Of all incoming conversations are fully resolved by HALO without colleague intervention
Customer Satisfaction Score, with HALO conversations scoring higher than conversations handled by employees
Increase in WhatsApp volume since the chatbot went live, without the team growing
Billink is a Dutch buy-now-pay-later solution for online retailers. Customers can place an order without paying upfront; Billink performs a credit check, pays the retailer directly, and takes on the full credit risk.
Behind the scenes, Billink's customer service team handles a significant share of the more than 100,000 customer questions that arise each year, with the personal attention the brand is known for.
As Billink grew, so did the number of repetitive customer questions. More retailers, more orders, and consequently more questions such as “How do I pay?”, “Can I get an extension?” and "What's the status of my return?" put increasing pressure on the customer service team.
The customer service platform Billink was using already included a chatbot, but it was entirely rule-based. Every exception had to be built as a separate manual flow, making it difficult to scale with the growing complexity of customer questions.
Since February 2026, Billink has been using Agent Inbox and HALO to take repetitive questions off the team's plate, freeing up time for the personal attention Billink is known for. Today, around 71% of all incoming conversations are handled fully automatically, with higher customer satisfaction and no need for a bigger team.
For Billink, customer contact is largely predictable. Most questions fall into two broad categories: payment questions and general order questions.
The customer service platform Billink used brought all channels together in one environment and included its own chatbot module. However, the chatbot was entirely rule-based. Every exception required a manually built flow, which meant the system could not keep up with the growing complexity of customer questions.
This created a clear limit for the customer service team. Without a different approach, growth in customer volume would automatically lead to more repetitive questions, and therefore more pressure on the team.
At the same time, automation could not come at the expense of quality. The level of service had to remain at least the same, and preferably improve.
You don’t want the number of repetitive questions to simply double when you scale your customer base. That’s not what scaling is about, and the same applies to every department. Bart den Outer, Head of Operations, Billink
One of the challenges was to automate as many tickets as possible while maintaining customer satisfaction. Customers should experience the same level of service as before, preferably better, but it should never get worse. Bart den Outer, Head of Operations, Billink
Billink chose Agent Inbox and HALO to replace its old, rule-based system. The difference is not simply that conversations are handled faster. HALO can make decisions within Billink’s business rules, rather than following a predefined flow.
Platform: Agent Inbox and HALO, in use since February 5, 2026
Channels: WhatsApp, live chat, and email
Key automation capability: HALO independently performs actions within predefined business rules, such as pausing orders and creating disputes
Key business benefit: the same team can handle a growing volume of customer conversations without increasing staffing levels
Build It Yourself with Ask HALO
What makes the transition particularly powerful for Billink is how easy it has become to build and refine AI agents. Bart den Outer, Head of Operations, has no technical background. Initially, he received some support from a CM.com specialist to set up the first agents. During the testing period, he quickly started doing it himself by working with Ask HALO, the AI assistant within HALO Studio that builds, tests, and adjusts agents based on instructions in natural language.
I can just do it myself now. I don’t really need anyone else to do it anymore. Bart den Outer, Head of Operations, Billink
When Bart wants to refine a rule or adjust an agent, he simply explains what he wants to Ask HALO as if he were explaining it to a colleague. No developer is needed, and there is no waiting for a release.
More about how Ask HALO makes this possible: Ask HALO: the agent that builds, improves, and maintains your AI agents
Most of Billink’s customer conversations revolve around two topics: payments and returns. Both require the same kinds of actions, again and again: pausing an order, creating a dispute, or explaining the applicable rules. Previously, employees handled each of these steps manually.
Result: these types of questions are now largely handled by HALO without employee intervention.
Answers questions about payment deadlines and extensions
Pauses orders based on business rules
Creates disputes for returned orders
Automatically escalates to an employee when it is unsure
Customers often communicate using screenshots instead of text: a screenshot of their payment app or a confirmation of their order. Because Billink is connected to numerous e-commerce platforms, each with its own order numbering system, identifying the correct order can normally be a time-consuming puzzle for an employee.
Result: HALO can recognize and identify order numbers directly from uploaded images, even when different e-commerce systems are involved.
Reads order numbers from uploaded screenshots
Recognizes which e-commerce system a reference belongs to
Asks targeted follow-up questions when a number cannot be identified immediately
Links the correct order to the conversation
After Billink introduced a chatbot on its website and customers experienced how effectively their questions could be resolved, customer behavior changed too. The number of WhatsApp conversations doubled, while phone calls and web form tickets decreased.
Once customers have been successfully helped through a channel, they tend to return to that same channel. This resulted in more conversations for Billink, but also in more of the type of contact that can be handled particularly well through automation.
A customer called back, very happy about the conversation she had “late the previous evening” with a colleague, without realizing that the entire conversation had actually been handled by HALO. Bart den Outer, Head of Operations, Billink
The results after the first months with HALO:
Of all incoming conversations handled automatically
Additional FTE needed despite WhatsApp volume doubling
CSAT score, higher than before working with CM.com
Increase in WhatsApp volume, without the team growing
For Billink, the key impact is not about needing fewer people. It is about allowing the team to keep up with business growth without staffing levels having to increase at the same rate.
When the team is temporarily understaffed due to illness or holidays, customer service may fall behind for a few days, but it no longer drops below the required quality level, something that could happen before HALO was introduced.
With the foundation in place, Billink is already looking at the next phase:
Allowing HALO to independently send emails on topics where confidence in the AI’s answers is high enough, instead of only providing suggested replies for an employee to review
Exploring how HALO could support phone agents with real-time information during a call, the phone conversation itself stays personal, handled by a person
Actively encouraging customers to interact with HALO themselves, further increasing self-service across the platform
Ultimately, when we know for sure that HALO’s answer is correct on a particular topic, I want it to be able to send those emails itself. Bart den Outer, Head of Operations, Billink
"You have to encourage people to just use it. Once you see the results, that naturally motivates you to keep going. I also tell people internally: just send a message to Billink and see what happens."
Bart den Outer, Head of Operations, Billink
Select a region to show relevant information. This may change the language.
