Retail, commerce and marketplaces: the operating job, and how a first project is scoped

The operating job in one sentence

A retail, commerce or marketplace business converts an item of supply into a completed order through records that several systems each believe they own. The work is not selling; it is keeping one version of the catalogue, the availability figure, the order state and the customer identity correct while every channel writes to it at a different moment.

Current pressures

What operators report in 2026, each a failure mode:

The workflows a project touches

A first project cuts into this sequence. Scope which steps it owns before choosing software.

1. Product data intake. Items arrive from suppliers, a bulk file or an internal team, and are normalised into one shape with required attributes, variants and media.
2. Channel publication. The normalised item is mapped to each channel's taxonomy and attribute rules, published, and monitored for rejection.
3. Availability. One figure per item per location, with reservations, in-flight returns and supplier lead times represented, not assumed.
4. Order capture. Orders from every channel are normalised into one record with a single identifier the customer can be told.
5. Fulfilment. Allocation, picking, dispatch and the delivery promise, including when it cannot be kept.
6. Returns and refunds. The reverse path, with stock, payment and supplier claim updated from one event.
7. Customer contact. Order status, returns and product questions, routed with the order record attached.
8. Reconciliation and reporting. Channel settlements, fees and adjustments matched against orders, with the difference explained, not absorbed.

Steps 1 to 3 and seller trust are detailed in marketplace and classifieds operations. Steps 3, 5 and 8 sit in ERP and back-office operations; steps 4 and 6, where payment and order state must agree, in order and payment operations. Step 7 is customer support automation. Reward shops add redemption rules: see the loyalty and rewards platform guide. Rooms, seats and tickets sold by date follow travel and hospitality; course sales ending in enrolment follow education and training.

Data and regulatory constraints

• Shopper personal data. Consent, retention and deletion apply to the customer record, the order history and the marketing identifiers attached to both. Decide the retention period per market before the first table is designed.
• Payment card data stays with the payment provider. Your application holds a token and a reference, never the card. That is a boundary decision, not an optimisation.
• Tax and invoicing differ per market. The order record must carry the rule in force at the moment of sale, because a later correction is judged against that moment.
• Channel terms restrict what you may store and export. A marketplace agreement often limits retention of buyer contact data and forbids re-use. Read it before designing a customer master.

The usual integrations are channel and marketplace APIs, a payment provider, a shipping aggregator, a search index, an accounting or ERP system, and analytics. Each needs a named owner, a sandbox and a defined behaviour when it fails.

Where does a first project start?

Netbase's delivery lifecycle: discovery and strategic alignment; team assembly and architecture planning; agile execution with outcome-based milestones; modular components; training and rollout; ongoing support. For a multichannel seller, discovery ends with one owner named per record, and rollout avoids the selling season entirely.

1. Count the real volume. Items, variants, channels, orders per peak day, returns and contacts per week; estimates are marked as estimates.
2. Name the owner of each record. Catalogue, availability, order, customer. Where two systems own one record today, that conflict is the project.
3. Pick one channel and one category for a first milestone, with written acceptance criteria.
4. Build the normalised record before the integrations. A connector built onto an undefined catalogue has to be rebuilt.
5. Run the reverse path in the pilot. A pilot that tests only the forward order path proves the easy half.
6. Widen by channel, then by category, against the measures baselined in step one.

Most projects are agreed as fixed-scope contracts after discovery. Milestone acceptance is on the project delivery page and evidence labelling on the methodology page. The marketplace matching and quotation guide frames the buyer decision, the commercial route is custom product engineering, and other shapes are under services.

Which published work sits closest to retail

Netbase has served more than 500 clients in the United States, Europe and Asia-Pacific, most from outside Vietnam; those are service markets, not offices, and no project or country counter is published. It began in e-commerce and added consulting and outsourcing in 2016. That is company origin, not a current sector claim.

One anonymised record here is published in full. For a founder who is not named, Netbase delivered a bilingual classifieds platform: a Laravel backend with REST APIs, one-time-password accounts, paid ads, search, messaging, event ticket ads, payments and multi-language SEO on AWS, in six milestones over four months. No result or traffic figure is claimed for it. The full record with its evidence limits is the classifieds platform with AI-assisted moderation page; every published record is listed under work.

Other published commerce work adds named results. For 4over4, which may be named, Netbase delivered e-commerce conversion work on its print store, including a product recommendation engine built on browsing history and previous purchases; the published 4over4 case study records revenue up 82% within six months and conversion up 48%. For an EU fashion-tech marketplace that is not named, gross merchandise value rose 47% and vendor onboarding time fell 60% within six months of launch. Netbase has also built web-to-print stores with online design tools for unnamed print businesses in Singapore, France, the Netherlands and South Korea, and Magento stores with Android and iOS apps for unnamed fashion retailers in the Netherlands and Nigeria and a parts and equipment seller.

Both sets of figures are one client's outcome, never a benchmark for yours.