Volatile raw material prices, disrupted supply chains, and rising sourcing costs: R&D departments are under pressure. New tariffs, geopolitical tensions, and supply bottlenecks are shortening planning horizons and rendering cost calculations obsolete faster than they can be completed. Price fluctuations of 10, 20, or even 50 per cent within a matter of weeks are no longer the exception. They are operational reality. Yet the goal remains unchanged: high-quality products, fast development cycles, and costefficiency.
What is frequently overlooked: one lever for cost reduction sits inside the laboratory. R&D has significant influence over its raw material consumption by cutting unnecessary iterations. The extent of that reduction depends on how systematically the team works with its data.
The hidden cost driver: unnecessary experiments
In formulation development, every experiment carries direct costs. Raw materials, laboratory time, instrument capacity, and documentation overhead add up quickly. The indirect costs are often underestimated: the time spent searching for data before a trial begins, the opportunity cost of an occupied lab, and the delays in time-to-market caused by every unnecessary experiment.
When experiments are repeated because of incomplete baseline data, those costs multiply without any additional gain in knowledge. The calculation is straightforward: cost per experiment, multiplied by the number of experiments per project, multiplied by the proportion of avoidable repetitions, multiplied by the number of active projects per year. Reducing that figure delivers structural savings.
Experiments are repeated because prior work isn’t accessible. Knowledge is lost because it was never collected in a structured form. Conventional laboratory information management systems (LIMS) only partially address this problem: they manage samples and measurement results, but are rarely designed for the interconnected, iterative nature of formulation development, where raw materials, formulations, process parameters, and experimental history are deeply linked.
Resource waste in the laboratory also has an ecological dimension. Every unnecessary experiment consumes not just money, but also energy, water, and raw materials. Efficiency and sustainability are, in this case, two sides of the same coin.
The silent risk: knowledge that leaves the company
There is another factor that is frequently underestimated: generational change in R&D departments. Experienced developers and laboratory managers who retire often take undocumented expertise with them. They were the go-to people for questions about material selection, process parameters, or why a particular formulation was rejected ten years ago.
This experiential knowledge is especially valuable for raw material substitution. Which alternatives were explored in the past? Which worked, which did not, and why? When these answers are not documented, teams are forced to start from scratch with every new supply crisis.
The pragmatic way out of this dilemma is to capture knowledge directly at the source. For every raw material, observations, handling notes, and process experience can be recorded in a central database. The same applies at the experiment level:
- What was observed with this sample?
- What behaved differently from what was expected?
- What conditions were unusual?
Structured notes at the experiment level transform raw data into structured knowledge and create the foundation for resource-efficient development.
When a raw material disappears: reaction instead of control
Beyond day-to-day resource consumption, there is a second and more acute problem: the loss of a raw material. Disrupted supply chains, whether triggered by geopolitical conflict, new tariff regimes, or production bottlenecks at key suppliers, can put an entire product line at risk.
A typical scenario: a supplier announces that a raw material will no longer be available from the third quarter onwards. The development manager recalls that a substitution project was underway two years ago, but who worked on it and where the results are stored is unknown. Experimental data spread across Excel files, the test report as a PDF in a different folder. After several hours of searching, the team starts over.
What follows is a laborious, error-prone analysis. Which formulations are affected? How many products depend on this single raw material? What alternatives exist, and have any of them ever been tested?
Without a structured data foundation, this analysis is painfully slow. With the right tools, the full picture can be made visible within minutes. A targeted search immediately identifies all formulations containing the affected raw material, making the full extent of the problem quantifiable.
Three questions every R&D team should be able to answer immediately in a crisis
In a raw material crisis, structured data delivers what nothing else can: fast, reliable answers to three critical questions.
First: which formulations are affected? A central raw material library linked to formulations, process standards, and experimental results delivers this answer in seconds rather than after hours of manual searching.
Second: what has previously been tested with this raw material? Targeted filters in a structured experimental database show not just whether something was tested, but with what result and under what conditions. This knowledge exists in most R&D departments. Without a structured data foundation, however, it is inaccessible.
Third: how do we capture what we are learning right now? What is tested under crisis conditions is often more valuable than dozens of routine experiments. A raw material pushed to its limits reveals its behaviour under real-world conditions. These insights should not be lost.
From risk analysis to alternative formulations
Data transparency is the prerequisite for the next step: developing concrete alternative formulations. This is where AI enters the picture. An AI assistant built on structured formulation data, defined product requirements, and the team's own experimental history can suggest possible substitutions based on established formulations. The output is a prioritised selection of alternatives ready for targeted testing.
Teams that continuously maintain their data foundation do not need to start searching for alternatives when a crisis hits. They can already be testing potential replacement materials before a supply gap occurs and are ahead of companies that only react once pressure has already been built.
The prerequisite: Structured data comes first
All of the scenarios described above share one requirement: a reliable, structured data foundation. Teams managing formulation data in Excel spreadsheets, siloed ERP systems, or handwritten laboratory notes are neither in a position to respond quickly nor to plan with confidence. Even conventional LIMS systems reach their limits here. They are designed for sample management and compliance, not for the interconnected search across raw materials, formulations, and experimental history that becomes decisive in a supply crisis.
A central database that integrates all of these data sources is not an optional add-on. It is the foundation for data-based decision-making and the prerequisite for AI capabilities to deliver usable results.
The sequence matters. AI in R&D works in three stages:
The first, and most important, is the data foundation. Data must be structured, linked, and searchable; without this, AI cannot produce reliable outputs. The second stage is assistive AI: search, analysis, and support for documentation. The third stage is predictive AI, where AI-assisted experiment planning achieves better results with fewer experiments than conventional trial-and-error.
Those who skip the first stage will find it difficult to reach the third. The organisations using AI productively today structured their data first. Those who establish structured data processes now will be able to respond faster in the next crisis and evaluate alternatives with greater precision.
What this means for R&D departments
Teams that want to actively manage resource consumption, quantify raw material risks, and evaluate material alternatives need:
- A single, centralised view of all formulations, raw material dependencies, and experimental results.
- Connectivity between formulation data, product requirements, and historical data.
- AI capabilities built on this data foundation that deliver concrete, traceable suggestions for action.
This is not a long-term vision for the future. It is an infrastructural prerequisite for modern R&D. The more volatile raw material markets become, the more critical the ability to make fast, informed decisions on a reliable data foundation.