Are you throwing away your most valuable fundraising data?

Auckland Grammar School Case Study

Amanda Stanes OStJ CFREDirector of Advancement helping build Auckland Grammar School’s future

In 2017, Auckland Grammar School’s Development Office transitioned  to Potentiality as its CRM system. The reason why? We wanted to integrate all our various connections/touch points with our Old Boys, past and current parents and donors to develop a much more informed and nuanced picture of how engagement linked with philanthropy.  We instinctively knew, and research has borne out, that the more engaged a person is with your story, the more likely they are to donate. Up until then, we had separate systems for event management, eDMs, emails, payment portals and donor management. Using separate systems required significant amounts of uploading and de-duping of data.  Not only was it a slow and onerous process which wasn’t getting us any closer to finding out how we were tracking with donor acquisition and conversion, paying multiple licenses was expensive for a New Zealand state school.  

Although we are only in the early stages of our donor discovery process, the work we are doing with Potentiality is throwing up some interesting data which we are now assessing. Please also note, we have a small base of donors. Many support the Academic Endowment Fund campaign which is an ongoing programme, however, since 2018 we have also been operating a capital campaign in parallel,  acquiring and engaging with new donors.

Once the initial setup was complete (changing daily habits and migrating existing data from numerous sources) the day-to-day management of our office was greatly simplified and the visibility and control we have on the personal information of our whole school community was greatly increased.

But with a data specialist on staff the focus then became how we could start using the wealth of engagement data automatically collected as part of our daily processes, to help with donor identification. 

Some examples of the data collected for analysis:

  • 4.6 million clicks across 3 connected sites over 3 years
  • Over 650,000 read and 125,000 click statistics on all bulk communications (Headmasters’ Bulletins, Old Boy e-newsletters, Auckland Grammar School Foundation Trust communications and many parent notifications)
  • 83,000 event attendances (all school and alumni events over 20 years)
  • 8,000 individual Outlook emails (tracked through an Outlook Plugin), 2000 attached notes as well as meetings and phone calls
  • Approximately 2000 survey results from login registrations (including expressions of interest in donating)
  • 34,000 payments to the school (excluding donations and events)
  • Relevant data from the school database such as leadership positions or awards received

So how can this data help identify donors? The approach was to use advanced statistical analysis in a community of 55,000 members to understand if and to what extent the different engagement data (e.g. email clicks and events attended) relates to the probability of a person becoming a donor or ‘donor propensity’.

The analysis

With the help of Potentiality we gave an anonymous version of the data to a data analyst to calculate the influence of each engagement variable on the likelihood of someone to donate. They ran a regression analysis involving the huge quantity of engagement data cross-referenced with 20 years of historical fundraising data. Once they completed their work we had a very good idea about how engagement influenced fundraising behaviour. 

For example if a community member logged in to the online community and expressed an interest in Capital Projects or Supporting Teachers, the likelihood of becoming a donor increased around 5000%. Some other interesting statistics were an increase of 327% per event attended, 288% per Outlook email communication (tracked via a plugin), 41% if they read a high proportion of school emails, 47% if they had an award or honor whilst at school, 280% if they took part in the community business directory, 335% if they’d viewed fundraising information on the school website, and 200% if they viewed the archives (the last two tracked thanks to the online community LinkedIn and Facebook integration).

Using the data

At this point we introduce our donor prospecting tool within Potentiality which plots each member of the community on a “propensity versus capacity to donate” chart based on live data in the database. A donor prospect’s propensity position on the chart is influenced by filters that can have variable scales which we set based on the results from the data analyst, and the capacity position is based primarily on census data, occupations and approximated property values.

The final chart has 55,000 dots representing every member of the community ranked between 1 and 55,000 indicating where they rate in capacity and propensity. The highest ranked capacity members appear across the top of the chart and the highest ranked propensity members appear on the right. The shape/colour of the dots represents their donor history.  With 55,000 records showing together interpretation is difficult but adding filters reveals trends in the data.

Figure 1

By filtering out non-donors and lapsed donors, we can see the different current donors on our chart. Of most significance the major donors (over $10K) are highlighted in yellow and our new donors this calendar year are light green stars.

You can see there’s a significant trend to the right side of the chart and also a huddle of dots in the top right corner. This tells us in broad terms that our analysis is working i.e. donors score high on propensity. Importantly however, the propensity calculation gives a stronger indication of a donor than our capacity data i.e there are more dots to the right than to the top.

This is good to see, but a donor might have increased communication whilst making a donation so to create the perfect donor prospecting analysis (or as close as we can get), we need to ensure that their position on the chart isn’t impacted by the fact that they’ve made a donation.

To achieve this we can look at donor engagement at different times before they became a donor to see if we can see the same trends.

Figure 2

Figure 2 shows only community members who made their first donation after January 1 2018 (when Potentiality had been running for a short time) and only counting engagement prior to that date. You can see there’s not much trend to speak of (approximately 40 future donors on the left and 60 on the right).

Figure 3

In Figure 3 we’ve run the same analysis in May 2019 (we have a lot more engagement data but it’s not long ago so there’s less stars) and all but one future first time donor is located on the right side of the chart, and around 60% of the future first time donors are within the top 10% propensity rating (the far right).

Turning the data into dollars

Now that we can see that it’s possible to identify community members that look like donors, the obvious next step is to see if we can turn them into donors.

Figure 4

Figure 4 is zoomed on the top right corner of the chart where we find the highest capacity and propensity community members and a filter has been applied to remove anyone who’s donated or been approached about donating in the past year.  The remainder show approximately 200 non-donors and 200 lapsed donors (of which 80 were major donors). Because we have sliding scales on the engagement data we can rule out community members whose engagement was a long time ago, these are all currently highly engaged members.

Each node comes with a brief description explaining its position by identifying with colours which filters matched and how strong the match was, for example based on filter numbers the highlighted node above matched with a medium capacity occupation, lives in a high income area according to the NZ census website, was in the top .1% (extremely high – red) of the community for; attending recent events, payments (excluding events or fundraising) and email read/click rate. They are in the top 2% (orange) for communications with the Development Office via Outlook and matched for a leadership position and awards and honors whilst at school. Our team can then dive further into their profile to have a closer look at their engagement history, what their interests are or who has been liaising with them (if anyone) to date. This builds a conversation template to start the cultivation process. 

At this stage, our strategy is to ask them for advice, not for money. These people are worth spending time on, we might invite them for a tour of the school or coffee with the headmaster. When the conversation is complete a new communication record is created and flagged as “involved a donation request” which automatically removes this person from the prospecting chart for now. Then we can choose the next person and so on.


All information used in this exercise is based on pre-existing data and if the analysis works then community members who don’t want to be contacted won’t be contacted.  A data-driven personal approach is clearly going to be received more warmly by the community and if you approach someone who gets annoyed or complains then there’s something wrong with the data or the analysis that informed the approach.  

In this case study, please note all information is held within the existing New Zealand privacy laws and any non-Grammar information is from publicly accessed websites.


What we’ve managed to show from this study is that engagement data is an effective indicator of future donors, and likely more significant than capacity analysis. In most schools, engagement data is either not considered important, or held in databases within other internal departments separate from any fundraising purpose. Based on our study they might want to rethink this approach.