How do you know you’ve been making a difference with your granting initiatives? That’s what every grantmaker is asking or being asked by their donors, funders and boards. While it’s important to know what you’ve accomplished, it’s even more important to have a solid understanding of how to gather, organize, analyze and share data, in order to ensure your granting activities achieve impactful, measurable results; this is where predictive analytics comes into play.
Any successful predictive analytics program starts with 5 key steps:
2. Develop a Data Understanding. Like most granting organizations, you likely have a large amount of data that you’ve collected over the years. Understanding your datasets is critical in ensuring that you’re properly organizing them for future use.
3. Bucket your Data. Once you know what data you have, you can begin to group your data into different categories. This is an important step because your data buckets will reflect what you want to achieve with your project plan. There are 2 specific variables you need to consider for this type of analytics exercise:
- Discrete variables, where no group of data is more important or valuable than any other, and
- Ratio variables, where data can be assigned a rating that is measurable and puts it into a hierarchy (eg. A is better than B).
4. Explore your Data. Step 4 involves extracting the relevant data from your buckets so that you can identify what will support your project plan and help achieve your goals. Remember, analytics is as much about using the right data as it is the actual collection of it.
5. Create the Data MODEL. The final step is to create a data MODEL. This is where you take a look at the datasets that may be causing some confusion around your results. Predictive analytics is only valuable to the extent that the data you’re using for analysis is the data you should be using to address the questions you’re looking to answer. This last step closes the loop on managing the integrity of your data. As the acronym MODEL suggests, there are 5 key items in the checklist that need to be reviewed:
- Missing - is there any information that isn’t included?
- Odd - the data may be correct but there’s a good possibility it’s not
- Duplicate - are there records of a specific piece of data that appear more than once?
- Erroneous - there is very clearly an error in the data
- Logical Error - is there an identifiable error based on the rest of the data collected?