What are the Weaknesses of Directly-Follows Approaches?

Directly-Follows approaches (Directly-Follows Graphs, DFG) build process models solely from which activity immediately follows another. This purely sequential view cannot correctly represent concurrency, loops, and genuine dependencies, and is also sensitive to noise and incomplete data. The result is often inaccurate or misleading models.

Why does this matter?

Because many process mining tools rely on DFGs, they frequently produce cluttered "spaghetti diagrams" and false causal assumptions. Understanding these weaknesses allows you to deliberately choose more expressive methods such as Event Knowledge Graphs and Causal Process Mining.

What does this look like in practice?

If a credit check and a stock check run in parallel, the log shows them sometimes in one order, sometimes the other. A directly-follows graph draws edges in both directions and suggests a loop that never existed. The model looks more complicated than the process — and the phantom loop attracts analysis that leads nowhere.

Not to be confused with

Not to be read as a general objection to process mining: directly-follows graphs are one — admittedly very common — modelling form among several. Techniques such as inductive mining, and object-centric and causal approaches, avoid precisely these weaknesses.