This week Dataro is excited to announce the launch of our new machine learning platform for not-for-profits. NFPs can now use this state-of-the-art propensity modelling technology to increase fundraising returns and eliminate wastage.
“Machine learning already touches our lives in so many ways, but NFPs have lagged behind the commercial sector when it comes to adopting this powerful technology,” Dataro CEO Dr Tim Paris said.
“Ultimately, that’s not really in the interests of the organisations themselves or their supporters. Our purpose is to drive more efficient fundraising, and we do that by giving NFPs access to the best predictive tools and technology to help them improve donor engagement and ultimately raise more funds for their cause”
Dataro is already assisting more than half a dozen well-known NFPs with upcoming appeals and fundraising campaigns. “We’ve had great results so far, and we’re really proud of the work we’ve done with organisations like Greenpeace, where our results showed a 10% reduction in donor churn in a controlled experiment,” Dr Paris said.
Dataro is building its service for NFPs around the power of machine learning to provide predictions about what donors are likely to do next. By training their predictive models on the entire history of an organisation’s fundraising, Dataro generates ‘propensity scores’ for every single donor. This allows NFPs to understand and cater to the future needs of their donors.
Dataro’s platform takes the struggle out of complicated data science. It allows NFPs to upload data extracted from their CRM using a simple drag and drop interface, to access advanced analytics and insights, and to download target campaign lists with scores for every donor.
Dataro’s platform already includes on-demand propensity modelling for several common fundraising campaign types, including Attrition Reduction, Regular Giving upgrades, and One-Off Giving (e.g. appeals). The ability to predict the best candidates for Re-activations, conversions to Regular Giving, major giving and bequests are also in the pipeline.
How it works
“Some NFPs are already experimenting with propensity modelling methods, and of course everyone has their own way of segmenting their database in order to choose which donors to include in a particular campaign or appeal. The problem with these traditional approaches is that they typically only consider a very limited set of factors, and individual donors end up being treated like a single group, rather than an individual with their unique background, desires and behaviour,” Dr Paris said.
“Our machine learning approach means we can take into account literally hundreds of different factors and spot trends that are invisible using more simplistic methods. Further, instead of breaking a database into rudimentary segments, our approach is also a lot more realistic because every single donor gets their own propensity score, which recognises the reality that no two people are exactly the same. This results in much better targeting, ensuring only donors interested in your messaging are contacted.”
Dataro CTO Dave Lyndon said that traditional rules-based approaches were the most common method NFPs currently used to select donors for direct mail and telemarketing fundraising campaigns. A rules-based approach may be as simple as ‘contact all donors who have not been called in the last 6 months’, through to much more complex frameworks.
“Rules-based approaches suffer from a number of issues, including that it is very difficult to write good rules, it is almost impossible for a human to write a good rule that appropriately takes into account more than 5 or 6 factors, it is very difficult to evaluate how good a rule is, it is very unlikely that a rule-based system designed for a Christmas appeal will be as effective when used in a tax appeal, and it is almost impossible to write effective rules to deal with continuous data or factors that vary non-linearly,” Mr Lyndon said.
Vs. Rules-Based Approaches
“There are many advantages to a machine learning approach, including that it is possible to learn models that are essentially optimal given the available information, the validation process gives an indication of how effective the model will be and how much its predictions can be trusted, it is possible to understand which features of the model are most useful, models can be trained for a specific moment in time and for a specific organisation, we can utilize data that is categorical, discrete or continuous and it is possible for the model to learn non-linear patterns.”
A traditional barrier preventing NFPs from adopting machine learning has been the price and skill set required, such as hiring a team of data scientists and data engineers. Dataro removes this barrier by offering access to its core platform on a cost-effective subscription model, with no minimum contract terms or cancellation periods.
“We’ve spent the last year building this platform so that it’s easy to use and massively reduces the barriers to entry for NFPs. We want NFPs to give this technology a try, so we decided to make it available in very low risk way by allowing people to simply sign up for a month. We’re confident they will stick around, because ultimately the evidence shows better targeting can significantly improve fundraising returns and reduce wasted costs,” Dr Paris said.