Why Your AI Provider Needs To Help You With More Than Just Technology

Artificial intelligence offers enormous potential to optimize operations and increase revenue for your nonprofit organization. But with the realization that AI is the future of fundraising, comes mass panic and worry about what kind of shape your data is in. Because in order for Artificial intelligence to work, there are a few requirements that are 100% necessary. AI cannot work without accessible, high quality, labeled, granular data, and you should be very wary of anyone who tells you differently.

Oftentimes nonprofits capture data to meet compliance requirements and regulations, for this reason there is a lack of focus on quality when it is recorded. The landscape is changing though, and as philanthropy evolves, data has become paramount to engaging the next generation of donors.

Just because you have purchased or are using a database, it doesn’t mean that the data is being stored in the way that is, or will be most useful. Collecting and recording data properly is still largely dependent on human actions and therefore requires a behavioural change in order to improve it. This behavioural shift is not something that any technology can do for you, but it is something that the right technology can help you to change. The reality is, that if you think you aren’t ready to start embracing AI, you’re probably never going to be. In order to take your organization and your data to where it needs to be, you need to stop planning and start doing.

This doesn’t mean that you should stop thinking or planning strategically about your long term plans for digital transformation, but not taking action on AI now, is a sure path to stall success.  It’s a bit like planning a whole health transformation, complete with diet changes and a gym regime, but then putting it off again and AGAIN...because you are not quite ready to do it perfectly. Whereas the people who take baby steps and make small changes to their diet and exercise routine over time, are the people who actually change their lives successfully and maintain success in the long term.

AI providers cannot offer a magic solution that actually works if they are not willing to get down and dirty with your data, and the people who keep your organization running. Any AI vendor that is looking out for your long term success, will develop a relationship with your team and your data from the ground up. Factors having to do with data quality and accessibility like completeness, consistency, accuracy, validity and timeliness, are things that a real AI partner can help you with, because it is necessary for your success AND theirs.

The more labeled data and the more granular your data set, the better for AI. If you can find a partner that will help you develop a more granular and labeled data set, that alone is invaluable. Granular data refers to how detail-oriented a single field is, the advantage is that it can be molded in any way required by a data scientist or analyst. Granular data can be aggregated and disaggregated to meet the needs of different situations, and it can be easily merged with data from external sources, effectively integrated and managed.

Labeling typically involves taking a set of data, and augmenting each piece with meaningful and informative tags. For example, labels might indicate what type of action is being performed in a video, or whether or not a hyperlink directs to a homepage or a donation page. This kind of data is significantly more expensive to obtain, but also much more valuable.

It is difficult to get changes in data discipline to stick across teams, likely because the people who have to execute these changes do not see immediate value to their important work. Data discipline is extremely valuable and should not be framed as an administrative task. A focused project where the outcome is directly correlated to a change in practice, can allow team members to see the importance of that work and in turn, be more motivated to learn how to do it better. Once these changes are established as necessary for success, the virality will scale itself throughout your organization.

While developing best practices in data discipline is a necessary first step towards applying AI, unfortunately, most of the advice on how to introduce it into your institution suggests top down organizational overhauls as the first step.  But it does little good to create a new enforcement structure, committee, or design a new work process, if you leave the people behind. We can’t fail to appreciate the human psychology around change, when ultimately what we are trying to do is help employees embrace, adopt, and utilize a change in their day-to-day work.