Community Discussion

GiveRank - Increase donor conversion with artificial intelligence

GiveRank is an artificial intelligence that increases lead-to-donor conversion by analyzing leads' online behavior and predicting which leads have a higher chance of converting into donors.

By filling this field, you are confirming that you have already acquired the consent of the concerned party you represent to share this idea on this challenge. 

I am representing W-Mind, a startup that applies artificial intelligence to fundraising

Remember to be user-centric. Think about your user’s needs/problems you are trying to address.

Thanks to the revolution of digital fundraising, organizations today gather a very high amount of potential leads every year through their campaigns. Many fundraisers would love to convert existing leads into donors using direct mail or telemarketing.

However, since leads haven’t donated yet they can’t be segmented with models such as RFM. As a result, understanding the most potential leads to include in a new appeal becomes really difficult. Fundraisers that wish to convert their current leads, have to make guesses or extract random leads lists for their conversion appeals.

Because of how hard it is to understand which leads are ready to convert, issues arise hurting both the NPO and the potential donor. On the fundraiser side, redemption rates get really low (usually below 0.5% according to a series of interviews conducted with a panel of large NPOs based in Italy) and unsustainable. As for the donors, they might hear from the charity too early in the relationship. This could stress them and make them perceive the organization as too pushy.

COVID-19 has further exacerbated the issue of untapped lead pools. The drastic reduction in F2F due to COVID left many organizations deprived of a key source of new donors and revenue. In this scenario, understanding how to better convert online leads into donors, becomes an even more important task for NPOs. 


For whom is your proven concept trying to solve a problem? (Examples of audience could be: parents between 30-50 years old, young professionals who live in the city, millennials, small non-profits, seniors and retirees, etc.)

Non-profits of all sizes, primarily medium and large organizations.

What are their pain points and gain points? What type of data and insight did you use for this analysis? 

We decided to work on the issue of low redemption on online lead pools after hearing about it several times from our NGO clients (all medium and large NGOs based in Italy). We conducted several interviews to further understand the problem and to have the right information to start developing our solutions.

Our clients were able to recruit large numbers of lead from their online properties for a highly sustainable cost. However, when trying to convert these leads into donors through telemarketing or direct mail they encountered unsustainably low redemption rates. This is because since leads haven't donated yet they were not able to segment or rank their rich pools of leads using any standard methods such as RFM (Recency, frequency, and monetary). 

We realized that an artificial intelligence that could drastically improve the redemption rate of leads conversion appeals by ranking leads for the likelihood of conversion would have been of significant help for our target audience.

A value proposition is a positioning statement that explains what benefit you provide for who and how you do it uniquely well. It describes your target customer/donor, the pain point you solve, and why this new solution is better than other alternatives?

  • Our target customers are NGOs of any size that fundraise through individual donors.
  • The pain point our solution aims to solve is that of a low conversion rate on the number of leads that chose to make a donation out of the total number of leads contacted through telemarketing, direct mail, or email. We also aim to fix the problem of donor stress generated by sending appeals to leads that are not ready to donate yet and might perceive a donation request as pushy.
  • GiveRank's is an original algorithm and is the result of three years of work on predictive modeling work for non-profits. We believe that it's accuracy in predicting donors is hardly replicable by competitors that do not have vertical expertise in donor behavior. We use a unique combination of machine learning algorithms (an ensemble in technical terms) to form the final model used by GiveRank, that we have carefully composed in months of trial and error. We also developed a unique data pre-processing and feature engineering procedure that is specifically geared towards the analysis of fundraising data, allowing us to further increase the accuracy of the AI algorithm. 

Please briefly describe the process of piloting testing for your proven concept, and explain why the results have demonstrated its potential. Make sure you have adequate evidence that to demonstrate that your proven concept has addressed the problems/needs you identified.

GiveRank was validated in an experimental setting using a validation set of leads that either converted to donors or never converted to donors in the past. 

In the test, we showed the AI a number of leads and their online behavior. These leads came from a historical dataset and we had information on which leads positively converted into donors and which didn’t. GiveRank was not given access to this information. 

GiveRank was able to identify most of the leads that would eventually convert into donors when shown the leads' online behavior only.

We then run an appeal simulation to compare the redemption rate of a random leads extraction Vs. the redemption rate of a list created by GiveRank. In the first scenario, we extracted a random list of 10.000 leads and calculated the redemption rate. In the second one, we calculated the redemption rate of  10.000 leads extracted by GiveRank. The redemption of the ‘random’ list was 0.4%. The redemption of the list extracted by GiveRank was 2.4% (+600%). 

We estimate that NGOs that implement GiveRank might experience an uplift in conversion (compared to random extraction of leads or manual segmentation) between 300% and 1200% according to a) quality of data b) quantity of data c) time given to GiveRank to learn from the data.



Please refer to the four evaluation criteria for this challenge:

  1. Innovative
  2. Impactful
  3. Replicable
  4. Feasible

1. Innovative

We believe that our concept is strongly innovative since it uses the latest development of new technology (artificial intelligence) to strongly improve a) the way in which organizations engage with their potential donors b) the future revenues of organizations that are willing to implement our solution.

2. Impactful

The concept would be able to generate strong and sustainable future revenue growth for organizations worldwide. Our artificial intelligence should deliver an increase in redemption rate by +300% to +1200% (compared to manual segmentation based on trial and error or random extraction). Results will vary according to the quality and quantity of the organization's data.

By implementing GiveRank across organizations, we could unleash millions of euros of new revenues from untapped lead pools. Fundraisers would have a way to supplement the acquisition gap brought by COVID and the disruption of F2F. They would also create a new sustainable source of revenue for the future.

3. Replicable

GiveRank it’s fully scalable, replicable, and can be easily implemented throughout different organizational sizes, languages, cultures, and geographies. Our algorithm dynamically adapts to understand the unique behavior of local users.

4. Feasible

GiveRank has been already developed and tested extensively (in experimental settings only). Tracking and storing digital behavior data is a requirement to implement GiveRank and we know that not all NPOs are at that stage yet. 

If some leads behavior has already been tracked by the organization, the NGO will be able to move forward with GiveRank in 3-4 months. If the organization has to start tracking leads behavior from scratch, GiveRank would need 10-12 months to be successfully implemented.



Please consider the following aspects of the investment:

  1. Financial resources (in USD)
  2. Time
  3. Tools
  4. Platforms
  5. Other relevant aspects

1. Financial resources

10.000-40.000 USD to set up GiveRank according to the size of the organization. Following the first implementation, the organization would only need to pay a small fraction of the initial cost to allow maintenance, update, and monthly extractions.

2. Time

3-4 months to make GiveRank fully operational if the NGO already stores lead behavior data. 10-12 months if the NGO has to start tracking lead behavior from scratch.

3. Tools

The NGO would need to have a CRM system or marketing automation software in place to track both current donors' behavior and leads' behavior (email reads, opens, clicks; website visits, opens, clicks; etc.). This data can be stored on different platforms or software or sheets. 

4. Platform

GiveRank supports any kind of CRM or database. No particular platform is needed. 

This video cannot be played in your browser. You can try downloading it and playing it with a different player on your device.

Download original video (.mp4)
  • Nov 11, 2020
Comments Connections
You have to be logged in if you want to comment. Do you want to log in now? Log In