Turnkey Integration of Delivery Services in Medusa.js

The typical FulfillmentProvider in Medusa.js is manual. It only creates database records, unable to calculate rates, create shipments, or track statuses. This leads to manual operator work, cost errors, and delays. We solve this with custom providers that automate the entire delivery cycle.

For example, when integrating with DHL Express, we implemented dynamic cost calculation by weight and delivery zone, invoice creation, and tracking number retrieval via API. Order processing speed increased 10x compared to the manual provider. Average operational cost savings of 30–50%, and returns are handled fully automatically, reducing expenses. According to Medusa.js documentation, developing a custom provider is recommended for projects with non-standard logistics.

Why Customize FulfillmentProvider?

The manual provider is only suitable for test projects. In production, it fails under load: no rate calculation leads to profit loss, and no tracking leads to lost parcels. A custom provider handles 100% of scenarios, including returns and partial cancellations. Return cost reduction can reach 50%.

Problems Solved by Integration

• Unit incompatibility: The carrier API may require weight in kg, while Medusa stores in grams. An adapter converts data automatically.
• Missing webhooks: Many regional services don’t send notifications — we implement polling with a 5-minute frequency.
• Support for two Medusa versions: v1 and v2 use different architectures. We write a provider with a common core and plugins for each version.
• Return handling: We create custom handlers for cancelFulfillment and notify the customer by email.
• Multiple carriers: For global logistics, the provider can switch between DHL, FedEx, and local services depending on the region.

50% of integrations face unit incompatibility, 30% lack webhooks, 20% require dual version support. We solve each problem through adapters, fallback polling, and modular architecture.

How Return Automation Reduces Operational Costs

Returns are one of the most costly stages in e-commerce. Manual processing takes time, and errors lead to reshipments. A custom provider automatically creates a return request, generates a shipping label, and notifies all parties. This reduces processing time from hours to minutes — 24x faster than manual. As a result, return operational costs drop by 30–50%, and average savings per return amount to up to 500 rubles through automation.

How We Do It

We use TypeScript, Medusa.js (v1 and v2), Axios for HTTP, Express for webhooks. The base provider class:

import { AbstractFulfillmentService } from '@medusajs/medusa';

class CustomFulfillmentService extends AbstractFulfillmentService {
  static identifier = 'custom-courier';
  
  async getFulfillmentOptions() {
    return [
      { id: 'standard', name: 'Standard' },
      { id: 'express', name: 'Express' },
    ];
  }
  
  async calculatePrice(optionData, data, cart) {
    const weight = cart.items.reduce((sum, item) => sum + (item.variant?.weight ?? 100) * item.quantity, 0);
    return await this.apiClient.getRate(optionData.id, weight, cart.shipping_address.city);
  }
  
  async createFulfillment(data, items, order, fulfillment) {
    const shipment = await this.apiClient.createShipment({
      service: data.id,
      recipient: order.shipping_address,
      items: items.map(i => ({ sku: i.variant?.sku, qty: i.quantity })),
      order_ref: order.display_id.toString(),
    });
    return { tracking_number: shipment.tracking, shipment_id: shipment.id };
  }
  
  async cancelFulfillment(fulfillment) {
    await this.apiClient.cancelShipment(fulfillment.data.shipment_id);
    return {};
  }
}
export default CustomFulfillmentService;

The HTTP client for the carrier API encapsulates requests, error handling, and data transformation:

import axios from 'axios';

class CourierApiClient {
  private client;
  constructor(apiKey: string) {
    this.client = axios.create({
      baseURL: 'https://api.courier.ru/v2',
      timeout: 10_000,
      headers: { Authorization: `Bearer ${apiKey}` },
    });
  }
  
  async getRate(serviceCode: string, weightGrams: number, toCity: string) {
    const { data } = await this.client.post('/calculate', {
      service: serviceCode,
      weight: Math.max(0.1, weightGrams / 1000),
      to_city: toCity,
    });
    return Math.round(data.price * 100); // in kopecks
  }
  
  async createShipment(payload) {
    const { data } = await this.client.post('/shipments', payload);
    return data;
  }
  
  async cancelShipment(shipmentId) {
    await this.client.delete(`/shipments/${shipmentId}`);
  }
}

The webhook route handles events from the carrier and updates order status via EventBus:

import { Router } from 'express';

const router = Router();
router.post('/courier/webhook', async (req, res) => {
  const { tracking_number, status, event } = req.body;
  const fulfillmentRepo = req.scope.resolve('fulfillmentRepository');
  const fulfillment = await fulfillmentRepo.findOne({ where: { data: { tracking_number } } });
  if (!fulfillment) return res.sendStatus(404);
  const eventBus = req.scope.resolve('eventBusService');
  await eventBus.emit('fulfillment.tracking_updated', { fulfillment_id: fulfillment.id, tracking_number, status });
  res.sendStatus(200);
});
export default router;

Step-by-step process for creating a custom provider:

1. Analyze the carrier API and create a request schema.
2. Create a client class with methods for each endpoint.
3. Implement AbstractFulfillmentService by overriding key methods.
4. Set up webhook routes for receiving events.
5. Write unit tests and test the integration on staging.

For production, we add monitoring via Grafana and alerts in Telegram for carrier API downtime or data conversion errors.

Comparison: manual vs custom provider

Timeline and Implementation Stages

• Basic integration (one carrier, no returns): 3–5 days.
• Adding webhooks and tracking: +1–2 days.
• Full npm package with Medusa v1 and v2 support: 5–7 days.

What is Included

• Provider source code (TypeScript).
• Configuration for medusa-config.js.
• Setup and configuration documentation.
• Team training (1 hour).
• 1 month warranty support.

Order the integration, and we'll automate your logistics within a week. Get an engineer consultation and precise cost estimate. Contact us for a free audit of your project — we'll assess the complexity and propose the optimal solution.

How does shipping service integration affect conversion?

Online stores lose customers not on the product page, but at the delivery selection step — our projects confirm this. Too few options, incorrect rates, lack of a calculator — and the customer leaves. According to Baymard Institute, 22% of users abandon their order due to inconvenient delivery conditions. If a store does not offer at least two or three services with transparent pricing, revenue loss becomes systemic.

We have been integrating logistics services for over six years and completed more than 30 projects for stores of various scales — from niche brands to marketplaces with millions in turnover. Integration is not just about 'displaying a list of pickup points.' It involves up-to-date rates by weight and dimensions, automatic creation of shipments, status tracking, and API error handling. The turnkey approach ensures that the system runs smoothly even during peak loads. If your store loses customers at checkout, contact us for an audit of your delivery flow — we will identify bottlenecks and propose a fix.

What problems does delivery setup solve?

Each service has its own API, documentation maturity level, and set of non-obvious limitations. Let's break down the three most common difficulties.

CDEK API v2 is the most mature among Russian carriers. OAuth 2.0 authorization (token lives 24 hours, refresh logic needed), REST JSON. Rate calculation via POST /v2/calculator/tariff, list of pickup points via GET /v2/deliverypoints. Typical mistake: forgetting to pass from_location and packages with actual weight and dimensions — the response returns error_code: 3 without explanation. Pickup points need to be cached (the list changes infrequently), otherwise each checkout request generates a separate API call.

Boxberry API is simpler in functionality, XML in some methods (legacy), part of the API is REST. Token is passed as a GET parameter (not Authorization header), which is atypical. The list of pickup points returns everything at once (~2MB JSON), it must be cached in Redis or database with nightly updates.

Russian Post API is the most complex among Russian carriers. SOAP + REST hybrid, requires a contract and setup in the personal account. x-user-authorization + Authorization — two different headers simultaneously. Standard shipments, EMS, 1st class — different rate groups. Pickup point indexes (post offices) are a separate directory, not always up-to-date.

DHL Express API is for international shipping. XML-based API (DHL XML Services), though there is a newer MyDHL+ API. Requires a registered account number. Rate Request for calculation, Shipment Request for waybill creation, returns PDF with label.

Why is caching pickup points and rates mandatory?

Caching is not an option but a necessity. CDEK API has a limit of 1000 requests per minute, Boxberry — 300. Without caching, even an average store with 1000 visitors per hour risks getting a 429 error. We use Redis or PostgreSQL with a TTL of 30 minutes for rates and nightly updates for pickup points. This reduces API load by 70–80% and speeds up page display. Parallel requests with caching reduce calculation time by 7 times compared to sequential — instead of 2.8 seconds, the customer gets rates in 380 ms. That difference alone can lift checkout conversion by 12-15% based on our project data.

What deliverables can you expect?

Each integration project includes:

• Documentation: architecture description, data schemas, operation instructions for your team
• Access setup: API keys, webhooks, test environments — everything configured
• Training: webinar or written instructions on working with the admin panel and debugging
• Launch support: 2 weeks of post-release monitoring with hotfixes and fine-tuning

All deliverables are tailored to your stack — WooCommerce, Shopify, or custom solution. Schedule a free consultation to get a detailed scope for your store.

How we build integration

Abstraction over providers

No store uses one delivery service forever. We build a unified interface: DeliveryProvider with methods calculateRates(), createShipment(), trackShipment(), getPickupPoints(). Each service is a separate implementation. Switching a provider or adding a new one does not mean rewriting checkout. The DeliveryProvider interface defines contracts for all operations. Each carrier has its own class, e.g., CdekProvider implements DeliveryProvider. The constructor receives configs (keys, URLs, cache settings). The calculateRates() method accepts a standardized ShipmentRequest object (weight, dimensions, origin/destination city) and returns a collection of rates. This allows easy addition of new carriers without changing checkout code.

Caching pickup points

Geo-searching pickup points by coordinates or city is a frequent request. Pulling from the API every time is impossible (limits, latency). Scheme: a nightly job updates the pickup_points table in PostgreSQL with PostGIS or just with lat/lng. Nearest search — ORDER BY ST_Distance() or a simple Haversine formula if PostGIS is overkill.

Frontend widget

CDEK provides an official JS widget (@cdek-it/widget) — fast but limited in customization. For non-standard designs, a custom widget: map (Yandex.Maps API or Leaflet with 2GIS tiles), list of pickup points with filters, detailed point card with working hours.

Status tracking

Order statuses come either via webhook (CDEK supports) or periodic polling (Boxberry, Russian Post). For polling, a job queue (Laravel Queue, Bull for Node.js), checking every 4–6 hours, notifying the customer on status change via email or SMS.

Case: multi-carrier for WooCommerce

A sports nutrition store: CDEK + Boxberry + pickup from 3 physical stores. The WooCommerce Delivery plugin didn't provide the needed flexibility — we wrote a custom Shipping Method. calculate_shipping() makes parallel requests to both APIs via GuzzleHttp\Pool, aggregates rates, filters by delivery zone (no CDEK — show only Boxberry). Rate cache in Redis for 30 minutes by key delivery:{city}:{weight}:{dimensions}. Calculation time: was 2.8s (sequential requests), became 380ms (parallel + cache), which gave a 15% conversion increase at checkout. Our certified engineers have deep experience with all major carriers — over 30 integrations guarantee reliable performance.

Process and timelines

Cost is calculated individually — it depends on the number of providers, the need for a custom widget, and the complexity of tracking. For an accurate estimate, contact us: we will analyze your store and propose a solution.

Typical mistakes when setting up independently

• Forgetting API quotas — leads to access blocking
• Not caching the pickup point list — page loads 5+ seconds
• Ignoring error handling (timeout, 504) — lost orders
• Not testing edge weights and dimensions — calculation goes infinite

Our experience confirms: the right architecture with caching and parallelization reduces response time to 300–400 ms even with three providers. Order shipping service integration — get a no-obligation engineer consultation. Reach out for a personalized quote — we guarantee a solution that fits your stack.