More orders should not mean proportionally more admin
In a marketplace, growth shows up immediately in the operations team’s inbox.
New cities need a courier supply plan. Existing cities need enough people on the road at the right hours. Every courier needs to be recruited, trained, contracted, equipped and ready to deliver. And none of that helps if delivery quality deteriorates as volumes rise.
During Wolt Hungary’s growth phase, I built and led a 15-person operations team covering supply planning, onboarding, courier experience, city management and delivery performance. We developed routines for managing supply and monitoring delivery quality — while looking for processes we could make less dependent on manual effort.
Courier onboarding became one of the clearest examples. The early version was entirely personal. It worked when volumes were small; it clearly wouldn't work forever.
Four versions of the same process
We didn't automate onboarding in one heroic leap. We kept redesigning it as scale exposed the next bottleneck.
Individual introductions gave us direct contact, but consumed team hours and constrained throughput.
We moved to larger groups and introduced knowledge tests at the end. The assessment was still paper-based, but the hiring decision was no longer based only on attendance or a quick conversation.
Applicants registered for a session and received the joining link automatically. After the call, they were sent a knowledge test; results were collected, scored and organised into a review queue. The operations team made the final selection decision.
We recorded the training so applicants could work through it on their own schedule, supported by the already-established digital steps. By the time COVID arrived, the backbone of remote onboarding was in place.
The practical stack combined video calls, forms, applicant-state tracking, automated messages, knowledge tests and digital contracting. By 2020, the workflow used tools including Zoom/VideoAsk, SurveyMonkey/Airtable and HelloSign.
Automate invitations, tests, scoring and administration. Keep the final hiring decision with the team.
Onboarding was only one piece
Courier onboarding is only useful when it is connected to supply planning. We needed to know when and where new couriers were required, how our existing supply performed, whether service levels were holding and how much the system cost to operate.
The team created city-launch routines, courier capacity monitoring, onboarding and equipment workflows, performance dashboards, and ownership for recurring problems. Instead of one person knowing everything, we gradually made the system legible to the whole team.
Our Hungarian operation grew to more than 50,000 deliveries a day while keeping average delivery time below 30 minutes. That scale reflects the wider business and team effort; my contribution was building and leading the local operations function that had to make it work every day.
Help the next team avoid our first mistakes
Once we had developed usable processes in Hungary, I spent time helping teams in other Wolt markets assemble and adapt similar operating practices. That included work with colleagues in Kazakhstan, Azerbaijan, Croatia and Czechia; similar approaches also became common in markets such as Poland and Greece.
It wasn't a copy-paste rollout. Country teams had different labour arrangements, local expectations and operational constraints. The useful part was transferring the underlying model — what to measure, what to standardise, what to automate and where local judgment remained necessary.
That experience later became a recurring pattern in my international strategy and operations work: a good process is not the one that looks beautiful in a central playbook. It's the one another team can actually use.
Scale, efficiency and international adoption
Hungarian marketplace operation, maintaining average delivery times below 30 minutes.
Capacity, city operations, onboarding, courier quality and delivery performance.
Final onboarding conversion in a historical internal comparison of non-interactive and interactive introduction approaches.
Adaptation of Hungarian operating processes to new local conditions.
The internal funnel comparison measured conversion across non-interactive and interactive introductions. It did not isolate the effect of recorded-video training.
Building a reliable operating function, using low-cost automation to remove administrative bottlenecks, and helping local solutions become repeatable playbooks for other markets.