Automated Data Preparation

Skip the hardest partof Process Mining

The Process.Science Data Preparation Tool turns raw data from SAP, Microsoft Dynamics 365 and your own databases into analysis-ready event logs. No data engineering project, no hand-written ETL. Pick a ready-made process, connect your system, and every scheduled run delivers a complete event log to Power BI or to Process.Science Intelligence.
See how it works
Speak with an Expert

Everything before the analysis,already solved

Fast to set up, flexible with your systems and serious about event log quality, because everything you analyse afterwards stands on it.
Overview of a Purchase-to-Pay pipeline in the Data Preparation Tool with run history and activities
Overview of configured connections to source and target systems in the Data Preparation Tool
Overview of configured pipelines in the Data Preparation Tool
Form for creating a new pipeline in the Data Preparation Tool with a selection of predefined processes
User interface displaying a list of custom activities to include in a transformation, with checkboxes marked for each: Application Submitted, Document Received, KYC Check Completed, Credit Score Calculated, Manual Risk Review, Additional Documents Requested, Offer Created, Offer Sent, Offer Accepted, Offer Declined, and Contract Signed. Each activity has an edit and delete icon. On the right, a pipeline navigation with steps: Source & Sink, Processes, Activities, and Column Mapping. Buttons for Back, Cancel, and Save are visible.

The 80/20 problem, reversed

In most process mining projects the majority of the effort goes into finding and preparing data. Templates and automatic enrichment move that work to the start of the project, where it happens once.

The system maze, already mapped

ERP, CRM and ITSM systems hold tens of thousands of tables. We have decoded the relevant ones for the processes most companies start with, so nobody has to find out which field carries the timestamp.

More than case, activity, timestamp

A bare event log is not enough. Every run arrives enriched with master data, durations, variants, costs and automation flags, so the analysis can start instead of the modelling.
The Process.Science formula

Less data engineering,more process analytics

Manual SQL scripting

Works, until the source system changes. Hard to document, hard to hand over, and one schema update can break the whole logic

Heavy ETL suites

Built for data warehousing, not for process mining. They move data reliably, but they do not know what an activity, a case or a variant is.

Process.Science Data Preparation Tool

Built for one job: turning operational data into event logs. It knows the processes, enriches them with master data, and delivers the same output structure for every process and every source system.
Ready for analysis? Choose target system:
Power BI Suite →
Process.Science Intelligence →

Three steps from raw datato a finished event log

Connect your system, choose your process, put it on a schedule. From there, every run delivers the same complete event log.

01. Connect

Choose your system: Microsoft SQL Server or Azure SQL, PostgreSQL, MySQL or MariaDB, Microsoft Dynamics 365 or a simple CSV upload. Enter the connection details, test and save, then pick where the results should go.

02. Choose your process

Start from a ready-made template such as SAP Purchase-to-Pay and select the activities you care about. The templates already know which columns hold the case, the activity and the time.

03. Run it, or schedule it

Start the pipeline once, or run it hourly, daily, weekly, monthly or on your own cron expression. You can also trigger a run from your own tools through the API.
Out-of-the-box setup

Proven templates, open to the last line of SQL

Don't start from scratch. We have decoded the table structures of the systems most companies start with, and every template stays yours to change.
Ready for your core processes:SAP P2P, from purchase requisition to invoice clearing. SAP O2C, from quotation to billing and payment. Microsoft Dynamics 365 P2P and O2C. For SAP P2P there is also an object-centric variant, covering requisitions, orders, goods receipts and invoices and how they relate to each other.
Nothing is locked:Switch activities on or off, edit the SQL behind each one, add your own activities and preparation steps, and adapt everything to your system with parameters. If your process is not in the catalogue, build it from scratch.
Written once, runs everywhere:A template is written once and runs on every supported database, because the tool translates the SQL into each system's dialect.
Template catalogue in the Data Preparation Tool with ready-made SAP Purchase-to-Pay and Dynamics 365 processes
Enriched event log from the Data Preparation Tool with case, activity and timestamp plus throughput time, variant and cost
What you get

An event log that is ready to analyse

A typical event log has three columns: case, activity, timestamp. That is where ours starts.
Enriched, not bare:Every case arrives with its throughput time, variant, loops, cost and automation rate. Every step carries its duration, waiting time and rework flag. Case attributes such as vendor, plant or payment terms travel with it and variant filters are calculated in advance.
Two zoom levels, two languages:Switch between the detailed view („Record Goods Receipt") and the management view („Goods & Invoice Receipt").
One structure for every process:Purchase-to-Pay, Order-to-Cash or a process of your own, from SAP, Dynamics or any database: the output tables have the same layout every time. Reports and dashboards work across all of them without per-project rework.
Automated quality assurance

Nothing half-finished ever reaches your reports

Dirty data does not announce itself, it just distorts the dashboard. The Data Preparation Tool is the gatekeeper in front of your analysis.
Complete or not at all:A new event log goes live in a single switch, only once it is complete. Your analysis never reads a half-written log and a failed run leaves the previous version untouched. Outdated tables are removed automatically.
Robust by design:If one activity fails, you get a warning and still receive everything else, instead of losing the whole run.
Every run is traceable:Full run history and logs, with event counts and durations per activity, so you can see what a run produced and what changed since the last one.
Quality assurance check in the Data Preparation Tool validating an event log before it goes live
Run history in the Data Preparation Tool with event counts and durations per activity
Built into every run

Every event log arrives with a first finding attached

While the tool prepares your data, it already sees which cases behave differently. It flags them and names the factors they have in common.
Problem cases, marked automatically:Slow cases above your target or the slowest quarter, cases containing rework loops, cases with a low automation rate, and violations of process norms across five rule types, with a default norm included for SAP Purchase-to-Pay.
Influencing factors, not guesses:The tool scans every case attribute and ranks what stands out among the problem cases, with effect size, case counts and statistical significance. For example: cases with one particular vendor miss the throughput target 3.2 times as often as the rest.
Where it leads:The findings land next to your Power BI tables and in Process.Science Intelligence. There you follow them through the process graph, compare populations, and keep what holds up as an insight with its evidence.
What's next

Describe your process, and let the tool find it

We are bringing the pattern-recognition expertise behind our IoT Miner to data preparation. You describe in one sentence what your data is about, for example „procurement, from purchase requisition to invoice payment", and the tool proposes process candidates with real activities from your own tables.
Checked against your data, not guessed:Every proposed activity is verified against what is actually there. Do the tables and columns exist, are there enough matching rows, do the timestamps look real, does the activity share case IDs with the rest of the process?
It asks only where it is unsure:Each activity carries a confidence score built from several signals, and you refine the result in a conversation before turning it into a runnable pipeline.
It learns from your experts:What your team confirms is remembered, so the next discovery on the same system starts from a better position.
Automated process discovery in the Data Preparation Tool proposing activities from your own database tables
Connections

Read from your systems,
write back where you work

A source is where the tool reads from, a sink is where it writes to. Every database connection works in both directions.
Diagram of source and sink connections of the Data Preparation Tool between databases, Dynamics 365, Power BI and Process.Science Intelligence
Missing something?
Talk to us →
The perfect setup

Built for yourPower BIandIntelligence

The Data Preparation Tool is the engine room of your process mining. The event logs it produces flow into the system you work in, without an extra modelling step in between.

Power BI integration

The output tables have the layout our Power BI report templates expect. Your event log lands in your own model, next to your existing dimensions, and the dashboards work without per-project rework.
Discover the Power BI Suite →
Automation rates by activity and over time in Process.Science

Process.Science Intelligence

A finished run becomes the data set your Intelligence projects work on, and refreshes the analysis automatically. From there you have Setup, Analysis, Insights, Chat and Alerts on top of it.
Discover Process.Science Intelligence →
Process.Science Intelligence interface with Setup, Analysis, Insights, Chat and Alerts
Get started

See your own event log, not a demo

Connect one system, pick one process, and see what comes out. We do it together on your data, so the result is your process and not a sample data set.
Runs on your own systems and databases
Ready-made templates for your core processes
No hand-written ETL required
By sending the request, you agree to Process.Science’s Privacy Policy
Thank you! Your submission has been received!
You will receive an E-Mail with a confirmation Link.
Oops! Something went wrong while submitting the form.
Any more questions?

FAQ:Technology & Architecture

The answers IT admins and data engineers look for first
Which source systems are supported?

Microsoft SQL Server and Azure SQL, PostgreSQL, MySQL and MariaDB directly, further databases via ODBC, Microsoft Dynamics 365 through the Dataverse Lake export, and CSV upload for several tables at once. SAP is read through its database, which is why the SAP templates work independently of your SAP release. Results can be written back to any of those databases, handed over as data files, or sent to Mavim.

Does data preparation have to take place in the cloud?

No. The tool reads from and writes to your own databases, so the data stays in the environment you operate. Passwords and access tokens are stored encrypted, access is organised by workspace, and runs scale out across workers when volumes grow.

Is the tool a replacement for Power Query?

It is not a general-purpose ETL tool and does not try to be. Power Query and similar tools move and reshape data without knowing what a process is. The Data Preparation Tool is built for one job: turning operational tables into an event log, with activities, cases, variants and master data already in place. You can keep using Power Query for everything else.

How long does the first event log take?

Once the connection is in place and you have chosen a template, the first run is a matter of minutes rather than weeks, because the mapping work is already done. What takes time is agreeing which activities belong in your process, and that is a conversation with your own people, not a data engineering task.

What happens when our source system changes?

A template is a set of SQL steps you can read and edit. If a field moves or a table is renamed, you adjust that step instead of rebuilding a pipeline. Every run is logged with event counts and durations per activity, so a change in the source shows up as a change in the numbers.

Can we build a process that is not in the catalogue?

Yes. You can start from a template and add your own activities and preparation steps, or build a custom process from scratch. In both cases the output has the same structure as every other event log, so your reports keep working.