- Believe the hype.
Examples: Visualisation Tools vendors that ‘demonstrate’ how their tool will solve all your problems but never really show you the data wrangling requirement to get those flashy graphs to work. Or how you can distribute insight cheaply. Or how to make it truly dynamic. Or what access to it would cost for a large userbase.
- Listen to Buzzwords.
Examples: Machine Learning. Story telling. Artificial Intelligence. Extract Load Transform. Visualisation. Data lakes. Data swamps. Big data. Data Science. Data Engineering. Again while there are many valuable elements that fit these terms in a healthy Business Intelligence strategy, many are probably not needed or wanted for most fo your decision support requirements.
- Skip the Business Analysts (or misunderstand the role)
Examples: No proper discussion with business groups on what they require. No determination on the companies first and most critical Decision Support need. No analysis of the existing system for design flaws, redundancy and aging, and systematic and technical decay of business rules implemented in crisis.
- Big bang approach.
Examples: Expect the project to run start to finish and deliver everyone’s needs. Common to this is forgetting how long a complete project would take and how much the business through market pressures could change in this time. Also larger consulting companies promote big bang approaches because the paperwork alone keeps them in cream for years.
- Rigid thinking/ acceptance of rigid thinking.
Examples: Not building in organic growth and shrinkage mechanisms in the data core. Not considering the management tools needed to keep the system fresh and adjusted to needs. Thinking the data warehouse will be the same for the next 20 years.
- ERP blindness (lazy thinking, risk averse thinking)
Examples: Assuming an ERP will do everything the vendor promises, including Business Intelligence. Assuming the ERP will perform all the functions of the business as you need. Forgetting that having the same ERP as competitors means you lose competitive flexibility.
Try a different approach.
- Start with understanding your business users, the coalface departments. Define the role data plays in their work. Look for the lowest hanging fruit, what decision support capabilities wins you the most market value.
- Load only the data that knowledge thread requires, staged, transformed, and data marted to meet your users immediate needs. Select the tool they will find most appropriate.
- Add the next most important need, scaling the data warehouse core tables horizontally while adding a new DataMart vertically.
- Keep going with connecting source systems to business user needs based on your original model as defined by the business and horizontally scaling the data warehouse.
The advantage of this approach is numerous, but the biggest is the saving ion time and money. You get reports in days and weeks, not months. You get a core operational data store that is aligned to source and user needs simultaneously. You get an organically managed data system that allows you to kick source systems to touch and replace them with minimal disruption.
We should know. We have being doing this for 15 years. Through a global financial crisis, a recession and a pandemic.