Emibra Case Study | Before Investing in AI, Find Out Where Your Company Is Losing Margin
In approximately 30 days, Dataside analyzed 19 processes across eight areas of Emibra’s business and identified six priority AI opportunities, five of which are directly related to protecting margins.
Emibra, a manufacturer of paperboard packaging, used Dataside’s AI Value360 to identify where artificial intelligence could create the most value for the business—before investing in any tools.
There was a clear reason for this approach. According to McKinsey’s The State of AI in 2026 survey, nearly nine out of ten organizations already use AI in at least one business function, but only 37% report any impact on EBIT from AI. Using AI has become the norm. Seeing it improve business results is still the exception.
Not every lost margin shows up as a loss. And not every AI opportunity appears in a technology backlog.
The challenge: quality data, but disconnected
Emibra didn’t come asking for a chatbot, an agent, or a machine learning model. It came with a business question: Where can data and AI create the most value in the operation?
Emibra’s problem wasn’t a lack of data. It was that information the company already had was kept separate. The challenge was to connect it and identify where AI could deliver the greatest return.
With more than 50 years in business and now in its third generation of family ownership, Emibra serves brands such as Johnson & Johnson and Aché. It is an industrial company with mature processes and an operation that keeps detailed records. In the words of Marcos Galvão, Emibra’s CFO: “The data side is very good. We even have more data than we need.”
The missing piece was bringing it all together. Budgeted, produced, consumed, and invoiced amounts were recorded in different systems and spreadsheets, with no single source showing the full picture for each production order.
This is a common situation in midsize industrial companies. The information exists, but it doesn’t reach the point where it can inform a decision. Emibra didn’t need to create more data; it needed to connect what it already had.
Not every lost margin shows up as a loss
Money left on the table is margin that should have been earned but never makes it into the company’s results. Unlike a loss, it doesn’t appear in any report.
A loss is visible. It shows up in the financial close, becomes a meeting topic, and prompts an action plan. But an order that ships with less profit than expected, a quote revised because materials weren’t in stock, or a machine changeover that a different production sequence could have avoided may all go unnoticed.
That’s the money AI Value360 set out to find at Emibra. The most revealing finding was that more than half of the identified potential wasn’t an accounting loss; it was margin that never materialized. Companies that look only at losses see less than half the problem.
Not every AI opportunity appears in a technology backlog
The opportunities uncovered during the discovery weren’t on an IT project list. They were spread throughout the operation: in material purchasing, quoting, production planning, and on the factory floor.
None could solve the problem on its own, because Emibra’s margin is lost in the handoffs between departments. What’s missing at one end shows up as a cost at the other. Only a view of the entire operation reveals how it all connects.
How AI Value360 works in practice
AI Value360 is a discovery process that starts with the operation and ends with a map showing where AI can create value, in what order, and with what level of effort. It applies Dataside’s AI Fast methodology. At Emibra, the process followed five steps.
Immerse in the operation. Dataside’s team met with teams across the company for more than 13 hours of discovery conversations and visited the factory floor to see the operation in action.
Size the problem using real data. Emibra shared two years of revenue and margin history. With this data, the conversation shifted from “this is a pain point” to “this is what it costs.”
Prioritize by value and complexity. Each solution was placed on a matrix based on its potential value and implementation complexity. Opportunities with potential gains but lower returns were recorded for a later phase.
Deliver the AI Value Map. The result was an interactive map showing each area’s workflows, the connections between them, the pain points identified, data maturity assessed across the 10 pillars of the DSGOV framework, and a solutions roadmap.
Prove the value before scaling. Where the potential gain is already clear, the solution moves to implementation. Where the estimate depends on real-world data, the next step is to build a prototype before committing to a project.
Example: AI-powered paperboard demand forecasting
At Emibra, paperboard is the largest cost component. When purchasing doesn’t keep pace with demand, one of two things can happen: capital gets tied up in inventory, or the right material runs short and production has to use an alternative board, putting pressure on margins as early as the quoting stage.
AI Value360 connected these dots. More accurate demand forecasting doesn’t just benefit procurement; it protects margins further down the line, in sales. In a backtest using the company’s historical data, the machine learning model indicated a baseline improvement in forecast accuracy of about 10%, with the potential to reach 15% to 20% if confirmed through a prototype.
This is the kind of connection between departments that an isolated analysis can miss—and that AI Value360 is designed to uncover.
The 6 AI opportunities prioritized for Emibra
AI Value360 prioritized six opportunities. Five address margin protection, from material purchasing to the factory floor, and one focuses on generating new revenue.
The data foundation with a conversational agent brings together budgeted, produced, consumed, and invoiced amounts for each production order, showing the margin for each order the next day instead of only at month-end.
The planning optimization agent recommends production sequences to reduce setup changes by simulating each machine’s actual capacity.
The quoting copilot brings actual inventory levels and production parameters into the quoting process. Paperboard demand forecasting improves purchasing accuracy.
The downtime cause classifier reads free-text explanations and identifies root causes. Lead list enrichment gives the sales team qualified prospects and the decision-makers’ contact details.
Three opportunities are moving to implementation, while two will first go through prototyping. The return on each will be measured against a baseline established using real data. When the results are in, Emibra will be the one to report them.
The discovery in numbers
8 areas of the business involved
19 processes mapped from end to end
133 process steps mapped
12 discovery meetings, totaling more than 13 hours of conversations
1 in-person factory visit
10 data maturity pillars assessed
6 AI opportunities prioritized, 5 tied to margin protection
Approximately 30 days to complete
100% of managers aligned on the roadmap
Customer rating: 10 out of 10
What Emibra said about AI Value360
Emibra validated the discovery in an internal meeting attended by all its managers. The point the customer emphasized most was the focus on what would deliver results.
“We were sure there could be gains in some areas, but the returns weren’t as high, so we made a point of not focusing on them. You really focused on where you could deliver the best results for the company.” Lucio Tanikawa, IT Coordinator, Emibra
“All the managers were aligned with what you presented.” Lucio Tanikawa, IT Coordinator, Emibra
“In a short time, you managed to map everything out very effectively. The team captured the essence of it.” Marcos Galvão, CFO, Emibra
Strategic impact
The discovery changed the question Emibra was asking. It went from “Where should we apply AI?” to “Which opportunity will protect the most margin first?” Leadership gained an integrated view of the operation and a roadmap validated by the entire management team.
That’s what separates using AI from seeing it improve business results. Many industrial companies are in the same position as Emibra: quality data, mature processes, and margin trapped between one system and another. AI Value360 is Dataside’s way of finding that value before a single line of code is written.
How much money is your operation leaving on the table?
Comments