Why process mapping is not enough: from static BPMN to digital twin with BPSIM in property insurance claims

You've mapped your property insurance claims process. Swimlanes and responsibilities are clear, gateways are not causing loopholes, and every stakeholder has signed off on the model. And then a workload spike hits, over 100 claims come in within a week, and your "well-mapped process" has to face a real-world volume.

The BPMN diagram is a structural model. It shows you what steps exist and in what order. It doesn't show you what happens when your usual workload is put under pressure, where the queue will build and which resources will break. Process mapping without simulation gives you a blueprint. And you can't stress-test a blueprint.

This blog post walks you through how to use Cardanit BPMN process modeling and BPSim process simulation to design a property insurance claim that holds up under edge cases and workload spikes. First, we'll see together how to model a standard property and insurance claim in Cardanit BPMN. Then we’ll show exactly what Cardanit BPSim adds when you turn that static map into an executable, simulation-based digital twin.

Why does your static BPMN diagram fall short?

Traditional BPMN models are declarative. They describe what should happen, not what is happening or what will happen under pressure.

Even if you've run process mining on your event logs, what you still get is a snapshot. A record of what happened, not a prediction of what will happen. The moment you decide to redesign a subprocess or shift resources, that snapshot is already outdated. And every change you make to the model creates new unknowns: new potential deadlocks, bottlenecks, wrong resource allocations. And no static BPMN diagram will flag those for you before they hit your live process.

Two problems show up repeatedly in property insurance process work:

• Conformance gaps. There's almost always a disconnect between the ideal model and its actual execution. Static maps don't capture deviations in real time and they rarely account for edge cases like belated first notification of loss (FNOL) submissions or claims that skip a step due to informal workarounds.
• Shadow steps. Loss adjusters spend 40 minutes calling back policy holders to collect missing documents. Claim service representatives have additional workload around cross-checking policy coverage. None of this appears in a BPMN diagram nor is predictable. But all these hidden shadow steps affect your process time and cost per claim.

Simulation doesn't replace BPMN. It extends it. Your BPMN model is the structure. BPSim (Business Process Simulation) is the OMG-standard extension of BPMN that lets you attach simulation parameters — durations, frequencies, costs, resource assignments, and more — directly to your BPMN elements. In other words, BPSim is what makes those elements actionable.

Cardanit BPMN combines BPMN 2.0 business business process modeling and BPSim simulation in one workspace. You can map workflows, run simulations, and test changes without switching tools or losing context. Most teams model in one tool and simulate in another, or don't simulate at all. Keeping both in the same environment removes friction and keeps the process model as the single reference point for every business change. As a digital process modeling tool, Cardanit also handles process documentation alongside simulation, so you're always ready for sharing your process with your stakeholders.

Use Cardanit BPMN to design your ideal process. Then BPSim to test and validate it.

Advantages of business process validation

How to model a property insurance claim in BPMN

A standard P&C property insurance claim, when modeled in BPMN, uses swimlanes to divide responsibility across actors (in our BPMN example model): the policyholder, front and back office with CRS (claim service representatives), the loss adjuster, and the claims settlement office with claims processor. Let’s model this in Cardanit with a property insurance claims process template.

The flow typically runs like this:

1. First notice of loss (FNOL). The customer reports damage, for example, a flood, fire or theft.
2. Policy verification. The front office receives the claim, confirms whether the policy is active and checks whether the incident type is covered. Claims that fall under standard exclusions are flagged or rejected here.
3. Damage assessment / on-site investigation. A field adjuster visits the property, takes photos, writes a manual report and estimates damage costs. For complex losses, an independent expert may be appointed. This case isn't included in the walkthrough model below, but you can map it easily in Cardanit using subprocesses.
4. Review and approval. A claim processor reviews the damage estimate. Claims above a defined value threshold may require secondary approval. Again, this case isn't included in the walkthrough model below, but jump into Cardanit and add it yourself.
5. Settlement and closure. Accounting processes the payment and notifies the customer. The case is filed.

On paper, this is a clean process. In BPMN, it looks manageable. But what about hidden deadlocks in the process design? Hidden costs? Realistic task execution? That's exactly where simulation picks up. Setting up BPSim simulation parameters in Cardanit allows you to transform your static BPMN diagram into a simulation-based "digital twin". You can run it against real inputs to identify bottlenecks and test improvements.

How to build a claim process digital twin with BPSim

A simulation-based digital twin for business processes is an executable replica of your business process model. It’s a calibrated process design you can run against realistic inputs before making any changes to your live process. The goal is to predict process deadlocks and performance and before you commit to changes.

To build one for your property insurance claims process, it’s enough to set BPSim parameters directly on your existing BPMN model.

In Cardanit, switch from the design tab to the process simulation tab and start assigning:

• Trigger time for the start event: this lets you set how many times the process is triggered (100 claims within a week in our example) and the time interval between each trigger (every 10 minutes). A calendar schedule is applied here to reflect a working schedule of Monday–Friday, 8 hours per day.

• Task durations: most tasks have a variable completion rate. In this example, two processing time parameters help describe it:
  • A uniform processing time parameter that takes a variable amount of time and distributes it equally across a minimum and maximum value. You can use this parameter to simulate task duration when you don’t have historical data to start from.
  • A triangular distribution describes a continuous probability distribution that allows you to simulate uncertainty of a task that can require for example, sometimes 1, sometimes 2, sometimes 4 days.

• Resources: set the number of CRS, loss adjusters and claim processors (2 per role in this example)
• Gateway probabilities: set likelihood for a task outcome to occur: for example, the likelihood a claim passes initial triage or the percentage of claims sent to the anti-fraud department.

What simulation reveals that BPMN can't

1. Hidden bottlenecks from resource contention

Your BPMN shows a box labeled formal assignment to the loss adjuster. It doesn't show what happens when 100 claims arrive and you only have 2 adjusters available, for example.

By assigning resources to tasks in BPSim, you can see exactly where their processing time increases because resources are under-allocated. Simulation allows you to push the model further. You can run a catastrophe scenario with over 100 simultaneous claims and the simulation data will clearly show you at which steps your process will get stuck. In this case, the process simulation data output can also help you determine where and how many resources you need to face this unexpected event. By running different simulations, you find out the right ratio of in-house adjusters to external contractors to maintain your target response time for example.

So, if your stakeholders are asking "how many adjusters do we need for next quarter," use Cardanit simulation to give them the answer. Simulation can give you a backed-by-data number to put in front of a decision-maker.

2. Cost and time variability under real conditions

Static maps rely on "ideal path" estimates: “this task usually takes an hour." In practice, it follows a distribution — sometimes 5 minutes, sometimes 50. BPSim lets you apply stochastic modeling: a technique that replaces fixed estimates with probability distributions so each task duration reflects a realistic range of outcomes rather than a single assumed value.

Take the on-site investigation task. Instead of a fixed "4 hours," you can apply a triangular distribution parameter: minimum 2 hours, most likely 4 hours, maximum 3 days. That maximum accounts for what actually happens: scheduling conflicts, site access delays or adjuster unavailability. Run the simulation across 100 claims and the model will show statistical outputs, like queue length, processing time, tasks durations, wait and lag time, resources used and so on.

3. The compounding effect of small delays

A 30-minute task duration in FNOL data ingestion might look harmless in isolation. But simulate it across 100 claims simultaneously and you'll see the policy coverage verification queue building up, pushing same-day assessment scheduling past capacity. BPMN shows the sequence. Simulation shows the compounding cost of that sequence under load.

4. Proving automation needs

Many automation projects start because a BPMN diagram shows a repetitive task. But if that task isn't the actual bottleneck, automating it won't reduce your process time. It will just move the pile faster to the next desk. Simulation proves whether your proposed "to-be" process actually delivers the expected savings before anyone writes a single line of code.

Simulate the current “as-is” and the proposed “to-be” models side by side. The simulation data output will tell you whether the redesign delivers the process time reduction you're targeting before any IT work begins.

Changing an approval routing in production is expensive. If the new flow fails, you lose money and frustrate claimants. Test it in the model first.

5. Calibrating the model against your actuals

A simulation is only as trustworthy as the inputs behind it. Before you run what-if scenarios, calibrate your current “as-is” model against your historical data. Pull data from your claims management system — average task durations, volumes, gateway split ratios — and compare them against your simulation output. Adjust your distributions until the model's behavior matches your historical baseline. Then run your ideal scenarios. Before suggesting process improvements, you need to find out the bottlenecks.

BPMN vs. BPSim: what each one does

How to read your simulation results

Cardanit showcases simulation output through three business process data visualization tools:

• Interactive heatmap. Tasks are color-coded by performance: red for high stress or long duration, yellow for mild stress, green for smooth flow. Applied to the claims process, you can immediately see whether it's the on-site loss inspection or the claim process stages that's driving the process time up. It's the immediate way to spot bottlenecks without browsing through a data table.
• Bar charts. Horizontal bar charts let you compare task-level performance across the process. Sort by processing time to spot which tasks exceed their target duration. Filter by resource to see whether a delay is a process design or resources issue itself.
• Interactive tables. Full simulation results in a filterable table format. Sort the data you need and hide columns that aren't relevant to the decision at hand. This is where you validate whether proposed changes actually optimize your property insurance claims process.

Once you've identified the bottlenecks and tested your what-if scenarios, you’ve got three concrete optimization levers:

• Redesign the workflow — remove, merge, or reorder steps based on simulation data output
• Reallocate resources — adjust staffing ratios or shift schedules to reduce queue times
• Adjust parameters — modify gateway probabilities, task durations, resources allocations and re-run the model to re-validate the impact

If you're curious to see how Cardanit BPSim simulation would pinpoint the bottlenecks specific to your property insurance claims process, start a 30-day free trial.