So your company’s leadership has decided that it is time to make better data-driven business decisions. Top management is tired of competitors making moves that seem to beat every clever intuitive strategy devised by your business leaders. You have a group of business analysts tired of collecting data from disparate sources and using basic Excel features to clean the data or worse applying brute force to clean it manually. You have a big responsibility and a tremendous opportunity to give your company a competitive advantage. Here are 3 keys to help you succeed on your project.
Generate a sense of ownership
It is one of the most important contributors to success. The main stakeholders will include business analysts, software architects, system administrators. Engaging them in the steps outlined below is critical
1. The proof of concept exercises to select the right tool
2. Developing the data warehouse and reviewing the data along the way
3. Validating data and building visualizations
4. Rolling out to the business users
A team that feels a sense of ownership will make reasonable compromises, trust that the developers will improve that solution in future sprints and be patient when the system hits inevitable snags such as slowness or dirty data etc.
Keep up the momentum.
There will be several momentum killers that you will face from vendors requesting NDA modifications, to ELT tools that do not deliver promised features, to data that is not clean. Do not let the perfect be the enemy of the good enough. Everyone seems to think that business intelligence (BI) solutions have to be perfect right from launch but a perfect solution that is never delivered is a total waste of time. It is important for leaders to watch the momentum and deliver in 2-week sprints. There’s always a chance to improve and fix. For example, if it will take 30 elapsed days to coordinate with an external party and automate a weekly data feed and 2 days to start manually importing the data, then manually import the data and let your business analysts use it while your developers work on the automation. BI requires several data feeds and the market will not wait for the perfect solution.
Fit the tool to your business analysts' abilities.
Involve all the business analysts in deciding the top features that they need. Conduct a proof of concept (POC) with large volume of your own data in your own server ecosystem. It is interesting how some tools work great during presentations and demonstrations but choke when used in other server ecosystems. These POCs also reveal how much processing power is required for the BI tools. Have the vendors conduct hands-on workshops for your analysts. It is interesting how much user experience design matters for success. Some tools may not have advanced features that a few genius employees desire but are easy to use for beginners or those with average skills. The cumulative effect of many employees making data-driven decisions is better for business than for a few genius employees using advanced features.
Tableau, looker, powerbi, knowi and thoughtspot are easy to use, performant and competitively priced data visualization tools. Snowflake, redshift, and google bigquery are excellent data warehouse options that all serious teams should evaluate in 2018. There are many many options in the BI market and this is by no means a full list. Please do your due diligence and have fun researching the BI products. It is a tremendous learning experience, almost every product has knowledgeable, helpful pre-sales engineers and sales teams who help organize demonstrations and workshops.