The Job-Search Problem a Buddy Can Change
Overstapbuddy.nl, an online platform studied in a Dutch randomized experiment, is the subject of this series. It connected unemployed job seekers who were considering a move away from occupations with poor prospects to former job seekers who had recently made that move themselves. The question is not whether encouragement feels good. It is whether a credible person who has crossed an occupational boundary can change the set of options a job seeker can see, evaluate, and act on. That distinction gives readers a practical way to judge both the promise and the limits of buddy-based career support.
The difficult part is not only finding a vacancy
Searching for work is often described as a volume problem: find openings, send applications, repeat. The study behind Overstapbuddy.nl starts from a more demanding description. Job search requires people to identify suitable opportunities, prepare a résumé and application, navigate interviews, and learn strategies that most people use only occasionally. Because unemployment is a temporary state, a person may have little chance to build a durable stock of experience about how to search effectively before the next search begins.
An occupational transition adds another layer. A job seeker is not simply asking, “Where is the next opening in the work I already know?” They may be asking whether their skills transfer, what the new work is actually like, whether an employer will recognize their experience, and whether the change fits their identity and life. A vacancy list can show that a role exists. It cannot necessarily make an unfamiliar role legible.
The Dutch experiment focused on people registered with the public employment service who were primarily searching in one of 31 occupations with poor job-finding prospects. To participate, they needed an online résumé, sufficient written and spoken Dutch, and at least 90 days of unemployment-benefit eligibility remaining. The buddies came from a different position: they had recently found employment after switching occupations. The intended relationship therefore crossed a meaningful boundary. One person was trying to leave a constrained search; the other had already navigated a comparable transition.
This matters because the problem need not be a lack of motivation. Before the intervention, the job seekers reported substantial motivation to search in new occupations, confidence in performing unfamiliar work, and a belief that their skills were transferable. The experiment did not begin with the assumption that people had to be persuaded to try. It asked whether they could benefit from a more useful social connection while they were trying.
Unemployed workers may have fewer connections to employed people, precisely when a transition makes such connections valuable. Someone who has already changed occupations can offer information about the practical path, but also something harder to obtain from a formal description: a grounded example that the path can be navigated. That example is not a guarantee. A successful switcher may have different skills, timing, geography, or personal circumstances. But they can make a possibility more concrete than an abstract instruction to “consider adjacent roles.”
Overstapbuddy.nl made a person part of the search
Overstapbuddy.nl was designed around contact, not just content. Job seekers and buddies created profiles with basic personal and occupational information and a short description of their labor-market experience. A job seeker could see up to seven possible buddies and send a request to as many as three. A buddy could accept up to five job seekers. After a match, the pair could use a dedicated chat, schedule a meeting, or exchange phone details and continue in another way.
The limits were practical as well as experimental. Showing a manageable number of profiles reduced congestion and gave the researchers a way to study whether occupational distance mattered, although the sample was too small to use that design feature decisively. Limiting requests and matches also acknowledged that a buddy’s time was scarce. A system that promises unlimited access to volunteers can create the appearance of abundance while producing very little actual contact.
The platform included a professionally edited video in which nine people described their own occupational transitions. That gave every invited job seeker at least some exposure to role models, even when a personal match did not occur. These accounts make transitions imaginable, but they should be read as examples of possibility, not as universal recipes. The personal exchange was the distinctive part of the intervention: it could respond to a person’s questions rather than merely display a story.
The researchers deliberately gave pairs general guidance but did not monitor or intervene in their conversations. This choice protected privacy and allowed exchanges to develop without turning a buddy into an extension of an employment-service compliance system. It also created a measurement limit. The study recorded activity and later asked matched job seekers what they discussed, but it did not retain the content of their messages. We therefore cannot identify which particular piece of advice, reassurance, or information made a difference.
A successful switcher may also occupy a different social position from a caseworker or an unemployed peer. Caseworkers may perform some of the same search-support tasks, but the researchers suggest that infrequent contact or a supervisory role can make some job seekers less willing to reveal uncertainty. A group of current job seekers may offer solidarity, but its members have not necessarily completed the transition and may compete for similar opportunities. These contrasts explain the design rationale; they do not show that buddies are better than caseworkers in a head-to-head trial. The platform was tested as an access offer against a control condition, not as a replacement for professional guidance.
The same caution applies to the idea that buddies “expand options.” The paper proposes several routes: a buddy may share information about suitable work, explain search strategies, reduce the psychological cost of considering a new occupation, enlarge a relevant network, or reinforce a norm of continuing to search. Those are plausible mechanisms, and they fit the relationship the platform was built to create. They are not separately identified effects of individual platform features.
The diagram’s central distinction is important. A buddy does not need to increase the number of applications to be useful. The relationship could instead prevent a search from narrowing too quickly around the first familiar occupation, help a job seeker judge an unfamiliar path, or make a transition strategy more actionable. Whether that happened is an empirical question.
The question the experiment could answer
Eligible job seekers were first invited to express interest. Across two waves, 6,048 people received the first information message and 4,972 received the second. Of those who expressed interest, half were randomly invited to create an account and the other half were placed in the control or waitlist group. In total, 713 job seekers took part in the experiment. The main analysis focused on 524 people who had worked zero hours in the month before the invitation: 264 in treatment and 260 in control.
That design gives the result its credibility, but it also defines the result precisely. The experiment mainly estimates the effect of being offered access to Overstapbuddy.nl. It does not estimate the effect of a completed, sustained buddy relationship for every participant. Among treated job seekers, 53% created an account and 19% found a buddy; 49 matches formed overall. Some people who received the offer never used it, while those who did form a match chose to do so. The causal comparison is therefore between an invitation and the control condition, not between “buddy users” and “non-users.”
Even with that modest level of take-up, the labor-market pattern was meaningful. Thirteen to 18 months after access was offered, treated job seekers were about 6 percentage points more likely to be employed and earned about €226 more per month than the control group. Over the preregistered 18-month period, they earned €2,185 more. The effect was not an immediate jump visible in the first few weeks. It emerged later, which is consistent with a transition process in which a person first considers options, gathers information, changes search direction, and only then reaches an employment outcome.
The average result also concealed an important difference. The strongest effects were among people who had been unemployed for more than 17 weeks at the time of the intervention. For this group, employment rose by roughly 9 percentage points initially and remained higher later; cumulative earnings after 18 months exceeded the control group by more than €5,000. For people unemployed for a shorter period, estimated effects were close to zero and not statistically significant. The study does not prove that unemployment duration itself causes responsiveness, but it suggests that a relationship preserving options may matter most after a search has had time to narrow.
The evidence is promising without being complete. The researchers did not find significant increases in application counts, motivation, or measured search intensity. They did find suggestive evidence that the control group reduced the number and breadth of occupations it considered while the treatment group maintained a broader outlook. Among people who found work and answered an outflow survey, treated workers also appeared more likely to move into occupations different from their initial selection occupation, although those estimates were imprecise. These patterns fit the option-expansion interpretation, but they do not prove that preserving search breadth was the mechanism that produced the earnings effect.
The sensible conclusion is consequently narrower than “a mentor gets people jobs.” Overstapbuddy.nl tested whether public employment services could offer a targeted social connection to people facing a difficult occupational transition. In this Dutch setting, access to that offer improved later employment and earnings, especially for the longer-term unemployed. The study leaves open which conversations mattered, which buddy characteristics are most useful, whether the relationship continued off-platform, and whether the result travels to other countries, age groups, occupations, or digital-access conditions.
The central question for the rest of this series is therefore not whether human contact is generally good. It is more specific: what can a person who has already made an occupational switch provide that a vacancy list, a formal service, or another job seeker cannot—and under what conditions does that contribution become a durable change in the search rather than an encouraging but isolated story?