123. SPN: AI Applications for EOY planning and insights
Creative is the best tool to improve the performance of your media spend
A very warm welcome to all the new subscribers. I’m thrilled to have you as readers and truly appreciate your feedback and support.
I don’t add up in public if I can help it and mental gymnastics isn’t a strong suit of mine. And yet too many donation forms are still an exercise in mental gymnastics!
Fundraise Up was built to solve a donor problem - give me the most frictionless, ecomm-like donation experience, and take my money now rather than give me 16 fields to fill in!
The result? Generosity unlocked. It’s time to elevate every donor’s experience of your Org.
Game changer? It is for me.
In this week’s SPN:
Creative is how you improve the performance of media spend
AI applications for EOY planning and insights
Taking inspo from other sectors
Jobs that took my fancy this week
Explore: Build Your Tech Stack (link)
Searching for new tools or trying to trim down your tech stack? Play around in this infographic. I add to it most weeks.
Before we get into today’s SPN, specifically the AI applications I’m leveraging for EOY planning and insights, it bears at least a mention that the energy requirements of AI data centers is truly astounding!
Apparently Google is ordering up to 7 small modular nuclear reactors - ones cooled by molten fluoride salt, rather than water. Microsoft is in talks about recommissioning the infamous Three Mile Island and Amazon is buying a stake in a nuclear energy developer in push to power data centers.
Given that the lead times for nuclear power plants is long and the ramp up in AI looks fast, what is going to happen to fill the energy gap over the next few years?
AdTech
I attended a pretty awe-inspiring Meta talk that outlined their roadmap of AI tools focused on improving Ad outcomes (by optimizing creative) recently. Google have done the same over recent months and at their Unboxed event this week Amazon showed off their new tech (which I nodded to in SPN 121).
Taking inspo as we like to do in SPN from other sectors, this look at how these new Amazon tools might be used by an Amazon seller was thought provoking - via bright minds at FordeBaker (an Amazon and ecomm marketplaces agency). Seeing different applications of technology is helpful. It refreshes how I approach hurdles with current Org’s.
Obviously look to see if and how you can enhance and expand on the platform tools you have before committing to any of this, but it’s worth a play and your agencies should be all over this.
Voice
And what are your thoughts about Voice being the main way of interacting and donating with AI? It was an interesting question Sal Salpietro asked me on a webinar last week. It’s something I explored in SPN 59, but only specific to Alexa.
I’ve a sense written prompt engineering will retain its place but it’s not for civilians per se. Apple intend to use Siri and Meta have enabled voice in its AI. Quite how Amazon plan to use Alexa is less clear but Google are sharing content on How to use Google Lens to ask questions out loud about what you see.
Creative
The maturity of this space underlines my belief that creative is now the best tool to improve the performance of your media spend - but who do you task with making the most of this? Your Creative support or your Media agency? Or do you do it in-house? That’s a tough call for most Orgs.
Jobs & Opps 🛠️
World Central Kitchen: International Fundraising Consultant (Spain)
New-York Historical Society: Manager, Development Operations ($70,000 - $80,000)
Operation Smile UK: Senior Acquisition Manager (£39,000 - £45,000)
Engage for Good: Senior Manager, MarComms & Growth
UNHRC: Private Sector Partnerships Individual Giving Associate
Doctors Without Borders/MSF USA: Director, Strategic Philanthropy ($149,836 - $224,754)
CSAA Insurance Group, an AAA Insurer: Community Impact Specialist ($71,000 - $107,0000)
UNICEF: Marketing Manager (Planning), LACRO, Panama (P-4)
British Red Cross: Digital Engagement Platform Manager (£35,292 - £40,000)
Anti-Defamation League: Director, Digital Transformation (up to $180,000)
National Gallery of Art (DC): Managing Editor, Brand ($139,395 - $181,216)
→ More jobs updated daily to SPN’s sister website: www.pledgr.com
AI Applications to Drive $$ Ahead of EOY
I’m often asked about which practical, data-safe AI applications I’m playing with so here are some fundraising-specific tools and approaches I’m tinkering with to increase revenue ahead of EOY and beyond.
I’ve broken them down into the following segments:
Current Donor Analytics and Activation
Prospecting, New Donor Acquisition
Creative and Website
Some of these are easy to orchestrate, some of them require some agency support. Take what feels relevant to your team's talents and available resources, and discard (or dream!) about the rest. There’s something here for everyone.
Current Donor Analytics and Activation
1. Use built-in Google Analytics 4 functionality to create “predictive audiences” of donors likely to donate in the next 7 or 14 days. Deploy these audiences to Google Ads for heavy-up retargeting campaigns with a higher budget than evergreen retargeting.
SPN Tip: Create two custom audiences by combining the predictive metrics and a random qualifier, such as the even or odd last digit in the transaction ID of their first donation. Heavy-up retargeting campaigns for one group and stop them for the other to measure incrementality.
2. Your email campaign cadences were likely automated a while ago and need to be revised. If you’re a Hubspot Org, they have a set of AI solutions, specifically an integration with Seventh Sense that’s available on their marketplace, which optimizes targeting and time of drops based on real-time engagement data.
SPN Tip: For US Org’s in particular, deploy it before the Election to automatically adjust your email schedule based on open rates. Manual adjustments will be too slow between October 20 and November 10.
3. If your Org has implemented Google Cloud Platform or Amazon Web Services, you may be in luck. Both of these platforms already have enterprise-grade security. All the data inside these platforms is subject to the Consumer Protection Acts in the US and GDPR in the EU, so you can upload donor-level PII. Both tools include catalogs of pre-built AI applications called AutoML for Google and Amazon.
Use Google’s pre-built segmentation models to automatically create audiences of donors based on GA data enriched with one of many public sources within the Google data “store.”
SPN Tip: A few segments I found helpful are:
a) “most likely EOY donors,” segmented based on last-year EOY donations
b) “most likely to increase their annual EOY donation,” segmented based on the in-year donation history
c) “most likely to churn,” segmented based on similarity to recently churned donors
SPN Tip 2: For these use cases, employ the help of your agency or in-house IT team. The lift here is MAX one day of work for them. But the payoff to deploying these audiences across your fundraising campaigns before year-end will pay off in multiples.
4. Implement an out-of-the-box tool like Dataro and use its “Predict” function. It’s AI-enabled platform will create an always-refreshed segment of donors most likely to donate again or move from one-time to regular donation, to deploy in your Email or Direct Mail campaigns.
Prospecting, New Donor Acquisition
By now, your Org is likely already using Google’s Performance Max (SPN #28) and Meta’s Advantage+ (SPN #117) campaigns for prospecting and acquisition. Both are applications of AI and I haven’t included them as stand-alone use cases here.
There are others:
1. Google’s and Meta’s solutions are black boxes with no access to back-end models or even solid reporting on what works and what doesn’t. It’s extremely limiting – and, most budget is still spent on traditional campaigns across both platforms.
Continuous testing helps ensure they perform, especially during the tumultuous and high-priced EOY period. In Q4, resources are stripped and teams likely don’t have time to conduct enough experiments. I’ve found AdEspresso and AdExt to be beneficial substitutes.
Both tools (AdExt worked better for me) complement built-in Google and Facebook AI models, and are Testing Tools rather than full-blown AI campaign builders. They help run multiple experiments simultaneously and segment your campaigns much more granularly.
AI only selects which experiments to run next based on the results of the previous ones. Both these tools will operate on top of your existing Google and Facebook accounts, so you retain access to the traditional reporting.
SPN Tip: Don’t deploy these on Performance Max (Google) or Advantage+ campaigns (FB). Models get into “conflict” with one another, and performance suffers. Think of AdExt as a tool to bring traditional Search and Display Network campaigns close to Performance Max in ease of use, while maintaining the level of insight of “normal” reporting.
2. Instead of adding more AI tools to your prospecting efforts, use the time spent on implementation to “train” Performance Max and Advantage+. Out of the box, these algorithms learn by “backpropagation.”
In simple terms (the best terms!), when a Performance Max campaign is launched it starts with the broadest possible targeting, slowly adjusting and getting narrower with every next impression.
Since the best prospective donor profiles for your Org are very different, these platforms get “lost” when a 65-year-old sports enthusiast from rural Arkansas with a household income of $60k and a 34-year-old gamer from San Francisco donate the same amount. As a result, campaigns target both audiences at the same frequency and with the same bid amount. Furthermore, they target all the people who look like either of them.
To improve performance, break up Performance Max (and Advantage+) campaigns into smaller ones, artificially limiting the breadth of targeting available to AI.
Four dimensions I’ve found to be particularly useful are:
Age and Gender – exclude the groups that never work for your Org and split out the others
Household Income
Affinity Audiences (Google) and Interests (Facebook)
Location (at a State level)
Aim to split your campaigns in two every week using the above dimensions, leaving the best-performing audience in the existing campaign (for example, if you currently have one campaign, are using Sports as the first dimension to split, and Runners generally perform better for your Org – create a new campaign to target Rowers, and adjust the settings of the existing one only to include Runners).
3. If your Org has access to Google Cloud and Search Ads 360, the Auto ML models explained in #3 (above) can be used for Prospecting via custom Bid Strategies in Search Ads. Here’s a good guide. SA360’s bid strategies rely on floodlights – tracking pixels within Google’s marketing environment. Floodlight conversion events can be automatically uploaded from the Cloud, tweaking the default algorithms.
SPN Tip: Work with your agency or an internal IT team to integrate the two tools. Once integrated, your team can implement “custom” Floodlight conversions based on the predicted lifetime value of a donor informed by Google’s AutoML models.
This will effectively create your proprietary version of Performance Max campaigns. Instead of public conversion and donor data used to train the algorithm, it’ll be based on your Org’s proprietary donor data from tools that “traditional” Google Ads campaigns don’t have access to - such as CRM or P2P fundraising platforms, which are likely already integrated into Google Cloud Platform as well.
Creative and Website
The most common use of new-ish conversational GenAI models is undoubtedly creative generation.
Without an Org-wide AI deployment plan and enclosed enterprise instances of Gemini, Copilot, or ChatGPT to ensure data privacy, applications are limited to those not involving uploading donor information into the tool.
NB. Org’s should be uncomfortable with trusting AI to generate content for donors without refinement, validation and double checking it, so I’ve limited the use cases below to enhancing your team’s workflow, not replacing it.
1. Refine your EOY email copy. ChatGPT, Copilot, or Gemini are great at making writing more concise, and attention is scarce during the EOY period. If your Org doesn’t trust AI tools to generate new creative, you can use them to improve the assets generated by humans.
Try the following prompt: “Shorten the below fundraising email to no more than two paragraphs. Retain the voice currently used in the email, and don’t add any new facts or adjectives”.
SPN Tip: The same can be done with Hubspot’s AI Email Writer tool, but instead of open-ended prompts, it only offers pre-selected functions such as “Expand,” “Summarize,” or “Rewrite in a different voice.”
2. Change your default donation amounts. Use Fundraise Up and this all happens for you. Game-changer? You know it is for me!
If you wanted to have a little fun though, you could use AI tools to help recommend default donation amounts without using personal donor data. Export a table of all your online donations from Google Analytics over the last three months with four columns: Date, Sessions, Conversions, and Revenue.
The prompt I’ve used is “Analyze the below transaction data for my Org and recommend what are the best three donation amounts to show as default on the donation page. These donation amounts should maximize the average donation value, calculated as Revenue divided by Conversions, while maintaining the same conversion rate, calculated as Conversions divided by Sessions”.
3. Use AI to analyze weak spots on your website. The newest release of ChatGPT has access to the internet in real-time, allowing you to explore live websites instead of copying/pasting text into the chat box. Use ChatGPT to analyze your landing pages and compare them against peer Org’s.
OK, that’s all for today!
What did I miss? What tools or processes do you have set up that you love?
If you enjoyed this SPN, please consider sharing with your network. Thank you to those that do.
If a friend sent this to you, get the next edition of SPN by signing up below.
And huge thanks to this Quarter’s sponsor Fundraise Up for creating a new standard for online giving.
Reads From My Week
B2B is having a moment - the LinkedIn event B2Believe this week surfaced lots of interesting content (LinkedIN)
How to Grow Your Brand (Mailchimp)
A global study from Adobe explores AIs impact on the future of Art, Film and wider culture (Adobe)
TikTok executives know about app’s effect on teens, lawsuit documents allege (NPR)
Enter the high street of the future with Snap Street – Augmented Reality - popping up in Shoreditch for one weekend in October (Creative Brief)
How Kraft Heinz's new CMO Todd Kaplan is defending creativity in the age of data (Digiday)
Why advertisers need a new omnichannel measurement approach amid a disconnected landscape (Digiday)
Accenture Life Trends 2025 (Accenture)
Did Apple Just Kill Social Apps? (NYTimes)