What is data integration worth to a manufacturer? What studies measured

No independent study has measured data integration on its own. The closest is a 2021 study of over 30,000 manufacturing plants that measured the use of predictive analytics, which depends on joined data. Plants using it were about 2.87% more productive than similar plants, or about 1.45% after controlling for management practices and other factors. The authors sum it up as "productivity benefits of 1 to 3 percent on average". The gain appeared only in plants with at least one of three complements: substantial IT capital, educated workers, or production designed for high flow efficiency. That, they write, explains "why some firms see no benefits at all".

What the usual answer says

The usual answer cites efficiency, faster decisions and cost savings, with return on investment figures. The top result for this question is a December 2023 summary of a Forrester study commissioned by AWS, which sells data integration services. It modelled "a composite organization within state government", built from interviews, and put its return at "an ROI of 33 percent" over five years. Forrester built it from interviews with organisations that had invested in data integration, so it rests on what adopters reported. Independent studies measure something narrower.

What the 2021 study measured

Brynjolfsson, Jin and McElheran, writing in Business Economics, worked with a national statistics agency on a mandatory survey with a response rate of 70.9%. They linked plants' reported use of predictive analytics to their productivity over 2010 to 2015. The authors' accepted manuscript was read in full.

Adoption was already common: "More than 70 percent of our representative sample adopted some level of predictive analytics as early as 2010." The benefits "increase with frequency of use". The authors read their results as support for claims that predictive analytics "can substantially boost performance", with a condition attached. The authors argue the link is causal, since "performance increases only after plants adopt predictive analytics, not before".

The central finding is the condition. Productivity gains are "almost entirely limited to workplaces that have high levels of accumulated IT capital", a large share of educated employees, or "high flow-efficiency manufacturing processes". And "the returns to predictive analytics are much higher in efficiency-focused contexts than in flexibility-focused production environments".

Reading it through a data lens

In our view, the study describes integration as a precondition. Joining the scale system, the production system and the ledger makes better decisions possible. The measured value comes when decisions change: a schedule built on joined order and inventory data, or a log purchase priced on joined recovery and sales data.

The efficiency finding may matter for timber. Our reading is that continuous processes, such as panel lines and paper machines, sit closer to the paper's efficiency-focused plants than job-shop operations do. The study doesn't test that.

The study suggests a practical test. Before a project starts, name the decisions that joined data will change, such as log pricing or weekly scheduling. Record the current result for each, then measure it again a year after go-live. If none has moved, the project hasn't yet produced the kind of gain the study measured.

The strongest objection

For a finance team, integration pays in ways a productivity regression can't see. Fewer hours reconciling systems, a faster month-end close, errors caught in invoices and accruals, and inventory counted once rather than argued over are cost and control effects, not productivity effects. The 2021 study says nothing about them, and neither does any independent study we opened. They are worth counting separately, before and after, in hours and in errors found.

What the studies can't show

The 2021 study measures analytics use, which needs joined data but isn't the same thing. Its averages hide spread: plants without the complements saw little or no gain. Its data cover 2010 to 2015, before today's AI tools, so current returns may differ in either direction. A single operation should treat the figures as a guide to what drives the gain, not a forecast of its size.

When it doesn't apply

Integration done for compliance, audit or regulatory reporting has a different payoff: avoided penalties and time saved. Very small plants with one system and one decision-maker have little to integrate.

Quarri for finance and strategy teams is built for the people who close the month, explain the margin and answer the board.

Quarri is an AI-native data platform for the timber supply chain. It connects buying, production, sales and inventory for forest management, sawmill, wood products and pulp, paper and packaging operators.