Time flies! In between balancing projects and vacations, the process mining summer continues. Microsoft has announced the general availability of their process mining product, Fluxicon organized an interesting process mining café on data preparation challenges, and we’ve developed most features highlighted in last month’s newsletter.
Addressing Data Preparation Duration
This time, let’s dive into the most common and challenging question we receive: how long does data preparation take in Konekti? The honest – and daunting – answer is: it depends. It varies based on scope, IT architecture, prior knowledge and more, making quantification difficult.
In general, data preparation often takes up to 80% of a process mining project’s time.1 That’s an enormous chunk, which is why Konekti’s vision is to make the data preparation for process mining 10x faster. As an example, for a process mining project that initially demanded 100 days, that means a data preparation time reduction from 80 days to 8 days, leading to a complete project turnaround in just 28 days – a substantial leap forward.
This vision drives us to continually develop Konekti. However, what could be more relevant to you is: what can Konekti do for you right now?
The importance of speed
In process mining projects, time-to-value is often as crucial as quality. Swiftly delivering results to your (internal) customers enhances project effectiveness. Furthermore, faster data preparation enhances scalability: spending less time on individual data preparation pipelines enables greater productivity. Speed matters, indeed.
The time needed for data preparation hinges on several factors. Let’s examine three key scenarios:
- Create an Event Log from Scratch: Creating an event log without prior knowledge on which process data to combine and how.
- Use a Template for a New Event Log: Leveraging existing knowledge or external resources for event log creation
- Refactor and/or Transfer Data Prep: Improving data preparation efficiency or transferring the data prep to another tool (for instance for migrating between process mining tools).
For each scenario, we compare the roughly estimated data prep time with and without Konekti. These estimates are derived from our experience as process mining consultants and discussions with other experts. Let the comparison begin!
Konekti’s Real Impact Today
The estimates are summarized in the picture below. We estimated that utilizing its existing features, Konekti accelerates event log creation to twice the speed. Starting from scratch saves you approximately 200 hours, and when using a template for a new event log, you can reclaim about 60 hours. In the case of refactoring, Konekti can increase your speed by a factor of 4, reclaiming 120 hours per event log.
To put these numbers into perspective: with Konekti, you transform from a turtle to a horse when creating an event log from scratch, from a default gazelle to a cheetah when using a template, and from a lion to a cheetah when refactoring data prep.
While there are definitely caveats to consider2, several valuable insights emerge:
- Konekti’s power grows with scale. For large-scale process mining, where multiple processes are mined, Konekti’s benefits add up, fortifying your Process Mining Center of Excellence.
- Templates are pivotal. A template customized for your precise process and IT system is a substantial time-saver. Frankly, a high-quality template can expedite data prep more than starting from scratch within Konekti. Notably, Konekti can amplify the template’s time saving, even when the template isn’t native to Konekti.
- The most intriguing scenario is one yet uncharted: what if templates become available within Konekti? While this requires a larger user base, be poised to be as fast as a falcon when the time comes!
That’s it for this edition! We haven’t delved into the details of how Konekti delivers these time savings. Eager to discover how Konekti can expedite your use case? Connect with Lotte for a special discount for new users mentioning their spirit animal. You can also book a demo with Maarten!
1 That’s the often-used rule-of-thumb. See for example Deloitte’s Global Process Mining Survey.
2 The numbers are based on five years of experience, but it remains speculation. Also, focusing solely on the time reduction for setting up the data model, foregoes other benefits of Konekti, like the time savings during the maintenance of the data model (perhaps impacting scalability even more). And, perhaps stating the obvious, there’s never “one” number for how long it takes to create an event log. It’s highly depending on the systems you try to mine and how it stores its process data, the complexity of the process you’re mining, how much background information you already have, and, of course, the tool you use (hello Konekti!).