5 Brutal Truths About Qlik Sense Architecture (Bad vs Good)

Read time: 2.5 minutes

How the Qlik Architecture Fails and What You Can Do About It

1. You Loaded Tables, Not a Data Model.

• No consistency in the grain.

• Duplicate or messy keys.

• Broken associations.

Fix:

• Start with your design.

• Clean keys and consistent grain.

• Validate your associations before building visuals.

2. Your Load Script Has Become a Dumping Ground.

• Too many joins and transforms.

• Logic is difficult to maintain.

• Performance is poor.

Fix:

• Separate out the responsibilities.

• Pipeline does the cleansing and prep.

• Script does the modeling.

• The app does the exploration.

3. You Created Chaos of Exploration.

• Too many fields and selection paths.

• No direction for users.

• Users are confused instead of having insights.

Fix:

• Intentionally curate fields.

• Design clear navigation paths.

• Limit what users have access to in order to drive their decisions.

4. Your Associations are Unreliable.

• Synthetic keys and circular references.

• Inconsistent behavior of selection on the same data.

• Users do not trust your data.

Fix:

• Have a single version of the truth.

• Maintain a clean key structure.

• Remove synthetic keys.

5. Insights Without Actions are Useless.

• Users explore, but do not act.

• No apparent outcome based on the analysis.

Fix:

• Design your dashboards around the actions that will result from your insights.

• Align insights with actions to deliver an outcome for users.

• Help guide users to what matters.

💡Key Takeaway: 

When the data model is incorrect, it leads to different interpretations of every selection… eventually, people stop listening.