking, adjusting, sending – now takes around three hours.
But for Stephen, the value was never just about doing it faster. It was about being able to do it at all, for everyone, without the responses feeling generic.
"Small changes like this allow us to meaningfully connect," he says. "It only takes a couple of sentences to extend our reach."
AI Replies gives the team a consistent starting point, but Priscilla still shapes every response – checking it reflects the customer's feedback, and that it sounds like Peter Nyssen rather than a template. The result is a process that scales without losing the personal judgement Stephen wasn't willing to hand over lightly.
"After testing, we felt confident in the abilities of the AI," Stephen explains. "Without it, we couldn't broaden our reach."
The results
Reply rate has risen from approximately 50% - to 75% of all reviews received. The shift isn't down to faster complaint handling; it reflects who the team is now responding to. The majority of the additional replies go to satisfied customers: people who left a positive review and, previously, heard nothing back.
That's a meaningful number of conversations that simply weren't happening before. An extra 2,558 customers a year – about 53 a week – now get a reply that previously wouldn't have been sent.
For a business built on repeat purchase, where the same gardener may return season after season, that closing of the loop works as a validator. A customer who buys tulip bulbs in September, gets a thoughtful response to their review in December, and feels genuinely heard is more likely to come back the following year than one who's ignored, "An email can be written in seconds, checked, and sent," Stephen notes. The speed is what makes the care possible, not what replaces it.
Peter Nyssen also surfaces customer reviews and photography on their testimonials page, making the review programme a visible part of how they present themselves to prospective buyers. The response discipline enabled by AI Replies feeds directly into that – a customer who feels valued is more likely to contribute again.
How Peter Nyssen works differently now
The most significant operational change isn't the time saved – it's the shift in what the team considers worth responding to.
Stephen's framework for managing reviews – recognition, recompense, assurance – hasn't changed, but the number of customers that applies to, has. Responding to a five-star review takes around thirty seconds and tells a loyal customer that their garden, and their feedback, genuinely matters to a business that has been talking to gardeners for over sixty years.
AI Replies made that sustainable. It moved review response from a one-person task with a high bar for quality to a shared team process with a consistent standard – and in doing so, gave Peter Nyssen a practical way to treat every reviewer as part of the community they've always believed their customers to be.
Most review programmes are set up to manage dissatisfied customers. Feefo's AI Replies is how businesses like Peter Nyssen start talking to all of them.
