Skip to main content
Every Allyy model answers one question about a supporter. They split two ways:
  • By outcome — a model either estimates a probability (classification, e.g. will they respond?) or an amount (regression, e.g. how much will they give?).
  • By scope — a model is either about a specific action (“if we mail this person now, what happens?”) or about a supporter’s trajectory (“what will this person do over the coming year?”).
You don’t need every model. Most teams start with one or two — typically a propensity model to prioritise a campaign, and later an expected amount or churn model to refine it.

​
Targeting a campaign — who to contact

DM propensity

Will this supporter respond to a mailing? Use it to rank a mailing list and cut volume without losing income.

TM reachability

Will we actually reach this person by phone? Prioritise dialling lists.

TM conversion

Given we reach them, will they accept the offer? Focus calls on likely converters.

Email engagement

Will this supporter open or click? Prioritise digital sends.

​
How much to ask — the value of an action

Expected amount

How much is this supporter likely to give in response to this action? Combine with propensity to target on expected revenue, not just likelihood.

Ask amount

What set of suggested amounts (the “ask string”, e.g. £20 / £50 / £100) maximises yield for this supporter?

​
Keeping supporters — retention & risk

Subscription churn

Will a regular giver cancel their direct debit in the coming months? Drives save / win-back programmes.

Active donor

Will a currently active supporter give again in the next 12 months?

Middle donor

Retention focused on the valuable “middle” segment — where small movements move a lot of income.

Subscription acquisition

Will a one-off donor convert to a recurring gift? Find your best regular-giving prospects.

​
Long-term value

Lifetime value (LTV)

What is this supporter worth over a multi-year horizon? Drives stewardship investment.

Donation frequency

How many gifts will this supporter make in the next year?

​
Timing — when something will happen

Time to next gift

How long until this supporter is likely to give again?

Time to churn

When is a supporter likely to stop giving?

​
Picking the right one

Think in terms of the decision you’re about to make:
  • Running a mailing or call campaign? Start with propensity, then layer expected amount to target on revenue.
  • Worried about losing regular givers? Use subscription churn.
  • Deciding where to invest stewardship? Use lifetime value.
Once you’ve chosen, head to Create & train a model.