← Back to Field Notes

February 28, 2025 · The Bloomfield Team

Every Manufacturer Needs a Data Strategy. Here Is Where to Start.

A 50-person job shop generates more operational data in a year than most SaaS companies with twice the headcount. Work orders, time entries, inspection records, material certifications, supplier invoices, customer communications, machine parameters, quality nonconformances, shipping logs. The data exists in the ERP, in spreadsheets, in email inboxes, in filing cabinets, and in the heads of the people who run the operation.

None of it is connected. The quoting team cannot see production cost data without exporting and manipulating it. The production manager cannot see quoting commitments without asking the front office. The owner cannot see profitability by customer or by part type without a week of spreadsheet work. Every operational decision is made with partial information because the data that would complete the picture lives in a different system, a different format, or a different person's memory.

What a Data Strategy Actually Means

A data strategy for a manufacturer is not a technology project. It is a set of decisions about what data the operation needs to capture, where it lives, how it connects, and who uses it to make decisions. The technology comes later. The decisions come first.

The strategy answers four questions. What decisions does each role in the organization make regularly? What data would improve those decisions? Where does that data currently live? What is missing, and how does it get captured?

Step One: Map the Decision Points

Every role in a manufacturing operation makes recurring decisions. The estimator decides how to price a job. The production scheduler decides what runs on which machine this week. The purchasing manager decides when to order material and from whom. The quality manager decides whether a nonconforming lot ships, gets reworked, or gets scrapped. The owner decides where to invest, which customers to pursue, and when to add capacity.

List every recurring decision by role. For each one, document what data the person currently uses and where they get it. This exercise reveals the gaps. The estimator uses their memory and a spreadsheet because the ERP data is too hard to access. The scheduler uses a whiteboard because the ERP scheduling module does not reflect reality. The owner uses gut feel because a real profitability analysis would take a week to assemble.

Step Two: Identify the Core Data Sets

Most manufacturing operations run on six core data sets: job cost records, customer and order history, material and supplier data, quality records, machine and production data, and employee time and labor data. These six sets, when connected and accessible, support 90% of the operational decisions the business makes daily.

For each data set, document where it lives (which system, which spreadsheet, which person's knowledge), how complete it is, and how current it is. A shop that enters time data into the ERP three days after the job closes has a completeness problem. A shop that has five years of job cost records with inconsistent part numbering has a quality problem. A shop where material pricing lives only in the purchasing manager's email has an accessibility problem.

Step Three: Fix the Foundation Before Building on Top

The temptation is to jump to dashboards and AI tools. That impulse is premature if the underlying data is incomplete, inconsistent, or inaccessible. A dashboard built on bad data produces confident bad decisions, which is worse than no dashboard at all.

Fix data quality first. Standardize part numbering. Ensure time tracking captures setup and run time separately. Confirm that job cost records include all cost categories, not just direct labor and material. Establish a process for entering data consistently going forward. The cleanup is unglamorous. It is also the foundation that everything else depends on. For a closer look at getting your data ready for AI tools, see how to prepare AI-ready data in a 50-person shop.

Step Four: Connect the Data Sets

Once the core data is clean and consistently captured, the next step is connecting data sets so that questions that span systems become answerable. What is the actual margin on aerospace work versus commercial work? Which customers have the highest quoting-to-win conversion? Which machines have the most unplanned downtime and how does that correlate with delivery performance?

These questions require joining data from the ERP, the quality system, the scheduling system, and the quoting system. The integration approach depends on your systems, but the principle is the same: data that cannot be connected cannot answer cross-functional questions, and cross-functional questions are the ones that drive the business forward.

Step Five: Put Data at the Point of Decision

Data that requires a 30-minute export and analysis session before it informs a decision is data that rarely gets used. The goal is to surface the right data at the moment the decision is being made. The estimator sees comparable past jobs when they open an RFQ. The scheduler sees real-time machine status when they adjust the production plan. The owner sees profitability by customer and by part family in a view that updates weekly without manual effort.

This is where AI tools create the most value for manufacturers. They do not replace decisions. They assemble the data that informs decisions and present it in the context where it is needed. The opportunity for AI in manufacturing is largest at exactly this point: connecting scattered data to recurring decisions.

A data strategy does not require a six-figure investment or an 18-month implementation timeline. It requires clarity about what decisions matter, what data supports them, and a commitment to making that data clean, connected, and accessible. The manufacturers who build this foundation now will have a structural advantage for every tool, every analysis, and every AI application that builds on top of it.