DMD CRM System

End-to-end CRM design for an auto-dismantling business. The goal was to remove the manual overhead from parts tracking, speed up warehouse lookup, and give the operator a single screen with everything needed to take an order and dispatch it.

DMD CRM System
Brand

DMD Auto Parts

Client

NDA

Location

Україна

Budget

>15 000$

Review
5/5

About project

End-to-end CRM design for an auto-dismantling business. The goal was to remove the manual overhead from parts tracking, speed up warehouse lookup, and give the operator a single screen with everything needed to take an order and dispatch it.

Challenges

DMD's auto-dismantling business handles hundreds of thousands of SKUs, each passing through a long chain: vehicle selection for dismantling → part removal → washing → refurbishment when needed → photography → marketplace listing → order intake → dispatch → accounting. At the start of the project every stage lived in its own tool: Excel files for the catalogue, notebooks on the warehouse floor, a separate accounting suite, manual photo uploads to OLX and Prom. Information kept getting lost at the seams, operators duplicated each other's work, and a single order intake stretched up to 12 minutes. The hard part was modelling the whole cycle in one system so a new operator could start working within a day and the critical 10-15-second part lookup still worked across 200K+ items.

Solutions

I shipped a full operational cycle for the client — from vehicle purchase to after-sales follow-up. Six stages, each with its own sub-operations and dedicated dashboard. Transitions are automatic — a part moves down the pipeline without manual handoff.

Stage 1 — Vehicle purchase and logistics. Vehicle search on auctions and from private owners, condition and dismantling-profitability assessment, purchase, delivery setup, customs clearance for imports, intake at the warehouse with VIN capture and an auto-generated assembly map.

Stage 2 — Dismantling and initial processing. Dismantling plan per vehicle model, part removal logged in the system, sorting by category (engine, body, interior, electronics), initial fault check with statuses «working / needs repair / scrap».

Stage 3 — Part preparation. Washing and cleaning, detailed function check, repair when needed, final condition grading (A — like new, B — working with marks of use, C — buyer-side repair needed).

Stage 4 — Content and listing. Photography with an angle template and fixed lighting, description authoring with specs pulled from the database, one-click listing on OLX, Prom, AutoRia and the company site — no manual duplication.

Stage 5 — Tracking and storage. Unique SKU labelling, packaging at the right protection level, warehouse placement by «hot / warm / cold» access — fast-movers closer to the dispatch zone.

Stage 6 — Sales and communication. Order intake from four channels (phone, website, marketplaces, messengers), cross-sell of complementary parts, dispatch with auto-generated shipping labels, after-sales follow-up and feedback collection for ratings.

Design

The design system is built around high information density without overload. Every screen has a clear visual hierarchy: critical actions top-right, secondary on the left, status at the bottom. Light theme is the default, dark theme is optional (the operator toggles it in one click).

Screen architecture. A left sidebar of modules (Vehicle intake, Dismantling, Refurbishment, Listing, Orders, Accounting), the main working canvas in the centre, and a contextual panel on the right with the selected object's summary and quick actions. The sidebar collapses to icons for experienced operators — freeing 180 px of width for the work surface.

Catalogue. Virtual scroll handles 200K+ rows without lag. Each row is a compact card with state (colour-coded: green/yellow/red), a 24×24 px photo, key attributes and price. Inline editing — click the price, type a new one, Enter; no modal, no form-filling routine.

Search. Cmd/Ctrl+K opens a global palette with autocomplete by VIN, make, model, supplier SKU. Debounce 120 ms, results in 80–150 ms across 200K SKUs. Recent searches are cached — in 70% of cases an operator searches the same thing they did five minutes ago.

Bulk operations. Shift+Click for ranges, Cmd+Click for toggling — standard macOS pattern, so even new hires get it intuitively. Select 50 parts → one action: change status, move to a different storage zone, publish to marketplaces.

Real-time. WebSocket updates with no page reload — a new part appears in the list with a soft animation, a status change is highlighted for 1.5 s. Toast notifications sit bottom-right and don't distract from the work area.

Empty + loading states. Every empty list has an illustration plus one CTA button («Accept the first vehicle», «Add the first part»). Skeleton loaders on 100% of screens — the user sees the layout before the data lands, the feeling is «instant».

Keyboard-first. Every primary action is mouse-free: J/K to navigate the list, E to edit, Space to preview, Esc to leave. On the real warehouse floor operators can log work with one hand while holding a part in the other.

Accessibility. AA+ contrast across all states, focus rings on every interactive element, ARIA labels for screen readers, reduced-motion respected for toast animations.

Technologies

FigmaReactNode.jsPostgreSQL

Development

Frontend — React with a custom design system on Radix + Tailwind, fully type-safe with TypeScript. Backend — Node.js with PostgreSQL and full-text search via pg_trgm + ts_vector. Real-time inventory updates over WebSocket. Integrations — telephony (Asterisk), 1C, Nova Poshta, marketplaces (OLX, Prom, AutoRia).

Result

Order intake time dropped from 8–12 minutes to 2–3 (-70%). Dispatch errors fell from 4% to 0.3%. The client retired three operator positions because one person now handles the load of three. The project ROI paid back inside 4 months of use.