Who should we contact first? Donor segmentation vs. AI prioritisation
Strategy & Frameworks
Static segments tell you who fits a bucket; predictive rankings tell you who to contact first.
Who should we contact first?
Contact the donors most likely to act right now, not the largest group that fits a rule.
Rule-based segmentation sorts donors into fixed buckets, such as recent givers or lapsed mid-value donors. AI-powered prioritisation ranks each donor by their likelihood to give, upgrade or lapse, and updates that ranking as new data arrives. For most teams with limited capacity, a ranked list answers "who first?" faster and more accurately than a static segment.
The two approaches are not rivals. Segments describe your file. Rankings tell you where to spend the next hour. The rest of this article compares them and gives you a framework for choosing.
What is rule-based donor segmentation?
Rule-based segmentation groups donors using fixed criteria you define in advance.
The most common methods are:
RFM: buckets based on recency, frequency and monetary value
Demographic: age, location, giving channel or acquisition source
Behavioural: event attendance, email engagement or campaign response
The output is a set of groups. Everyone in a group is treated the same, and the rules stay fixed until someone rewrites them.
What is AI-powered audience building?
AI-powered prioritisation scores every donor individually and ranks them by a predicted outcome.
Instead of one label per bucket, each donor gets a propensity score for a specific action: likelihood to give to the next appeal, likelihood to upgrade, or likelihood to lapse. The model reads your CRM history and refreshes as behaviour changes, so the ranking reflects this week, not last year.
The output is a ranked list you can cut at any point to match your budget or capacity.
Rule-based segments vs. AI prioritisation
Factor | Rule-based segments | AI prioritisation |
|---|---|---|
Unit of analysis | The group | The individual donor |
Output | Fixed buckets | Ranked list with scores |
Updates | Manual, periodic | Continuous, as data changes |
Answers "who first?" | Roughly, by bucket | Directly, by rank |
Setup effort | Low to moderate | Moderate, then automated |
Explainability | High and familiar | High when scores are transparent |
Best for | Small files, simple choices | Large files, tight capacity |
Main trade-off | Treats unlike donors alike | Requires clean data and trust in scores |
The trade-offs in plain terms
Segments are quick to build and easy to explain, but they treat everyone in a bucket as identical. A "lapsed mid-value" segment can contain a donor about to give again and one who will never return. You can't tell them apart, so you either mail both or guess.
Rankings separate those two donors. The cost is that you need reasonably clean CRM data and scores your team can inspect and trust. When scores are transparent, the ranking is as easy to justify as a segment, and more precise.
When are static segments enough?
Static segments work well when the choice is simple and the file is small.
Use rule-based segmentation when:
Your active file is small enough to contact most of it
The decision is coarse, such as a broad newsletter or a one-off event invite
You need a fast, familiar cut and precision won't change the outcome
Your data is too sparse to support reliable scoring
In these cases, adding a model won't change who you contact. A clear segment is the right tool.
When is predictive prioritisation necessary?
Predictive ranking earns its place when capacity is tight and every touch has a cost.
Use AI prioritisation when:
You can't contact everyone and need a defensible cutoff
Mail, media or staff time make over-contacting expensive
You want to mail fewer people without losing revenue
You need to spot upgrade candidates or at-risk donors early
Last year's rules no longer feel predictive
This is where rankings do what segments can't: they tell you where to draw the line and who sits just above and below it.
A framework for choosing
Work through these questions in order:
Can you contact your whole active file affordably? If yes, a segment is often enough. If no, rank and cut.
Does precision change the result? If separating donors within a bucket would change who you mail, use a ranking.
What's the cost of a wasted touch? High cost favours prioritisation. Low cost tolerates broad segments.
Do you need to defend the cutoff? A ranked list with transparent scores gives you a line you can explain to stakeholders.
Is the goal retention or upgrade? These depend on individual signals that buckets miss, so lean predictive.
A practical pattern for growing teams: use segments to frame the audience, then rank within it to decide who to contact first.
Practical takeaways
Segments describe your file; rankings tell you who to act on now.
Keep segments for small files and coarse, low-cost choices.
Switch to prioritisation when you can't mail everyone, or when retention and upgrades are the goal.
Insist on transparent scores so the cutoff is easy to explain.
You don't have to replace your CRM. A predictive layer sits on top of it and returns a ranked list into your existing workflow.
Conclusion
"Who should we contact first?" is a ranking question, not a grouping question. Rule-based segments still have a place for simple choices on small files. But when capacity is tight and every touch costs money, a continuously updated ranking lets you mail fewer people with confidence and protect results. Match the method to the stakes, and let the data draw the line before you spend the budget.
Related articles
Discover more insights into nonprofit fundraising and data from the Dataro team.
