Skip to main content
GullySystem
A practical guide

Build or buy software for EV component traceability

Ganesh HSStrategy and Technology, GullySystem
Practical advice · 5 min read

In this guide 6 sections
The short answer

Buy a ready product when your records are serial numbers, cell lots and end-of-line test results that most EV suppliers share. Build custom when your test rigs, customer portals or cell grading logic are unusual enough that every release needs constant workarounds and the vendor’s screens fight your process. Many makers do both.

Write down the records you must be able to produce

Expect these questions after a field failure. Which cells went into this pack, which torque tool tightened the busbar, what did the insulation resistance test read on the day, and who ran it?

List each record as a field. A battery pack needs pack serial, module serials, cell lot, cell grade, BMS firmware version and end-of-line result, and each field needs a named owner. A motor needs stator lot, magnet batch, balancing result and hipot reading, and a charger needs board serial, calibration file and output test reading.

Count what your files hold. The gaps decide the project. Be strict. A field that someone remembers to type is not a record, and a pass mark without the reading will not satisfy a customer asking for the number months later.

  • Serial and lot numbers for each child part
  • Test readings, not just pass or fail
  • Firmware and calibration versions
  • Operator, station and tool identity
  • Customer-specific labels and data formats

Where a ready product wins

A product wins on the common work. Scanning a child serial into a parent pack, blocking a module that failed the earlier hipot station and printing a pack label are well-trodden ground. Genealogy reports come built in. Vendors have already met the awkward cases.

Upkeep also sits with the vendor. Barcode scanner drivers, label printer changes and operating system updates are not your problem, so your own people spend their hours on the line, not on driver fixes.

With two lines and standard stations, buying is usually the quicker path to a working traceback, because setup means mapping fields to your serial formats, not writing code.

Where a custom build earns its cost

Custom wins on unusual equipment. A cell grader that exports a CSV, a leak tester with a serial port or a customer portal that demands a fixed XML upload each call for integration work.

Process can be the advantage. If you sort cells into packs by matching capacity and internal resistance, the matching rule is yours, and burying it inside a vendor setting makes it hard to change.

Be honest about ownership. Custom code needs someone who can fix it on a Saturday, and if nobody at your plant or developer answers when a line stops, the licence saving is gone. Name that person before you sign.

  • Test rigs with no standard interface
  • Cell or magnet matching rules unique to you
  • Customer portals with fixed upload formats
  • A line layout that changes every quarter

Test both options against one real recall drill

Pick a past pack serial. Ask each option to show every part inside that unit and every other unit built from the same cell lot, then time the answer.

Do the same in reverse. Start from a cell lot and list the finished packs, their shipment dates and their customers, and time that answer as well. One direction only is a weak buy. Skipping the test is a weak build.

Check how each option behaves with a station offline, because paper fallback and later entry should be possible without breaking the genealogy chain. Test it. Unplug a scanner.

A hybrid is often the honest answer

Buy the core genealogy and label engine, and build the small connectors for your testers and customer uploads around it, which keeps the unusual logic in your hands. Nothing standard is rebuilt.

Before signing anything, write one page covering the records, the stations, the integrations and the drill above, then show it to two vendors and one developer. Compare what each says about the same page, because differences in their answers show where the real risk sits, and weigh those gaps before you weigh the price.

Traceability option scorecard

Each record field gets its own row, such as cell lot, module serial, torque result, firmware version, insulation reading and customer label format. Columns show the data source, whether a ready product captures it, whether a build is needed and who owns the fix, so the call is made field by field. A bottom block times the recall drill.

Open a blank worksheet to print

Questions owners ask

Does traceability need a full MES?

Not always. A smaller system linking child serials to a parent, storing test readings and printing labels may be enough, while an MES adds scheduling and machine monitoring you may skip.

How long should we keep EV component records?

Follow the retention period written in your customer contracts and any rule that applies in your market, since many customers ask for years beyond warranty. Confirm with your quality head before deciding storage.

Can we start with Excel and move later?

Yes, if the fields are clean from the start, because messy serial formats and free-text lots make migration painful later, so fix the formats first, even inside a spreadsheet.

What if our customer demands its own system?

Then your software must feed it, so check the required file format early, since that integration often becomes the main custom piece of the project.

Browse all guidesBack to top ↑
Next step

Have a specific situation to work through?

This article covers the general case. Tell us what you’re actually dealing with and we’ll respond directly.

Share Your Requirements