E-Commerce Chat Library and Admin Tools
A New York e-commerce platform faced a classic build vs buy call on chat: extend an existing React Native library that couldn't reach the web, or bring in an embedded engineer to build a custom one.
- Geography
- New York City, New York
- Year
- 2023
- Stage
- 11 to 50 employees
- Team
- One Leanware engineer embedded with the in-house team
- Duration
- April 2023 to ongoing
- Engagement size
- Confidential
The situation
An e-commerce platform for independent food distributors was weighing build vs buy on two long-requested back-office features, and needed a partner that could execute either path without disrupting a product already in daily use by customers. Creating accounts for a customer's employees, setting permissions, and linking them to the right entities was still a manual process on the team's side, which meant staff time spent on account setup instead of anything that grew the business.
The second gap was chat. The platform's mobile app already had messaging, but it was built on a React Native library that could not be reused in the back office, so support and account teams had no equivalent way to message customers from the web product. Both features had to slot into an existing design system and an established codebase rather than introduce a parallel stack, which is what tipped the decision toward building rather than buying an off-the-shelf widget.
What we built
Leanware supplied one engineer, embedded directly with the in-house team rather than delivered as a separate project team. That single point of contact carried both features from scope to ship.
The employee management work added a new page to the back office so customers can create employee accounts, set permissions, and link them to entities themselves, closing out the manual process that used to run through support. The chat work was the heavier lift: a custom chat and messaging React library built from scratch, matched to the client's design system and wired into their existing stack of Next.js, React, React Native, Testing Library, Apollo, GraphQL, and Hasura.
The engagement ran on scrum, week-long sprints, with check-ins three times a week to keep the embedded engineer aligned with the in-house team's priorities. Product specs arrived incomplete more often than not, so a working pattern for an embedded engineer here was asking the right clarifying questions early rather than blocking on documentation that did not exist yet. The same held for the codebase itself: conventions and patterns were largely undocumented, and the engineer picked them up by reading the code rather than requiring a walkthrough.
Outcome
Both projects shipped within the expected timelines, alongside a series of smaller fixes and enhancements across the applications. A Clutch review from the client credits how quickly the engineer onboarded to undocumented code, the steady on-time sprint delivery even through mid-sprint pivots, and a collaboration that felt like one team rather than a third-party vendor.
Per the confidentiality terms of the engagement, the client is identified only as an e-commerce company, and the engagement value stays confidential.
"We were thoroughly satisfied with Leanware's performance and the overall collaboration."
— Confidential , Engineer , E-Commerce Company · New York City, New York
Engagement FAQ
How do you decide between building a custom feature and buying an off-the-shelf tool for an ecommerce back office?
Buying makes sense when a third-party tool can match your design system and codebase conventions without heavy customization. Building makes sense when it can't, here, the mobile chat library was tied to React Native and couldn't reach the web, and an off-the-shelf widget would have looked and behaved like a bolt-on rather than a native part of the product. When a feature has to slot invisibly into an existing system, that's usually the signal to build.
How long does it take to build a custom chat library for an ecommerce back office?
A chat and messaging library built from scratch and matched to an existing design system is a multi-sprint build, not a single-sprint bolt-on. In this case, it shipped within the engagement's expected timeline, running on week-long scrum sprints with check-ins three times a week — alongside a second, smaller feature (the employee management page) delivered by the same embedded engineer in parallel.
Can one embedded engineer handle a feature build inside a live ecommerce platform without disrupting it?
Yes, when the engineer works inside the existing team's process rather than around it. Here, a single Leanware engineer ran scrum with the in-house team, checked in three times a week, and delivered two separate features (a chat library and an employee management page) without a separate parallel workstream.
Why build a custom chat library instead of using an off-the-shelf messaging provider?
A platform's existing mobile chat was tied to a React Native library that could not be reused on the web, and the back office needed to match the client's own design system rather than a third-party widget's look and behavior. Building from scratch was the path that fit both constraints.
How do you replace a manual employee account process in an ecommerce back office?
By giving customers a self-serve page to create accounts, set permissions, and link employees to the right entities themselves, removing the step where staff handled account setup manually. That is the shape of the employee management feature shipped in this engagement.
How much does it cost to build a custom chat library instead of buying an off-the-shelf provider?
The engagement value here is confidential, so there's no dollar figure to share. What is public is the shape of the cost driver: this was staffed with a single embedded engineer running week-long sprints, not a dedicated project team — that staffing model is usually what keeps a custom build's cost closer to a buy decision than teams expect, especially when the alternative (an off-the-shelf messaging widget) still needs custom work to match an existing design system.