A fundraising consultant can use AI for five client jobs: drafting appeals, writing donor reports, segmenting donor lists, building thank-you and follow-up sequences, and researching prospects. Donors tend to object to invented stories, impersonal letters and careless data use, so each job needs human approval, disclosure agreed with the client and data kept out of unvetted tools.
This is a menu you can sell, not a tool review; for tool choice see AI tools for fundraising consultants. For each job below you get what it is, an example with figures, effort and cost, and the objection you are most likely to meet, with a way to answer it.
Everything is illustrative: the figures are there to show how to reason, not as benchmarks. I name no prices for tools, because plans change often; check the vendor's page when you buy.
Before the menu: three rules that apply to every job
- A named human approves every outgoing word. You or the client's fundraiser. The tool drafts; it never sends.
- Donor data stays out of unvetted tools. Use aggregated or anonymised extracts for analysis, and read what donor data must never be pasted in before you begin. If a task needs personal data, use a tool with business data terms and training switched off, and record that choice.
- Facts come from the client. Beneficiary numbers, outcomes and quotes are supplied and checked, never generated.
Job 1: Appeal drafts and variants
What it is
You turn the client's brief, case for support and impact facts into an appeal letter or email, plus two or three variants for testing. The consultant's value is the structure, the ask and the judgement about which version suits which segment.
Example with numbers
Say a small environmental charity wants a year-end email appeal. The client supplies: the project (replanting a woodland), a goal of $12,000 for 600 saplings, the cost per sapling ($20), and a quote from a volunteer, used with permission. You supply the brief to the tool:
Draft a year-end email appeal of 250 words. Use only these facts: [facts]. Ask: a gift of any size; example gift: $20 plants one sapling. Tone: warm, plain, no pressure. Give two subject lines and two versions of the opening paragraph: one starting with the volunteer's quote, one starting with the goal. Do not invent any numbers, names or stories.A realistic output is competent but generic: the phrase "together we can make a difference" will appear, and it adds nothing. You cut it, tighten the ask, and check that "600 saplings" is $12,000 divided by $20. It is, so the arithmetic holds. The deliverable is the final text, the variants and a one-page note explaining which segment gets which opening.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowEffort and cost
About 90 minutes including client review, against perhaps three hours from scratch (my estimate; time your own). Tool cost is a small subscription you likely already carry.
The objection
"Donors will know it was written by a machine." Answer: they will if nobody edits it. The test is whether every fact is true, the voice sounds like the charity, and the ask is clear. A deeper workflow is in the donor appeal workflow tutorial.
Job 2: Donor reports and impact updates
What it is
You convert the client's raw outputs (project notes, spreadsheet totals, photos, case notes with permission) into a readable report for major donors or a trust or foundation funder, structured around what the funder asked for.
Example with numbers
Say a youth charity received a $25,000 grant for after-school sessions. The client sends attendance records: 40 sessions, 312 total attendances, 38 individual young people, 27 attending more than five times. You ask the tool to produce a first-draft report from those figures and the funder's headings, instructing it to use only supplied numbers. You then run the check yourself: 312 divided by 40 is 7.8 average attendances per session, which you can state, whereas the tool might round to "about 8" or misstate it. For percentage claims, "27 of 38 attended more than five times" is 71 per cent, rounded, and you write it from your own calculation.
Effort and cost
Two to four hours per report, depending on how clean the client's data is. Quality depends on inputs: if the client cannot supply outcome data, the report cannot invent it.
The objection
"Funders want authenticity." Yes, and they get it from real numbers and real quotes. AI helps with structure and plain language. For narrative stories, ask the client for permission-based quotes and never let the tool create a beneficiary voice.
Job 3: Donor segmentation and lapse-risk lists
What it is
You analyse the client's giving data to propose segments (recent donors, regular givers, lapsed, major prospects) and suggest who gets which message. Most of this is spreadsheet logic; AI helps write the rules, explain results and draft segment-specific messaging.
Example with numbers
Say a client has 4,200 donors. You export a table with anonymised IDs, gift dates and amounts, no names. Rules you define: Active (gave in the last 12 months), Lapsing (gave 13 to 24 months ago), Lapsed (older than 24 months), and Regular (three or more gifts in the last 24 months). You ask the tool for spreadsheet formulas that assign segments, run them yourself and get illustrative counts: 1,150 active, 640 lapsing, 2,410 lapsed. You then suggest a lapsing-donor reminder aimed at the 640. The method is covered in using AI to spot donors who are about to lapse.
A quick sum shows the value: if 640 lapsing donors give an average of $45 and a reminder brings back 10 per cent, that is 64 gifts, $2,880 in total. Both the 10 per cent and the $45 are assumptions for illustration; ask the client for their own history, and never promise a response rate.
Effort and cost
Half a day for a first segmentation, much less to repeat. The main risk is data protection: personal data should not go into a general tool. Use IDs, or a tool and agreement that allow processing, and confirm the client's privacy notice covers analysis.
The objection
"Donors will feel profiled." Segmenting by gift history is ordinary fundraising practice; the line is predicting sensitive characteristics or using data in ways donors would not expect. Keep to giving behaviour, say so in the client's notice and avoid inferring anything personal.
Job 4: Thank-you and stewardship sequences
What it is
You design a sequence of messages after a gift: immediate thank-you, a follow-up with impact after a few weeks, an invitation to give again later. AI drafts variants by gift level and channel; you set the rules and review the tone.
Example with numbers
Take three tiers for a client: first-time donors, regular givers and gifts above $500. The sequence for first-time donors: a thank-you within 48 hours, an impact email at day 21, an update at day 60. You draft one message per step. Then you ask the tool to produce a first-time donor email in the charity's voice using three sample letters the client approved. The output usually follows the sample's structure; you check that it does not promise outcomes beyond the facts, such as "your $20 will plant a tree" when the real figure is that $20 plants one sapling.
More on tone and personal touch is in donor thank-you letters with AI that still feel personal. For high-value donors, write by hand; the tool serves volume tiers.
Effort and cost
One day to design a three-tier sequence with a reviewed message set. Ongoing cost is the client's email platform. If the client already has automation, you configure the messages; if not, you recommend a platform within their budget.
The objection
"A thank-you from a machine feels hollow." It is hollow if it is generic. Specific details, a real signatory and a prompt response solve most of the problem. Keep humans for major gifts and any donor who writes back.
Job 5: Prospect and funder research briefs
What it is
You prepare a short brief on a potential funder or major donor prospect from public information: giving priorities, past awards, application dates and fit. AI speeds the reading and summarising; you verify against the funder's own pages.
Example
Say a client wants to approach a charitable trust. You give the tool the trust's own published guidelines (pasted text from its website) and ask for a one-page brief:
From the text below, list: what the funder says it funds, what it excludes, grant size range if stated, deadlines if stated, and any condition that could rule out this client (a youth charity with an annual income of about $300,000). Quote the source sentence for each point. If something is not stated, write "not stated".
[funder guidelines text]The value is the "quote the source" instruction, which makes each point checkable. A realistic failure: the tool states a deadline from an old page. You always open the funder's live page and confirm dates before the client relies on them.
A limit on prospect research about individuals
Research on private individuals raises privacy and fairness questions. Stick to public, professional information, avoid compiling personal profiles beyond what the client's policy and the law allow, and take advice if unsure. Many charities have a written policy on wealth screening; ask for it.
Effort and cost
One hour per funder for the first brief, much less once you have a template. The risk is stale or wrong detail, so verification is non-negotiable.
Pricing the five jobs
Price the deliverable and your judgement, not the time AI saves. If you charge by the hour and AI halves the time, your income falls while the client gets the same value. A fixed fee per deliverable avoids that and gives the client a clear price.
| Job | Deliverable | Illustrative fee basis | Includes |
|---|---|---|---|
| 1. Appeal drafts | One appeal, three variants, segment note | Fixed fee per appeal | Two revision rounds |
| 2. Donor reports | One report per funder | Fixed fee per report | Data check and one revision |
| 3. Segmentation | Rules, counts, messaging plan | Fixed project fee | Data handling agreement |
| 4. Stewardship sequence | Three-tier message set | Fixed project fee | Tone review with the client |
| 5. Funder briefs | One-page brief per funder | Fee per brief or small bundle | Source links and dates checked |
A way to set the number: estimate your hours with AI assistance, add a review margin, then check the price against the value to the client rather than against the time. An illustrative sum: if an appeal brings in a client $8,000 in gifts, a fee of a few hundred dollars is easy to justify, but I would not tie your fee to a percentage of funds raised. The Association of Fundraising Professionals' code of ethics, for one, bars members from accepting pay based on a percentage of contributions, and finder's fees, and your client or its staff may be bound by that code or a similar one. Check any code your clients follow, and put the fee basis in writing.
The proposals tutorial shows how to package a menu like this into a proposal in an hour.
Disclosure to donors: a policy you can agree with each client
Whether and how to disclose AI use depends on the client, its funders and any codes it follows. What I can say with confidence is that you should not leave it to chance. Put three lines in the engagement agreement (illustrative):
1. AI tools may be used to draft and edit materials; all outgoing content is reviewed and approved by [named person].
2. If a donor, funder or regulator asks whether AI was used, we will say so plainly.
3. No donor personal data is entered into any AI tool unless the tool and the purpose have been approved in writing by the charity.Where a funder publishes guidance on AI in applications, follow it; read each funder's current instructions rather than assuming either way.
Accuracy: the failure that damages trust fastest
A single invented statistic or story in an appeal can embarrass a charity. Three controls:
- A fact sheet supplied by the client, with source, for every number.
- A second reader who checks each claim against the fact sheet.
- A rule that quotes and case studies need written permission and are never generated.
An illustrative mistake: a draft says "last year we helped over 500 families". The client's figure is 478. The tool rounded up. A reviewer comparing against the fact sheet catches it; a reviewer reading for flow does not. Make the fact-check a separate pass.
Data protection basics for the consultant
As a consultant you will often be a processor of the client's data, which usually means a written agreement. Ask what data protection obligations apply to the client, what the privacy notice says about analysis, and whether the tool counts as a subprocessor. The general risks are described in charity AI risks: data protection, deepfakes and donor trust. Where you are unsure, ask a qualified adviser; none of this is legal advice.
Effort and cost at a glance, with an honest total
For a client wanting all five jobs once, the preparation time might add up like this (illustrative): appeals 1.5 hours, report 3 hours, segmentation 4 hours, stewardship 6 hours, two funder briefs 2 hours. That is 16.5 hours, which you would price as fixed deliverables rather than 16.5 hours at an hourly rate. Your own figures will differ; time two or three real jobs and set your fees from that.
Answering donor objections: wording you can adapt
Clients worry about the day a donor writes in. Prepare short, honest answers in advance, and let the client's own voice shape them. Illustrative wording:
| Donor says | A reply to adapt |
|---|---|
| "Was this letter written by AI?" | "We use AI tools to help draft some of our materials, and [name] reads and approves everything we send. The facts and figures come from our own records." |
| "Why are you spending money on AI instead of the cause?" | "We use it to save staff time on drafting, so more of the working day goes to the cause. The tools we use cost [amount or a short description], and we review that choice each year." |
| "Is my data being fed into an AI?" | "We do not put donor personal details into AI tools. Where we analyse giving patterns we use anonymised data. Our privacy notice says more, and I can send it." |
| "Please don't contact me with generated messages." | "Of course. I have noted your preference and you will hear only from a member of our team." |
Never write a reply you cannot stand behind. If the client does put personal data into a tool, the third answer would be false, so settle the data rules first and write the answers second. Record each objection in a simple log, because a pattern of complaints about one type of message is a reason to change it.
A pilot plan you can propose to a first client
A small pilot shows value without large commitment. An illustrative six-week plan for a client with a lapsing-donor problem:
- Week 1: agree the data rules, the disclosure policy and the baseline (last year's response to a similar mailing).
- Week 2: segment an anonymised extract; present counts for active, lapsing and lapsed donors.
- Week 3: draft the reminder message and one variant; the client's fundraiser edits both.
- Week 4: send variant A to half the lapsing group and variant B to the other half.
- Weeks 5 and 6: collect results, write a one-page note on what worked and what to change, and agree whether to continue.
Be careful with the conclusion. With 640 lapsing donors split into two groups of 320, a difference of a few gifts is not strong evidence that one variant is better, so say so in your note and avoid overclaiming. The pilot proves the process and gives a baseline; it does not prove a response rate.
Choosing which job to sell first
Start where the client's pain and your confidence meet. A client with a large lapsed list probably benefits first from segmentation and a reminder sequence. A client with a funder deadline needs the report or brief. A client with no appeal plan needs Job 1. Offer one job as a pilot, measure the result against a baseline the client already has, and expand if it works. If you want to talk through the packaging of your own menu, my AI implementation consultation covers it.
Questions consultants hear from charity clients
Do charities have to tell donors a letter used AI?
There is no single rule I can cite for every client. Some sectors, funders or codes ask for disclosure, and donors may ask directly. Agree a policy with each client in writing: when to disclose, what to say, and that a named person approves every message. Never deny AI use if asked.
Can AI write a fundraising appeal on its own?
It can draft one, but a draft is not a finished appeal. Facts, figures, beneficiary stories and tone need checking by someone who knows the cause. The risky errors are invented numbers and stories that read as real. Keep a human approving every claim before it goes out.
What donor data can go into an AI tool?
Only what the client's privacy notice and data terms allow, in a tool whose data terms you have checked. Prefer anonymised or aggregated data. Never paste names, contact details, giving histories or health or beneficiary details into a tool you have not vetted.
How should I price AI-assisted work?
Price the outcome and your judgement, not the hours AI saves. A fixed fee per deliverable, with a stated number of revision rounds, works well. Say plainly that you use AI to speed first drafts and that review time is included.
Further reads
- AI Agents for Startup Fundraising: Investor Lists, Outreach, Questions — The investor-side version of fundraising workflows.
- How Charities Use AI Chatbots to Answer Supporter Questions — Answering supporter questions with a chatbot.
- HeyGen for Nonprofits: Donor Update Videos and Disclosure Rules — Video updates and the disclosure rules around them.
- Make.com for Charities: Automate Donor Thank-You Emails — Automating thank-you emails after gifts.
- How to Mention AI Use in Client Contracts and Proposals — Contract wording for AI use in client work.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: the Association of Fundraising Professionals' Code of Ethical Standards (PDF on afpglobal.org, checked 4 November 2026) for the percentage-pay point; no external statistics used; examples and figures are illustrative. Check each client's privacy notice, data-protection obligations and any sector code or funder rule on AI disclosure.