10 questions to ask before signing with any factory automation vendor

Most SMB factory owners we talk to have shopped at least 3 vendors before getting to a real conversation. The shopping experience is generally bad — slick demos, vague timelines, optimistic pricing, and a creeping sense that the gap between what's promised and what gets delivered will be expensive.

This is the list of questions we'd want every owner to ask before signing with any vendor — including us. We answer them honestly on discovery calls.

1. "Where exactly does my data live, and who owns it?"

The honest options are: vendor's cloud (you rent), your cloud (you own), or your factory's server (you own and host).

Watch for: vague answers about "the platform", anything that implies your raw production data is mixed with other customers' in shared schemas, anything that won't put data ownership in writing. If you can't get a clean answer here, every other question gets harder.

2. "What happens if I leave?"

Ask for the data export specification. Format, frequency, completeness. Ask for it in writing.

A good vendor: "Daily snapshot of all your data in standard formats, exportable any time, with a documented schema. If you leave, you take everything." A bad vendor: "We can export reports for you" — meaning you don't get the raw data, just summaries.

This question alone separates serious vendors from rentier ones.

3. "How fresh is the data on the dashboard?"

Sub-minute is the modern standard. 5-minute refresh is acceptable for some metrics. 15-minute "real-time" is a red flag — it means batch processing somewhere, which means alerts arrive late.

Ask specifically: "If a line stops at 14:32:05, when does the supervisor see it on the dashboard?"

4. "What's your stack at each layer?"

Sensors, gateway, backend, API, dashboard. (We wrote a pipeline overview covering each layer.)

A vendor who can answer "we use Modbus to OPC UA via Advantech UNO, MQTT to AWS IoT, TimescaleDB for time-series, GraphQL with Redis cache, React dashboard" has built systems before.

A vendor who can only show you screenshots probably hasn't, or doesn't want you to know what they use. Either is a yellow flag.

5. "How are downtime reasons captured?"

The right answer: auto-categorised by equipment state where possible (machine fault codes, planned PMs, line stops), supervisor-tagged where not, with structured reason codes — not free-text.

A bad answer: free-text only. The data looks fine for a week. After 6 months you have 800 unique "reasons" that nobody can analyse.

6. "What happens if the network drops for an hour?"

A robust system buffers locally on a field gateway and resyncs when reconnected. Without local buffering, every dropped packet is data lost.

Ask specifically how long the buffer holds — minutes, hours, days. For a Malaysian factory with occasional weather-related connectivity blips, you want at least an hour of local buffer.

7. "Have you shipped at least one similar project in Malaysia?"

Local context matters. Industry 4.0 projects fail in country-specific ways: TNB peak-tariff billing, Industry4WRD criteria, MyInvois deadlines, Bahasa-only floor staff, public-holiday density.

A vendor who's never shipped in Malaysia learns these on your project, on your dime. Ask for an anonymised reference.

8. "What's the milestone payment structure?"

Standard SMB payment split: 30% on signing, 40% on production-ready milestone, 30% on sign-off. Larger projects split further.

Watch for: 50% upfront with vague "balance on completion" terms (incentivises starting fast and shipping anyway), or 100% at start (no), or all-cash-on-delivery (suggests scope creep is already expected).

9. "What does the support model look like after launch?"

The honest answer: first month on-call (typically included), then a small monthly retainer for fixes / dashboard tweaks / alert tuning / hosting / monitoring.

A bad answer: "We hand it over and you're on your own" (reality: the dashboard breaks, you can't fix it, the vendor sells you a new contract at emergency rates). Or: "We charge per ticket" (incentivises slow response).

Ask for the retainer cost up-front, not after the project ships.

10. "What would you advise against doing first?"

This is the question that separates honest vendors from sales-driven ones.

A good vendor will name a specific scope they wouldn't recommend. ("Don't try to instrument 8 lines on the first project — start with 1.") A bad vendor will agree to everything you propose.

If they can't think of anything they wouldn't recommend, they're either too eager to close or haven't shipped enough to know where projects fail.

Bonus question: "Can I talk to your last 3 customers?"

Most vendors will say yes; the test is what they actually deliver.

A great vendor will give you 3 real names and contact details, not curated case studies. They'll let you talk to a customer whose project went sideways too — those references tell you the most.

A vendor who only gives you their best 2 success stories from years ago is hiding something more recent.

What we say to our own list

For full disclosure, here's how we answer our own list:

1. Data: AWS, Singapore region by default, your account if you prefer. Daily backups. Schema documented. Yours.
2. If you leave: clean export, structured formats, your code repo handed over.
3. Data freshness: sub-minute on dashboards, sub-second on alert paths.
4. Stack: documented in our pipeline article.
5. Downtime reasons: auto-categorised + supervisor-tagged with structured codes.
6. Network outage: local gateway buffer ~6 hours by default, configurable.
7. MY references: yes, two anonymised case studies on the site, real references on request.
8. Payments: 30/40/30 standard, can split further for larger projects.
9. Support: first month included, then a small monthly retainer covering hosting, monitoring, fixes, and 1–2 small dashboard tweaks per month.
10. What we'd advise against: instrumenting too many lines on the first project, building everything custom when commodity SaaS works for some pieces, and rushing past the "trust-building" phase after go-live (covered in our timeline article).

If you'd like to walk through any of these for your specific factory, drop us a line — free 30-minute discovery call, no obligation.