Scaling Operations at Wolt

Feb 2021 — Jan 2022 · Operations Associate · Berlin (nationwide)

Scaling Operations at Wolt

Operations Associate · Service operations & gear operations

Berlin office · Wolt Germany, 6 markets

~50%↓Support demand

30%↓Equipment spend

2,000+Courier partners

Two operational systems built with the same method — service demand structuring and gear circularity — during Wolt's Berlin growth phase. From the courier's side they were different problems. From the operations side they were the same one: unstructured demand entering a single physical channel, with no categorization and no clear ownership behind it.

Service designDemand routingReverse logisticsCircular economy

01 · The office as system

One role, two operational surfaces.

During Wolt's Berlin scaling phase, a significant part of the courier workforce was managed directly by the company. The relationship extended far beyond delivery operations — couriers needed support throughout their lifecycle: onboarding, documentation, contracts, gear, payroll, vacation, HR-related issues.

Much of this happened at the Wolt office. My role — Operations Associate — sat at the intersection of two operational surfaces simultaneously.

Service operations. Handling courier support demand as it arrived at the office.
Gear operations. Managing equipment distribution, recovery and refurbishment for the courier base.

Both surfaces were connected. Every courier interaction was also a potential gear moment. Every gear distribution was also an information exchange. But they were treated as separate workflows without shared operational structure.

Starting volume

At the initial volume — approximately 5 couriers per hour, across 3 agents, 7 hours per day, 5 days per week — the office was processing around 525 courier interactions weekly.

~5 couriers / hour
3 agents · 7 h / day
5 days / week
~525 interactions / week

Enough volume to strain capacity. Not enough structure to plan for it.

02 · Making the invisible visible

Categorizing 525 interactions a week.

The first problem was measurement. Different types of requests entered through the same physical channel regardless of whether they required in-person interaction. Scheduled onboarding could happen alongside urgent payroll escalations. Gear distribution alongside general HR questions.

Cardboard boxes and stacked courier jackets and thermal bags piled against the windows of the Wolt Berlin office, without shelving or intake system. — Fig. 1 — Berlin office, before structured demand routing: inbound material without categorization channels.

We started systematically categorizing courier visits. Rather than treating each interaction as an isolated support case, we recorded the purpose behind the visit.

The visit taxonomy

Payroll HR / employment questions Vacation Contracts Missing documentation Onboarding Gear — distribution, returns, damage Urgent escalations Other operational requests

Weekly monitoring consolidated into monthly reporting. That gave Operations something previously lacking: visibility into why couriers were using the physical support channel — and what percentage was actually resolvable in-person vs routable elsewhere.

The same categorization discipline that surfaced support patterns also surfaced gear patterns: which equipment was returned in what condition, which items reached end-of-life fastest, which distribution moments coincided with high-volume support hours. Both dimensions became visible through the same operational data.

03 · Two problems, one method

Data → categorization → routing → redesign.

Once demand was visible, both operational surfaces revealed the same underlying pattern: unstructured demand entering through one physical channel, symptoms of upstream problems, and no clear ownership boundaries between what Operations owned vs what other teams owned.

The methodology that worked for one applied to the other.

Data collection at every touchpoint.
Categorization of recurring patterns.
Attribution of ownership per category.
Routing to the appropriate resolution channel.
System redesign to reduce dependency on ad-hoc walk-ins.

The methodology loop

Five steps that turn a queue of individual cases into a system that shrinks its own input.

The loop closes on itself: the redesign changes what the next measurement sees, so the baseline moves and the categories re-rank. Applied to service it produced appointments and routing; applied to gear it produced recovery and refurbishment. Same sequence, different output.

Fig. 2 — The loop is the method, not the org chart: the same five steps produced the support system and the gear system. Drawn for this case study.

This became the operating principle across both service and gear:

Support the courier where support adds value. Own the process where ownership belongs to Operations. Route everything else to the team that actually owns the underlying issue.

04 · The support system

From walk-in chaos to service architecture.

For service operations, the redesign moved courier support from reactive to structured — not by adding capacity at the counter, but by reducing what had to reach the counter at all.

The Wolt Berlin office in daily operation: desks in use, stacked thermal boxes, packed bags and jackets on rails along the window wall. — Fig. 3 — Berlin office as service surface: onboarding, gear distribution and support demand handled in the same physical space.

Structured appointment management with categorized appointment types.
Self-service materials: FAQs, in-app guidance and chatbot flows for recurring questions.
Clearer channel routing: which issues went to Support, which to HR, which to Payroll.
In-person support reserved for cases that genuinely required it.
Weekly reporting escalating recurring bottlenecks to the teams that owned them.

Supporting is not owning

One of the sharpest distinctions this project surfaced was between supporting and owning.

A courier waiting several days for a payroll issue could arrive at the office frustrated. Because they were physically present, Operations naturally helped escalate. But resolving payroll was not an Operations responsibility. Our reporting made that distinction visible — helping route accountability to the teams that actually owned each recurring category, instead of absorbing everything at the office.

The new support journey

A problem no longer starts at the office door. The office became the exception, not the entry point.

Self-service where possible. FAQs, in-app guidance and chatbot flows for the questions that repeat.
Correct digital channel. Routed by category, not by whoever answered first.
Categorized appointment when necessary. Scheduled, with the right person and the right documents ready.
In-person for exceptional cases. Reserved for what genuinely needs a person and a counter.
Escalation to the responsible function. For the underlying issue behind the recurring category.

Two responsibilities anchored my ownership in this system: I independently took over onboarding and offboarding of couriers in the Berlin office, and I coordinated a small team of Gear Associates — as documented in the Wolt reference letter.

05 · The gear system

From consumable to circular.

For gear operations, the same measurement discipline surfaced a different problem: equipment was single-use by default. Every new hire received a full gear kit — thermal bags, weatherproof jackets, thermal boxes — imported from overseas manufacturers and distributed across Germany. Departing couriers kept theirs. The cost curve tracked hires and turnover, not active courier count.

Distribution station at the Wolt Berlin office: jackets on rolling rails, folded shirts sorted by size in open cases, and pallets of boxed stock. — Fig. 4 — Berlin office after implementation: distribution station with structured gear allocation by size and role.

Rows of Wolt courier jackets hanging on rails at the Berlin office, sorted and ready for re-issue. — Fig. 5 — Graded and cleaned jackets held on rails as the re-issue pool.

Three systemic gaps in the gear lifecycle

No structured recovery at offboarding. Departing couriers walked away with their gear. No return channel, no accountability, no incentive to bring it back.
No financial mechanism to enforce return. Missing gear generated no visible cost signal — neither to the courier nor to the company.
No downstream path for damaged gear. Even when items came back, worn or irreparable pieces had no recycling channel. They went to landfill.

The redesign moved the gear lifecycle from linear to circular.

Structured gear recovery at offboarding, integrated into the offboarding checklist.
Financial accountability. Missing items charged to the courier at offboarding — a behaviour-change mechanism, not a revenue generator.
Warehouse refurbishment at the Berlin office. Intake workflow, condition assessment, cleaning and minor repair, and a re-issue pool for new hires.
Downstream recycling. Irreparable gear routed to Wolt HQ Denmark via the existing partnership with TARPREC, which processes bags and heating plates in Tørring for material recovery: bags and equipment are shipped and dismantled, and up to 96% of materials are diverted from landfill and repurposed into products such as road signs and stuffing, as documented in Wolt's global sustainability communications.

The loop, drawn

Four stages that keep gear inside the system, and one downstream route for what cannot be saved.

Procurement stops being the engine of the system and becomes its residual input: new units are bought only for the demand the returned pool cannot cover. Everything the lime path touches is equipment that was already paid for once.

Fig. 6 — The lime path is the loop the program created; the dashed input is what it displaced. Drawn for this case study.

Carbon footprint quantification

The program's environmental effect was calculated rather than asserted, so it could be published externally and audited against the material flows it described. That calculation is what turned an internal cost programme into a reportable sustainability contribution, feeding into Wolt's broader sustainability reporting.

Gear that leaves with a courier is not a logistics problem. It is a design decision nobody made on purpose.

06 · What moved

Two outcomes, one methodology, honest attribution.

Service demand ~50%↓ Weekly courier support demand approximately halved during the period (from ~525 interactions/week baseline to ~262/week at close)

Procurement cost 30%↓ Equipment procurement spend, first 12 months at Wolt (Feb 2021 – Jan 2022) baseline vs. next 12 months post-implementation (Feb 2022 – Jan 2023)

Scale 2,000+ Couriers covered under both the support and gear systems across 6 German cities: Berlin, Munich, Hamburg, Frankfurt, Düsseldorf, Stuttgart

Attribution matters more than headline numbers.

The 50% reduction in weekly support demand was not caused exclusively by the service redesign. Wolt's introduction of external fleet owners during the period reduced the pool of directly-managed couriers, which mechanically reduced walk-in volume. The service design changes contributed to the shift — but so did operating model decisions I did not own.

The 30% procurement cost reduction is more directly attributable to the gear redesign, comparing baseline vs post-implementation 12-month periods.

Both outcomes are validated in the Wolt reference letter. The gear program's environmental contribution — quantified for external publication — fed into Wolt's broader sustainability reporting, contributing to the 25,000+ kg of material diverted from landfill via TARPREC across Wolt's European operations (source: Wolt Blog, October 2023).

07 · What I learned

One method, two systems.

The strongest lesson from this role was structural: two operational surfaces that looked completely different from the courier's perspective — needing support for a payroll issue vs returning a damaged bag — were the same problem from the operations perspective.

Both were symptoms of demand entering the physical channel without categorization. Both benefited from the same discipline.

Measure before optimizing.
Categorize before routing.
Attribute ownership before designing intervention.
Design for exceptions, not for defaults.

That is not customer support methodology or supply chain methodology. It is operations methodology. Applied to service, it reduces walk-in dependency and clarifies cross-functional accountability. Applied to gear, it turns consumables into circular systems and decouples cost from turnover.

Fifty people asking the same question

Another lesson worth naming: repeated support requests are often not primarily a Customer Support problem. They can be evidence of a process, information or ownership problem upstream. If fifty people ask the same question, answering it fifty times may be necessary — but it does not solve the underlying issue. The more scalable intervention is to understand why they are asking, where the information should exist, and which team owns the underlying problem.

Not case-by-case handling. System-level redesign around recurring user needs.

External validation

Reference letter and published sources

Both outcomes are documented in the Wolt reference letter, available via the Certificates section of this portfolio. Selected accomplishments confirmed by the letter:

“He led the project of cleaning the couriers' work clothes to reduce Wolt's costs by more than 30%.”

“He took responsibility for a nationwide recycling project to reduce Wolt's carbon footprint.”

“He independently took over the onboarding and offboarding of couriers in our local office in Berlin and coordinated a small team of Gear Associates.”

“He demonstrated his strong analytical skills in supporting the team leader through data-driven analysis and the preparation of weekly reports.”

External validation of the gear program's environmental impact appears in Wolt's global sustainability communications: 25,000+ kg of material diverted from landfill via TARPREC across Wolt's European operations (Wolt Blog, October 2023).

Certificates & reference letters → Wolt sustainability report ↗

The team

Jorge Aguilar with the gear team at the Wolt Berlin office, surrounded by thermal boxes and packing cartons at the end of a shift. — Fig. 7 — The Berlin gear team. A system that runs daily is a team habit before it is a process.

On the figures and photographs

This case study describes work carried out at Wolt Germany as Operations Associate in Berlin (Feb 2021 — Jan 2022). It is written from my own scope of work, and it is deliberate about which numbers come from where.

Photographs are mine, from the Berlin office. Faces and personal details are redacted in every frame.
The diagrams were drawn for this case study. They are models of the method and the lifecycle, not reproductions of internal documents.
The support-demand figure is approximate and shared. The ~525 → ~262 weekly interactions range is an operational estimate from the period's reporting, and the reduction is attributable to the service redesign and to Wolt's shift toward external fleet owners — a change I did not own.
The procurement comparison extends past the role. The 30% figure compares Feb 2021 – Jan 2022 against the post-implementation window Feb 2022 – Jan 2023.
Downstream recycling figures are public. The TARPREC partnership, the 96% material-diversion rate and the 25,000+ kg recovery figure come from Wolt's published communications, and describe Wolt's European operations — not this program alone.
No internal financials. Absolute procurement values, unit costs and commercial terms are left out; the 30% reduction is stated as a relative change only.