Problem Definition
User Story
As a developer
I want to view how decision rules were applied when analyzing process data
So that I verify that the business rule task performed as expected
Use Cases
- (MH) Download data to evaluate whether DMNs behaved as expected in Excel (SUPPORT-11648)
- (SH) Serve as backup of decision data so it can be deleted (SUPPORT-11648)
- (NTH) Easily view the DMN table associated with a business rule task
- (NTH) Filter process data based on decision data (OPT-5891/SUPPORT-12581)
PM Notes (Why, Customer requests, context, ideal state, assumptions)
Why
DMN is a major feature of our platform and allows banks, telcos, and insurance companies to leverage business domain knowledge in automating decision-making, especially when they're complex. This drives business-IT alignment, and it's popular. About 40% of our customers have filed a support ticket for DMN. However, customers have only opened [6 FRs|project = OPT AND labels = SUPPORT AND labels = dmn] in Optimize, and we don't have decision reports in Cloud so we don't have usage data. One customer said this ticket would resolve the "missing link" between decision and process data.
If we can solve this problem, our customers can better leverage DMN data to improve their process performance.
Complicating Factors
- Complex DMN tables might have over 100 rows
- One customer has 200 DMNs in production with ~50-100 users
- Per-instance decision data is unavailable when the engine's DB is wiped, so we have to store it in Optimize
Workarounds
- The user must click each process instance to see how rules were applied
- If decision instance data is unavailable, the user has to manually recreate the decisions
Ideal State
- This is an incremental step towards OPT-5891, where we enable process instance filtering based on decision rule data
- HOWEVER, the leading solution is to include decision rule data at the flow-node level as peers to variables. Flow-node level data is not available for filtering or visible in the raw data report.
- Business rule tasks can be repeated. How should they be displayed? (e.g. TableName-1, based on comparing timestamps).
- Here's how the existing workflow goes
Solution Definition
Solution Ideas
- Add a column to the raw data report with decision data
Design Notes (Videos, mockups, guidance)
User Scenario
GIVEN I am viewing a raw data decision report
(MH)THEN I am able to see a column for matched rules for decision instances
GIVEN I am viewing a saved raw data decision report
(MH)THEN I am able to download the CSV containing the matched rules
GIVEN I am limited by the number of downloadable records
THEN I can perform analysis on the sample
OR I am able to increase the configuration to allow a higher download
Metrics (if applicable)
- Number of raw data reports