← Back to Field Notes

July 7, 2025 · The Bloomfield Team

7 Questions Every Manufacturer Should Ask Before Buying Software

A 2024 survey by the Manufacturing Technology Centre found that 62% of manufacturers who purchased new software in the previous three years reported that at least one system was underused or abandoned. The average investment in those failed implementations was $147,000. That number does not include the months of disrupted operations or the team morale burned in the process.

Software purchases in manufacturing carry a specific set of risks that vendors rarely discuss during the sales cycle. These seven questions, asked before you sign anything, will surface the problems that kill implementations before they start.

1. Does This Solve a Problem We Have Measured?

The first question is also the most frequently skipped. A vendor demos a feature that looks powerful. The team nods. Someone says "we could use that." A purchase order follows.

The problem: "we could use that" is a different statement than "this solves the specific bottleneck that costs us $X per month." Software that solves a measured problem has a built-in success metric. Software purchased on instinct gets evaluated on feeling, and feelings shift the first time the implementation hits a rough patch.

Before evaluating any tool, define the problem in numbers. Quoting takes 4.5 days. On-time delivery sits at 78%. Scrap rate on first-run parts exceeds 6%. These are problems you can measure before and after. Everything else is a wish list.

2. How Does This Connect to Our Existing Systems?

Every manufacturer already runs an ERP, probably a CAD/CAM system, possibly a quality management system, and almost certainly a collection of spreadsheets that fill the gaps between all three. New software that does not connect to these existing systems creates another data island.

Ask the vendor specifically: does this integrate with our ERP through a native connector, an API, or a manual export/import? What data flows automatically and what requires someone to re-enter it? How does the system handle changes in the ERP after initial sync? For a broader look at how manufacturing systems should connect, see our complete guide to AI in manufacturing.

The answers to these questions determine whether the software becomes part of your workflow or sits alongside it as one more thing someone has to maintain.

3. Who on Our Team Will Own This Daily?

Every piece of software needs a human operator who uses it as part of their daily work. If you cannot name that person before you buy, the software will not get adopted. Full stop.

The owner is the person who enters data, runs reports, flags issues, and tells the vendor what is working and what needs adjustment. In shops under 50 people, this person usually has another full-time job. That means the software needs to make their existing job faster, or it will be the first thing they stop using when the week gets busy.

4. What Does Implementation Actually Look Like?

Ask for a week-by-week implementation timeline with named milestones. Ask how many hours per week your team will need to dedicate during setup. Ask what data you need to prepare before go-live and how long that preparation takes in similar shops.

A vendor who cannot answer these questions with specifics has either never implemented at a shop your size or is deliberately keeping the scope vague. Both are disqualifying signals. The best implementations we have seen follow a pattern: two weeks of data preparation, two weeks of configuration, one week of parallel running where the old and new systems operate simultaneously, then a clean cutover with the old system available as backup for 30 days.

5. What Happens When Something Breaks?

Manufacturing is a Thursday-afternoon-the-machine-is-down business. If your quoting system goes offline during a rush week, or your scheduling tool throws errors when someone enters a non-standard work order, how fast does support respond?

Get the specifics in writing. Response time SLA. Resolution time SLA. Whether support is domestic or offshore. Whether you get a named contact or a ticket queue. The difference between a four-hour response and a four-day response can be the difference between a minor inconvenience and a week of manual workarounds that nobody remembers how to do.

6. Can We See It Working With Our Data?

Generic demos use generic data. Generic data looks clean, complete, and well-structured. Your data does not. Your ERP has description fields with 15 different naming conventions. Your job records span three system migrations with inconsistent formatting. Your material pricing data has gaps from the years when purchasing tracked it in email instead of a database.

Ask the vendor to run a proof of concept with your actual data. A subset is fine: 50 job records, 30 quotes, one year of purchase orders. The way the software handles your real data, with all its messiness, tells you more about the product than any slide deck.

7. What Is the Total Cost Over Three Years?

The license fee is the number the vendor wants you to focus on. The total cost includes: implementation fees, customization charges, annual maintenance, per-user license increases as your team grows, data migration costs, training time for current employees, training time for every new hire over three years, and the opportunity cost of the hours your team spends maintaining the system.

Build the three-year total. Compare it to the measured cost of the problem from Question 1. If the software costs $180,000 over three years and the problem it solves costs you $50,000 annually, the math works. If the problem costs $20,000 annually and the software costs $180,000, you are buying capability you do not need at a price you cannot justify.

The Question Underneath All Seven

Every question on this list reduces to a single concern: will this tool become part of how we work, or will it become one more system we maintain alongside the way we actually work? The manufacturers who answer that question honestly before they buy are the ones whose implementations succeed. The manufacturers who skip it join the 62% who spent six figures on software their team stopped using.