21/01/2022 | Data Analytics | Gabrielle Gean
Opening the Gateway to Donors: Is Data Analytics The Golden Key?
- Data analytics enable charities to activate as well as retain their donors.
- Data analytics work best when five principles are followed: segmenting donors based on value, automating forecasts, predicting donor loyalty, determining what causes attrition, and using a test-and-learn approach.
Incumbent NGOs and charitable organisations may not notice how organisations with a bold digitalisation strategy are gaining share. Donor attrition rates tend to understate the issue, attributing the cause of defection mostly to donors contributing elsewhere, when in fact charities fail in obtaining the correct donor contact information. Maintaining a healthy base of primary donors may help to mask the grim reality of a declining share of donations.
Retaining donors is not a useful goal in and of itself. Rather, it should be propped up with deliberate and well-thought-out objectives, targeting the strongest segments. Indiscriminate retention reduces a charity’s funding because it requires spending money on loss-making donors or focusing on donors who are resistant to intervention.
One-time tactics rarely produce long-term results and are less effective than a systematic approach to donor loyalty that places donor priorities at the centre of the charity’s retention strategy. Donor leakage is caused in part by shifting donor needs and preferences, and in the event that the charity’s core strategy fails to meet these needs, its options become limited.
Recent advances in analytics enable charities to activate their high-value donors more effectively, through granular segmentation and personalisation. One charity in Poland, for example, collaborated with a software company and implemented personalised and interactive videos for donors in 2020 that yielded a 188% increase in funds raised year over year and encouraged nearly 13,000 people to share these personalised videos on social media.
Five principles for donor analytics
Charities that began down this path have discovered that deploying advanced analytics works best when it is supported by five principles.
Evaluate donors’ true value. Measures that indiscriminately target large donor groups by age, donation volume or other variables tend to be counterproductive, inadvertently triggering attrition or annoying disengaged donors. Moving beyond crude segmentation, advanced analytics can help develop a more granular view of target donor groups that are anchored in their value. For instance, a charity can segment its fundraising campaign strategy by age, donation amount, frequency, and preferred communication.
Marketing technology embedded in software platforms can help identify target segments who are the most profitable donors, by using donor lifetime value (LTV) metrics. Granular segmentation statistical methods include k-means clustering, latent class analysis, artificial neural networks, and random decision forests. The best method for a given situation is usually determined by practical considerations such as data availability and quality, team skills and tools, and a charity’s willingness to embrace dynamic segmentation.
Automate forecasting at scale. Many aggregate forecasts produce untrustworthy numbers, in part because it is difficult to attribute the effect of a single driver when a charity is launching multiple initiatives and campaigns. A more effective strategy would be to separate time series for cash deficits, donation flow, and other variables while performing proper baselining for each.
Another barrier to large-scale forecast adoption is the dependence on the manual approach, which is time consuming when preparing and collecting data, setting and tuning the model, and determining appropriate next steps. Automated forecasting processes for many subpopulations result in greater accuracy and pragmatism. Proven algorithms used by digital fundraising management organisations can tap the predictive power of a broader set of data sources, such as past donations, donor segmentations, macroeconomic variables, and market trends.
A crucial element of forecasting at scale is automatically incorporating expert judgement for a specific time series. Such expert forecasts include more information than statistical forecasts alone and adapt more gracefully to changing conditions or disruptive events. Having an analyst in the loop at scale necessitates an automatic evaluation of forecast quality and intuitive visualisation tools. With too many forecasts to manually check each one, the system must be able to identify those that may require expert intervention automatically.
Predict donor loyalty accurately for those who don’t respond to digital communication. Donor loyalty and advocacy are necessary for a charity’s long-term growth and profitability. A donor is a promoter when there’s a high chance of fulfilling a donation that is recommended by a close contact rather than pushed through advertising. Consequently, donors, who are promoters, donate more to their charities, cost less to retain, and are more likely to refer friends and colleagues.
Detractors can be harmful because donors are more likely to share negative feedback rather than positive feedback. They not only have a higher attrition rate, but they also take potential donations with them. To counteract one detractor, a minimum of two promoters are required. It is therefore critical to identify a detractor and improve their donor experience to avoid a chain reaction of negative referrals.
However, measuring loyalty has become more difficult in recent years. Surveys typically reach only a small percentage of donors at a prohibitive cost, and survey overload has forced charities to seek insights into donors’ perceptions of the brand and various episodes through alternative means. One option is to use datasets to train a predictive model on questions of advocacy and attrition.
A predictive model begins with basic features found in traditional segmentation, such as channel usage, the frequency and nature of donation interactions, and revenue. It then accesses more sophisticated data sources, such as the electronic direct mail (EDM) content or natural language processing of contact centre conversations (including volume and tone).
EDM analysis can also be useful in this situation. Constructing a social graph for the target donor segment based on transactional data or EDM open rates will yield a measure that captures an individual donor’s overall connectedness. Over a 3-year period, we conducted a case study with our charity partners in Thailand with a wide range of donor dataset. The results validated connectedness as a factor in attrition – a higher EDM open rate may result in lower attrition. This indicates the EDM content (rather than the amount of EDM sent) is one of the attributes to keep donors engaged.
Understand the factors that increase donor attrition and take actionable steps to reduce it. Traditional charity attrition models provide a single aggregate number based on customer churn, which is ineffective for frontline employees. It is unclear what factors account for the figure, what action should be taken, or how to address root causes.
Predictions without actionable recommendations are nearly useless. Explainable artificial intelligence (XAI) has the potential to change the game in this regard by revealing the relative importance of various variables on a donor’s likelihood to churn.
Charities can improve their models by asking departing donors why they are leaving. Once a charity understands the significance of each attrition factor, it can devise specific actions to address them. Frontline teams must have the right tools to carry out saving and recovery programs, whether it’s a customised strategy, sales material, call centre scripts, or other means. Those steps will be refined over time as new data and experience become available.
Rapidly test and learn which tactics connect with donors. As it is unlikely for a charity to identify which workable actions will yield results, ongoing experimentation is necessary.
Digital experts like Facebook and Amazon have run thousands of digital experiments to establish a huge userbase. Conversely, out of these experiments, many only run a few dozen demonstrations that have a negligible impact. Many charities that emphasise efficiency, predictability, and “winning”, regard the many experiments that fail as wasteful although in truth, these experiments are critical for staying in tune with the market.
Despite failing the experiments, charities can benefit. For analytics and donor engagement, they can create testable hypotheses about the best next steps among a plethora of options. Simply put, a testable hypothesis compares two options, Variant A and Variant B, to each other. Adept testers assign one of the options to similar but distinct groups and compare the result based on meaningful metrics to consistently retain and activate donors (see Figure 1).
Figure 1: A/B testing identifies the best fundraising strategies and campaigns for a specific donor segment and goal.
Adopting a granular, end-to-end conversion perspective has been shown to be the most effective in producing immediate results. Relevant metrics for converting prospects in the funnel could include the percentage of customers who open an email, click on a link in that email, receive advice, make a purchase, and stay with the charity for at least a year. Adept testers will zero in on appropriate measures to boost conversion at each stage, while keeping the larger challenge of retaining donors in mind.
Charities that avoid data-driven retention and activation practices, be it due to organisational design, or technological challenges, or unfamiliarity with regulatory norms, will fall behind in the race for high-value donors. Haste is necessary as donors are becoming more accustomed to receiving personalised experiences, and while technology allows for significant benefits, developing and effectively deploying the necessary skills takes time and practice.
In sum, the pandemic has altered the macro environment and donor behaviour, making historical data less reliable. The crisis has increased the importance of accurate forecasting, personalised communications, rapid testing and adaptation. After all, activating current donors who are open to growing the relationship with their charity of choice is far more effective than seeking new donors at any cost.
You might also be interested in: