How Welten Connects Intake, Data, and AI

The Challenge

Volume has never been an issue for Welten, and it remains strong today. Processing around 600 to 800 applications per month, there is no shortage of intake. Where friction did arise—partly due to these high numbers—was in the quality and speed of follow-up, the compliance requirements of a heavily regulated sector, and ensuring consistent interview quality during periods of internal team turnover. This last point becomes key later in this case study, as it is precisely why Welten is currently experimenting with an AI-driven digital intake.

From WordPress to HubSpot

Four years ago, Welten’s website and surrounding ecosystem ran on WordPress. That worked, up to a point, but in practice felt like a collection of isolated islands. For example, any website modification required a developer; email, WhatsApp, and dashboards ran through standalone tools without integration; and candidate website behavior was visible only if you were well-versed in GA4. The biggest gap went deeper: data from Byner did not flow back to marketing, meaning marketing and recruitment effectively worked with two different versions of reality.

Together with Loyall, Welten made the step to HubSpot, with Byner alongside it as the core platform. The difference was immediately noticeable:

• Make modifications, test, and optimize independently without needing a developer
• A single platform for ads, website, email, CRM, automations, dashboarding, and Byner data
• Full visibility into candidate click behavior
• Real-time flow of Byner data back into HubSpot

The division of roles that emerged is now sharp: HubSpot handles intake, while Byner serves as the core platform. Everything that happens before someone applies runs through HubSpot; as soon as a candidate applies, Byner takes over—from application management to interviews, job offers, and final placements. Both systems communicate continuously with each other. Where these were two separate worlds before the migration, they have now become one continuous workflow.

What a Year of Implementation Teaches You

Wander (Loyall) summarized the key lessons from the implementation, which extend beyond Welten:

• Data migration almost always takes longer than anticipated, and data cleaning is the least glamorous part of the process, while simultaneously being the most critical.
• Lead scoring is never truly finished: it is an ongoing refinement process, not a one-time setup.
• A small team with the right infrastructure can easily outperform a large team equipped with the wrong tools—a reality Welten now sees reflected in its numbers.
• The most important lesson: without a solid data foundation, AI has little to build on, so that basis must be in order before the layer above can deliver value.

The Numbers That Prove the Difference

The results highlight what a clean data foundation delivers in practice:

• Of all applicants, 97% voluntarily fill out a second, optional form detailing specific client requirements—such as zip code, willingness to travel, education level, and professional certifications (WFT)—on top of the minimal initial form asking only for name, email, phone number, and CV. Because both forms connect seamlessly within the same system, it feels like a single unified process to the candidate, with only 3% dropping off after step one.
• In terms of email performance, across over 5,600 emails sent this year, Welten achieved an open rate of 54.29% (well above the HubSpot sector benchmark of 42.35%), a click-through rate of 8.23%, and a delivery rate of 98.97%.
• Additionally, chatbots now handle over 100 distinct candidate questions that previously went straight to recruiters, freeing them from routine queries to focus on interviews.

The Digital Intake

The core of the AI application lies in the digital intake, allowing candidates to choose when to complete their initial interview rather than waiting for recruiter availability. Speed is the main advantage here, alongside consistency: an AI-guided intake ensures interview quality no longer depends on who happens to be working at a given moment.

Welten CEO Thijs Vermaat added an observation extending beyond process optimization: the common assumption is that people prefer speaking to humans, but in practice—especially among younger target groups—the opposite often holds true. A conversation with an avatar feels low-barrier rather than impersonal, precisely because younger candidates already interact daily with AI as a coach or sparring partner and feel less immediately judged by a human.

Key Takeaways

Wander concluded the session with three observations summarizing the core thread of the journey:

1. Don’t overestimate what you can accomplish in a week, but don’t underestimate what you can achieve in a year.
2. Make onboarding as easy as possible for candidates, because qualification will follow naturally.
3. Build a scalable process, even if your team is currently small.

For Welten, it comes down to a clear order of operations: establish clean data first, then layer AI on top. That AI layer is not there to replace human conversation, but to ensure every candidate receives consistent attention at any time of day. People make the decisions; technology enhances the connection, not the other way around.