How to Choose an ERP System for a Small Shop
Decide on an ERP by starting from your shop's real problems and process—not a feature brochure. For a 5–100 person job shop or small manufacturer, the right system is the one that runs one of your real jobs end to end, that your floor will actually use, and that you can leave with your data if you outgrow it.
This guide is written from a shop-floor point of view: plain language, shop examples, and no fluff. Use it with the SmallShop ERP Picker as a shortlist starter—not a recommendation.
Quick checklist
- Write down the three problems that cost you money or sleep this month.
- Map quote-to-cash on one page before you watch a demo.
- Decide job-shop vs make-to-stock vs mix.
- Inventory the data you already keep (spreadsheets count).
- Budget software plus implementation, training, and your people's time.
- Trial with one real job—not the vendor's sample part.
- Confirm data export before you sign.
- Plan floor adoption like a process change, not an IT install.
- Baseline metrics before go-live; re-check at 30 / 90 / 180 days.
Answer a few questions about your shop and get a shortlist to trial.
What guiding principles should I use when choosing an ERP?
Start from problems and process, not features
Software vendors sell feature grids. Shops buy relief from pain.
Before you open a comparison table, list what broke last quarter. Examples from real floors:
- You cannot answer "when will it ship?" without walking the aisle.
- Quotes are guesses because nobody compares estimated hours to actuals.
- Material shows up short at the machine, not at the purchase order.
- The whiteboard schedule and the shared drive disagree.
- A customer asks for lot history and someone rebuilds it by hand.
Map quote-to-cash on one page: quote → order → material → traveler/routing → operations → ship → invoice → cost close. Mark where data is retyped. That map is your requirements doc. Features that do not touch those pain points are distractions.
Lean lesson that still holds: an ERP will automate the process you have. Fix naming, travelers, and handoffs first. Clean item masters are most of an implementation.
Match job-shop vs make-to-stock fit
Job-centric work is quoted, routed, and costed as its own job. Think custom brackets, one-off fixtures, or low-repeat CNC with different setups every week.
Item-centric / MRP-style work plans material from known BOMs and demand. Think repeat assemblies, kits, or light make-to-stock.
Mixed shops are common. Ask a blunt question: what share of last year's revenue was a part number you had made before? Mostly repeat → item-centric tools can fit. Mostly new quotes → job-centric systems usually fit better. Small-shop MRP tools such as MRPeasy or Katana are examples in the item-centric lane; judge them (and every other vendor) against your mix in a live trial—not against a brochure.
Use the data you already have
You already have an ERP. It is scattered across Excel, email, QuickBooks, a whiteboard, and people's heads.
List it:
- Part numbers and revisions (even if messy)
- BOMs or traveler PDFs
- Supplier and customer lists
- Open jobs / WIP
- Inventory counts you trust (or know you do not)
- Quoted vs actual hours on a few recent jobs
Bring that list to demos. Ask: "Show me where this lands in your system." If the answer is "we'll customize that later," write it down as risk.
Count total cost, not just subscription
Price the whole change:
- Software subscription or license
- Implementation help (vendor, partner, or your own weekends)
- Training for office and floor
- Integrations (accounting, CAD, shipping)
- Parallel running (old process + new for a stretch)
- The hours your estimator, buyer, and lead hands will spend cleaning masters
A cheap monthly fee with a brutal data cleanup is not cheap. A higher fee that imports your travelers cleanly can be cheaper in year one. Compare on a 3–5 year view with the same assumptions for every finalist.
Trial with your own real job
Do not accept only the polished demo part.
Pick one recent job that had a hiccup: a late supplier, an engineering change mid-build, a scrap loop, or a rush ship. Ask the vendor to walk that job through quote or order, BOM/routing, material issue, operation complete, ship, and cost close.
Time how long it takes your people to enter it after a short training. That single afternoon beats a week of feature checklists.
Watch vendor lock-in and data export
Before you sign, ask for a written answer:
- Can I export item masters, BOMs, routings, inventory transactions, open and closed jobs, and quality records?
- In what formats?
- Who runs the export—me, or a paid services ticket?
Run an export during the trial. If you only get screenshots, you do not own your history.
Plan adoption on the floor
The best system fails if operators will not touch it.
Decide who enters labor and material at the machine: the operator, a lead, or the office. If the UI needs a mouse and three screens on a greasy bench, expect paper travelers to stay. Prefer simple shop-floor screens, barcode scans, or tablets you will actually mount.
Pick a floor champion. Run a short pilot cell before full go-live. Measure whether travelers get updated the same day—not whether IT checked a box.
What data can an ERP manage for a small manufacturer?
Here is the core data most small manufacturing ERPs (and many MRP tools) touch. Plain language, with a shop example for each.
Inventory
On-hand quantity, location, and often reserved vs available stock.
Shop example: You have 40 pieces of 6061 bar in "raw." Three open jobs need 12 pieces each. Without reservation, the third job discovers the shortage at setup. With inventory tied to jobs, the buyer sees the gap when the job is released—not when the spindle is waiting.
Bills of material (BOMs)
The structured list of components and quantities that make a finished or subassembly item.
Shop example: A weldment needs plate, tube, fasteners, and a bought-in hinge. Change the hinge supplier and every open job that uses that BOM revision should pull the new part—not a sticky note on one traveler.
Routings / travelers
The sequence of operations, work centers, and standard times to make the part.
Shop example: Saw → mill → deburr → inspect → ship. If "inspect" is missing from the routing, you will under-quote capacity and skip a gate your customer expects.
Purchasing
Purchase orders, expected receipts, and often supplier lead times tied to jobs or stock.
Shop example: Job 4521 needs a special fastener with a 3-week lead. Purchasing should see that requirement when the job is planned—not when assembly is already late.
Sales orders
Customer orders, line items, promise dates, and links to jobs or shipments.
Shop example: A customer adds five more pieces to a live order. Sales, planning, and purchasing should see one change—not three emails and a rewritten spreadsheet.
Scheduling
Which job runs when, on which work center, given capacity and priorities.
Shop example: Two rush jobs both need the same 4-axis machine Friday. A schedule that shows load beats a whiteboard argument at 6 a.m.
Quality and traceability (lot / serial)
Lot or serial identity from material receipt through operations to shipment, plus nonconformance hooks in stronger systems.
Shop example: A customer asks which heat lot shipped in box 12. You should answer from the system in minutes, not from a binder hunt. (This is operational traceability—not a claim about any regulatory certification.)
Job costing
Estimated vs actual material, labor, and overhead by job.
Shop example: You quoted 6 hours of mill time. Actual was 11 because of a bad fixture. Next quote for a similar part should see that variance—or you will keep buying work at a loss.
Accounting hooks (e.g. QuickBooks / Xero)
Many small shops keep QuickBooks or Xero for the books and use manufacturing software for inventory, jobs, and purchasing. The ERP (or MRP) then pushes invoices, bills, or journal entries across.
Shop example: When job 4521 ships, the invoice lands in QuickBooks without retyping line items. Ask for a live demo of your chart of accounts mapping—not a generic slide.
How do I know I'm getting ROI from an ERP?
You cannot prove ROI after go-live if you never measured the before. Baseline first.
Baseline metrics before go-live
Track these for a few weeks (or pull the last quarter from what you already have):
| Metric | What to capture |
|---|---|
| On-time delivery | % of lines or jobs shipped by the promise date |
| Quote turnaround | Calendar days from RFQ to sent quote |
| Inventory value / turns | Dollars on hand; turns if you can compute them |
| Expedite / premium freight | Dollars spent last quarter to "save" a late job |
| Admin hours on data entry | Hours/week retyping between quote, PO, schedule, and invoice |
| Quoted vs actual job cost variance | Average % or dollar miss on closed jobs |
Write the numbers on one sheet. Date it. That sheet is your baseline.
Simple payback math (EXAMPLE — illustrative only)
These numbers are illustrative. They are not a study, not an average, and not a promise. Plug in your own.
EXAMPLE assumptions (made up for teaching the math):
- Combined software + implementation + training in year 1: $18,000
- Estimated annual savings you believe you can defend:
- Fewer premium freight events: $6,000
- Admin time recovered (5 hr/week × 50 weeks × $40/hr): $10,000
- Less scrap / rework from clearer travelers: $4,000
- Total annual benefit (example): $20,000
Payback (example):
$18,000 ÷ $20,000 per year ≈ 0.9 years (~11 months).
If your honest savings are half of that, payback doubles. If implementation slips and costs rise, payback stretches. The point is the worksheet—not the sample dollars.
What to measure at 30 / 90 / 180 days
30 days after go-live
- Are travelers and inventory transactions entered the same day?
- How many workarounds (side spreadsheets) still exist?
- Support tickets: training gaps vs software gaps?
90 days
- Re-run on-time delivery, quote turnaround, and expedite freight vs baseline.
- Sample 10 closed jobs: is quoted vs actual easier to see than before?
- Is purchasing working from system demand or from tribal knowledge?
180 days
- Full baseline rematch on the table above.
- Inventory accuracy spot-checks (cycle counts vs system).
- Decision: expand modules, freeze scope, or fix process debt you automated too early.
If floor adoption is weak at 30 days, pause new modules. Fix the process and training first. More features on top of unused screens will not create ROI.
Summary: how should you decide?
- Name the problems. Map the process.
- Pick the planning model that matches your revenue mix (job vs stock vs mix).
- Bring your real data and one messy real job to every demo.
- Price the total change, including your people's time.
- Confirm export and floor adoption before you sign.
- Baseline metrics; rematch at 30 / 90 / 180 days.
When you want a starting shortlist filtered by shop type and needs, use the SmallShop ERP Picker. It is a directory, not a verdict—fit still depends on your routings, quoting, and quality process.
MRPeasy, Katana, QuickBooks and Xero are trademarks of their respective owners. No affiliation or endorsement is implied.
FAQ
- How do I choose an ERP system for a small manufacturing business?
- Start from the problems and process that hurt you today—not a feature list. Map quote-to-cash, decide whether you are mostly job-shop or make-to-stock, inventory the data you already keep in spreadsheets, budget total cost including implementation and training, and trial the shortlist with one real job before you sign.
- What is the difference between ERP and MRP for a small shop?
- MRP focuses on material planning from bills of material and demand. ERP usually covers that plus sales orders, purchasing, scheduling, inventory, quality, job costing, and often accounting hooks. Many small-shop tools blur the line and call themselves MRP or manufacturing ERP—judge the fit by your process, not the acronym.
- Should a job shop buy the same ERP as a make-to-stock manufacturer?
- Not always. Job shops live on quoted, routed, one-off or low-repeat work with job costing and traveler control. Make-to-stock or light assembly shops lean on item masters, BOMs, and material planning. Mixed shops should estimate what share of revenue is repeat part numbers before picking a model.
- How long before an ERP pays for itself?
- It depends on your baseline waste and how clean your data is at go-live. Capture on-time delivery, quote turnaround, inventory value and turns, expedite freight, admin hours on double entry, and quoted-vs-actual cost variance before you switch. Re-measure at 30, 90, and 180 days. Use simple payback math with your own numbers—not a vendor slide.
- What data should I clean up before implementing an ERP?
- Part numbers, units of measure, BOMs or travelers, supplier and customer lists, and open jobs or WIP. Software automates whatever process and master data you feed it. A messy item master is most of a painful implementation.
- Can I keep QuickBooks or Xero if I add a manufacturing ERP?
- Many small shops keep their accounting system and use the manufacturing tool for inventory, jobs, and purchasing, then sync invoices or journal entries. Ask each vendor to show the live integration path with your books—not a slide—and who owns the chart of accounts mapping.
- How do I avoid getting locked into an ERP vendor?
- Before you buy, ask for a written data-export path: item masters, BOMs, routings, open and closed jobs, inventory transactions, and quality records. Run an export during the trial. Prefer systems that let you leave with usable files, not screenshots.