Every trial is a bet: a new chemical program, a furnish change, a different setpoint. TrialMiner compares the trial period against your own baseline across every variable you track, links process changes to quality outcomes, and returns one standardized report your team can decide on.
The problem
A trial only pays off if you can prove what it changed. In most plants that proof is assembled by hand: historian exports, spreadsheet passes, a slide deck for the review meeting. By the time the verdict arrives, the engineering hours are spent and the result still gets argued.
After the trial ends, an engineer exports the data and starts cleaning, aligning, and charting it by hand. Assembling the PowerPoint often takes longer than the trial itself, and it lands on the desk of exactly the person who has the least time to spare.
One engineer compares averages, another filters outliers differently, a third picks a different baseline window. Same data, different conclusions, and the trial review becomes a debate about method instead of a decision about the result.
Without a standard record of what was tried and what it showed, trial knowledge lives in inboxes and memory. Teams re-run experiments that already failed, and pay for the same answer twice.
How TrialMiner works
TrialMiner needs three things from you: a data export, the trial window, and the baseline window to compare it against. The statistical work, the outlier screening, and the report assembly run automatically from there.
The analysis starts over from scratch with every trial.
The same evaluation, standardized and automated.
Inside a trial run
If your historian can export a CSV, you can run a trial through TrialMiner. There is nothing to install and no connection project to schedule.
TrialMiner works from a CSV export of process and quality data covering both time periods. Pull it from your historian the way your engineers already do and drop it in. The format check flags anything missing or malformed before the analysis starts, so bad inputs surface in minutes rather than in the review meeting.
Upload the export
You define what counts as the trial and what counts as normal running: trial start and end, baseline start and end, on the same machine. Filter to a specific grade and let outlier screening exclude abnormal stretches, so the comparison reflects how the process actually ran rather than the noise around it.
Two windows, one machine
Every variable is compared across the two periods and tied back to the quality outcomes you care about, from strength to consistency to the cost drivers behind them. The report lands in the same format every time, with an executive summary up front, so a mill manager and a process engineer can read the same document and reach the same conclusion.
Trial vs. baseline (illustrative)
What the report answers
A trial report is only useful if it settles the decision. TrialMiner's analysis is built around the questions a trial review actually has to answer.
The report links process variables to quality outcomes across the two periods, so you can see whether strength, consistency, or any other tracked metric actually changed, and by how much.
For chemical program trials, the comparison shows whether the new program held quality at a lower dose. Chemistry is one of the fastest cost levers in continuous manufacturing, and a trial is where a supplier’s claim meets your data before that program gets bought at every machine on site.
Statistical evidence replaces gut feel. When the report shows a real effect, you roll the change out plant-wide with confidence. When it doesn’t, you’ve avoided committing capital to a change that never worked.
Beyond a single trial
The report isn't just a verdict on one experiment. Standardized evaluations compound: they build a history your team can check before anyone proposes the next trial.
Each report is stored with the trial’s data, windows, and notes. What was tried, what it showed, and who ran it stays on record instead of leaving with the engineer.
The same evaluation framework runs on any machine, grade, or plant, so results are comparable across your operation instead of shaped by whoever did the analysis.
Trial findings tell you which levers matter. For plants running the ProcessMiner platform, those findings point directly at the setpoints and programs worth optimizing every day.
Built for
TrialMiner is a standalone product. It doesn't require the ProcessMiner platform, a historian connection, or a deployment project, which means it fits anyone whose job depends on a trial's answer.
Evaluate a chemical program, furnish change, or setpoint change without giving up a week to spreadsheets. Drop in the export, set the windows, and spend your time on the decision instead of the charting.
When you run a trial at a customer’s machine, the readout decides the contract. A standardized, statistically grounded report is a stronger close than a slide deck your customer has to take on faith.
Paper, packaging, nonwovens, food and beverage, wastewater: if the process runs continuously and the trial has a start and an end, TrialMiner can evaluate it.
At a glance
Put us to the test
The fastest way to judge TrialMiner is on a trial your team has already evaluated by hand. Send the export, set the windows, and compare our report against the conclusion that originally took weeks.
Already seen TrialMiner in action? If you're ready to start a pilot, click below.
Request a PilotIf you can export a CSV, you're ready.
See it on your process
See TrialMiner compare a real trial period against baseline, then explore how the rest of the platform keeps the gains once you roll the change out.
A standalone product that works from a data export, no platform deployment required.
Questions, answered
A CSV export of process and quality data covering the trial period and a baseline period, typically pulled from your historian. A sample CSV and format guide are built into the product. No software installation and no historian connection are required.
You set four dates on the same machine: trial start and end, and non-trial (baseline) start and end. You can filter to a specific grade, and outlier screening lets you exclude abnormal data before the comparison runs, so the result reflects normal running conditions.
Hours from upload to finished report, instead of the days or weeks a manual spreadsheet-and-slides evaluation takes. Once the file is uploaded and the two time periods are set, the statistical work runs automatically.
No. TrialMiner is a standalone product that works from a data export. If you already run the ProcessMiner platform, you can analyze trials on exports of the same data your live deployment collects.
Any A-versus-B question on a continuous process: a new chemical program, an additive or furnish change, a setpoint or equipment change. If you can name the two time periods, TrialMiner can compare them.
Process engineers and technical teams at continuous manufacturers, and the chemical and additive suppliers who run trials at customer machines and need a defensible readout of how their program performed.
Yes. A supplier running a trial at a customer’s machine can upload the same data export the customer’s engineers use, generate a standardized report, and bring a statistically grounded readout into the contract conversation instead of a slide deck the customer has to take on faith.
Get in touch
Whether it's TrialMiner, your process, or a trial you'd like to see analyzed, let us know.