July 28, 2025 • 19 mins
Every fundraiser faces the same dilemma: we need efficiency, but we can't sacrifice accuracy. When that major donor asks for one critical data point before finalizing their six-figure gift, the pressure to respond quickly meets our professional obligation to be absolutely correct.
ChatGPT makes this balancing act even trickier. Its confident, polished responses can mask a dangerous problem: AI hallucinations—fabricated information delivered with complete assurance. These aren't intentional lies but rather the model filling knowledge gaps with what sounds plausible. The danger comes when these fabrications slip into donor communications, undermining the trust foundation that every successful fundraising relationship requires.
This episode introduces a powerful principle: accuracy is stewardship. When donors receive information from your organization, they're not just reading facts; they're making judgments about your credibility. Even small inaccuracies can create cracks in the foundation of trust, and in fundraising, these cracks directly impact your ability to secure and retain support for your mission.
The good news? A simple five-word prompt can dramatically reduce this risk: "Cite sources for every claim." By implementing the "source-then-write loop" workflow, you can transform how you use AI—requesting verification before content creation rather than after. This approach doesn't just protect against errors; it lets you leverage AI's efficiency while maintaining complete confidence in what you send.
Whether you're writing donor updates, impact reports, or quick responses to board inquiries, this verification-first approach ensures you never sacrifice integrity for speed. Because in fundraising, trust isn't just nice to have—it's everything. Join us to discover how the most successful fundraisers are using AI to enhance relationships rather than potentially damage them.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Keith Greer, CFRE (00:00):
A major donor emails she's ready to move
forward with a six-figure gift,but asks for one last data point
before wiring the funds.
You type the question into chat, gpt, hit, enter and wait.
A polished answer appears inseconds.
But then it hits you when didthat number come from?
You scan the response again nosource, no citation, just AI's
(00:22):
confidence.
But now your confidence israttled because here's the thing
Send a made-up stat and thedamage doesn't stay on your
screen.
It follows you into donorconversations, into credibility
calls, into missed renewals.
But with one five-word promptyou could have surfaced the
source and safeguarded the trustyou've worked so hard to build.
(00:43):
So let's talk fundraising.
There's a golden rule infundraising that often gets less
airtime than it should.
Accuracy is stewardship.
We talk a lot aboutrelationships, emotion,
storytelling, and rightly so.
But the foundation underneathall that warmth and connection
(01:04):
is trust, and trust hinges oncredibility.
When a donor receives a messagefrom your organization, whether
it's an impact report, acampaign update or a one-line
email with a statistic, they'renot just reading facts, they're
making decisions about whetherthey believe you, whether they
believe in you.
(01:25):
And when the details in thatmessage are even slightly off,
even if the mistake isunintentional, the donor's
belief wobbles.
That wobble might not make asound, but it does leave a mark.
This is especially true when itcomes to numbers.
Donors love stories, but theyalso crave evidence.
They want to know the gift madea difference.
(01:47):
They want to feel thetransformation, yes, but they
also want to see the proof.
So when they ask for a datapoint, something specific and
verifiable, they're handing youan opportunity to reinforce
trust.
And the moment that data pointis inaccurate or fabricated,
even if it's well-written orwell-meaning, that opportunity
(02:09):
can vanish.
So here's the challenge thatmakes this even more pressing.
Today, Tools like ChatGPT areconfident communicators.
They're designed to soundfluent, polished and persuasive,
but they're also predictionengines, not search engines.
That means they're pulling frompatterns and probabilities, not
always from hard verifiablefacts.
(02:31):
And when that nuance goesunacknowledged, it opens the
door for something we callhallucinations, confidently
delivered falsehoods.
Let's be clear Hallucinationsaren't intentional deception.
They're not lies in the way aperson might lie.
They're simply the modelfilling in gaps with its best
guess.
The trouble is that guess cansound so smooth and so
(02:55):
authoritative we forget toquestion it.
You've probably seen thisyourself.
Maybe you asked ChatGPT tosummarize a research study.
It named a journal that soundedfamiliar but didn't actually
exist.
Or maybe you asked for thehistory of a donor program and
it gave you a beautifullywritten paragraph, none of which
matched your records.
(03:16):
These aren't rare glitches.
They're built into how largelanguage models operate.
This is why the principlematters so much.
Language models operate.
This is why the principlematters so much.
If stewardship is about honoringyour donor's trust, then
verifying your facts, especiallyAI-generated ones, isn't
optional.
It's part of the job, and thegap between that sounds right
(03:37):
and that is right can beenormous.
But here's what's mostencouraging Most errors don't
happen because fundraisers arecareless.
They happen because we'removing fast.
We're under pressure.
There's an email to send aproposal to finalize, a board
member waiting for a statisticand in the flurry of activity,
we assume the information issound, especially when it looks
(04:01):
polished.
This is where a simple practicecan change everything.
Imagine if, before askingChatGPT to write your next donor
message, you took a beat andran one short check, cite
sources for every claim, fivewords, that's it.
Suddenly, the model shifts fromjust drafting content to
(04:22):
backing it up, and if it can't,that's your cue to pause, check
and revise before anything goeslive.
So let's be honest.
You probably alreadyfact-checked your own work, but
AI output can feel deceptivelydone, which makes skipping the
step even more tempting, andthat's why building the check
into your prompt up front is sopowerful.
(04:44):
It prevents you from beinglulled into false confidence by
a well-written hallucination.
There's also a deeper truth hereworth naming.
When a donor receivesinformation that isn't accurate,
it doesn't matter whether themistake came from a rushed human
or an overconfident AI.
To them it feels the same andthe risk to the relationship is
(05:06):
the same.
Think about your own recentwork when the last time you sent
a message with a number or afact that you assumed was
correct but didn't double check.
Maybe it was a stat from a pastreport, maybe it was a program
outcome you remembered from lastyear.
Nothing sinister, just a quickassumption.
Most of us have done it andmost of the time it goes
(05:27):
unnoticed.
But that, most of the time, isexactly the issue, because it
only takes one time for a donorto notice and once that trust is
questioned, it's incrediblyhard to rebuild.
In a profession where trust isthe foundation of every major
gift, every renewal, everylegacy commitment, that's not a
gamble worth taking.
(05:49):
The beauty of a proactiveapproach is that it doesn't
require you to become afact-checking machine or a data
analyst.
It simply invites you to leadwith curiosity, to build a
culture of asking where did thiscome from?
That single question can saverelationships and, even better,
it can model for your team, yourinterns, your colleagues that
(06:09):
accuracy isn't aboutperfectionism, it's about care,
because when you care enough tocheck, your donors feel it and
that kind of trust isn't justpreserved, it's deepened.
So here's the fundraisingprinciple we're standing on
today.
Accuracy is stewardship, and AIis only as accurate as the
(06:29):
prompts and follow-up questionsthat you give it.
When you bring that mindsetinto your writing process,
everything shifts.
Your messages get sharper, yourstats get cleaner, your
confidence gets stronger and,most importantly, your donors
stay rooted in trust.
Take 10 seconds now.
Think about the last stat ordata point you included in a
(06:51):
donor communication.
Did you verify the source?
Did you cross-check it or didit just sound right?
No shame here, just awareness.
And with awareness comes theopportunity to lead with
integrity, even when you'reworking fast.
In a moment we'll break downexactly how to build this
verification into your chat GPTworkflow, but for now, remember
(07:14):
this Stewardship isn't justabout stories and smiles, it's
also about precision, and, inthe age of AI, the fundraisers
who pair speed with accuracywill be the ones who earn and
keep the deepest donor trust.
Let's turn now from principle topractice.
You know that accuracy isnon-negotiable and that AI,
(07:36):
powerful as it is, doesn'tguarantee truth.
But here's the good news Withthe right workflow, chatgpt can
actually make your fact-checkingfaster, more structured and
less stressful.
The key is what I call thesource-then-write loop.
It's a two-step process thatflips the usual sequence
fundraisers follow when using AI.
(07:59):
Instead of asking ChatGPT towrite the whole paragraph and
then hoping it got the detailsright, you start by asking for
the sources first.
Here's how it works Beforedrafting anything, you prompt
ChatGPT like this Cite sourcesfor every claim.
That's the five-word safetycheck.
It's simple, but it's powerful.
(08:20):
By asking for sources up front,you signal that you're not just
looking for smooth sentences.
By asking for sources up front,you signal that you're not just
looking for smooth sentences,you're looking for verifiable
facts.
You're not delegating judgment,you're guiding it.
So let's say you're preparing adonor update about a $25,000
endowed gift that supportedunderrepresented engineering
students last year.
Normally you might say chat GPTwrite a 150-word impact
(08:45):
paragraph thanking the donor anddescribing how their gift made
a difference.
But in the source, then writeloop.
Your prompt would sound morelike this Before writing, list
three verifiable sources thatprove how last year's $25,000
endowed gift supportedunderrepresented engineering
students.
Cite program statistics fromthe attached files or my
(09:07):
organization's website.
After I confirm them, draft a150-word impact paragraph in a
warm, donor-centric tone.
In response, chatgpt might lista program report from your own
organization's website, a pressrelease or a blog post featuring
a student spotlight, a fundingupdate published in last year's
(09:28):
annual report.
Then, and only then, do youmove forward to the writing step
.
Once you've confirmed that thesources are real, accurate and
appropriately cited, you inviteChatGPT to create the donor
message.
This workflow has a few bigadvantages.
First, it prevents you fromrelying on confident sounding
(09:49):
guesses.
When you ask for sources first,you're much more likely to
notice if something feels fishyor vague.
You won't be seduced by smoothlanguage.
You'll be focused on truth.
Second, it saves time, and thatmight sound counterintuitive.
Doesn't adding a step makethings slower?
Not really, because it's muchfaster to validate three sources
(10:11):
before you write than torewrite a message after
realizing something's wrong.
In most cases, you'll shave offrevision cycles, clarification
emails and back and forth withcolleagues.
Third, it builds muscle memory.
After a few repetitions, thisprocess becomes second nature.
You start spotting red flagsfaster, you trust your draft
(10:32):
more fully and, best of all, youstart sending AI-assisted
content with real confidence,not quiet anxiety.
Now let's talk about the secondpart of this process, the
one-click reference audit.
After ChatGPT gives you itssources, usually as URLs or
citations, take a minute toverify them.
Paste the links into yourbrowser.
(10:54):
See if they actually lead tothe source that's described.
If one fails, don't panic.
Just flag the sentence andeither adjust it or find a
replacement.
This isn't about being perfect.
It's about protecting yourcredibility.
You don't need to become afull-time fact checker.
You just need to stay in thedriver's seat One of the
simplest ways to spot trouble.
(11:16):
Watch out for citations thatsound almost right but don't
quite exist.
If a URL looks plausible butdoesn't lead anywhere, that's
your cue to dig deeper.
If ChatGPT references a statbut can't tell you where it came
from, treat it like aplaceholder until you can
confirm it.
Here's another real-world tip.
(11:36):
If you're pulling from your ownorganization's materials, like
annual reports, campaign updatesor board decks, drop those into
your ChatGPT prompt as part ofthe source list.
That way, the model hasgrounded content to draw from
and is less likely tohallucinate.
You're narrowing the focus,giving it guardrails and
dramatically increasing the oddsof a clean and accurate draft.
(12:00):
So let's recap the practicalprompts that power this approach
.
First, source first promptBefore writing list three
verifiable sources that showimpact statement and include
URLs.
Second, after you verify, nowwrite a 150-word donor update
(12:20):
based on the verified sourcesabove.
Keep it warm, professional andspecific, avoid generalities.
This structure worksbeautifully for a wide range of
communications, whether that'sdonor updates, a grant report,
board briefings, maybe afundraising appeal or even
year-end recaps.
And the best part Once you'veconfirmed the sources, once you
(12:41):
can often reuse the languageacross multiple messages or
segments with small tweaksinstead of the full rewrites,
you don't just save time, youreduce the cognitive load, you
stop wondering is this right?
And you start knowing yes, it'ssound, it's supported and it's
safe to send.
And when you're writing underpressure which, let's be honest,
(13:03):
is most of the time having thatconfidence is priceless.
So as you move into your nextdonor message, pause for a
second.
Ask yourself am I trusting theoutput because it looks good or
because I know it's grounded.
If it's the first, take theextra step, because accuracy
isn't a bonus, it's the baseline.
(13:24):
And when you pair thatcommitment with a tool like
ChatGPT, you don't just writewell, you lead well, you
fundraise with integrity andyour donors feel the difference.
Let's take a breath here,because this section isn't about
prompts or protocols.
It's about you and what mightbe running through your mind
even after learning the source.
(13:44):
Then write loop.
You might be thinking this allmakes sense.
I see the value, but doesn'tverifying AI output just cancel
out the time I was trying tosave in the first place?
And that question is honest andit deserves an honest answer.
Yes, adding a verification stepmeans spending another 60 to 90
seconds checking links orconfirming a statistic.
(14:06):
But let's zoom out.
What's the real cost of skippingthat check?
If a donor catches a factualerror in your message,
especially one tied to theirgift or program area, you don't
just lose face, you lose trust,you lose the sense of
reliability they felt when theyhanded you that check or they
wired those funds.
And rebuilding that trust thatdoesn't take 60 seconds.
(14:29):
That takes weeks, months, maybeeven years.
And it's rarely just one donorthe moment.
A stat is wrong in a boardreport or a fabricated ranking
slips into an appeal, the rippleeffect begins.
Donor, trust is relational,it's not transactional.
The word spreads, even subtly.
They're not buttoned up.
(14:50):
That number didn't check out,that wasn't accurate.
Those are dangerous impressionsto leave behind.
So if you're wondering whethera quick fact check slows you
down, I'd offer this it's not aslowdown, it's a safeguard, it's
speed that lasts, because thefastest route to credibility is
consistency, and that comes fromdetails that hold up.
(15:14):
Here's another shift that'soften necessary.
Many fundraisers, especiallyhigh-capacity performers, see
revisions or fact-checking as arework, as something to be
minimized.
And if you're using ChatGPT,there's a temptation to see its
fluency as final.
The writing sounds so good, sofast, that editing feels
(15:35):
optional.
But let's reframe that.
The goal isn't to cut editing,it's to edit smarter, to bring
discernment back to the surface,instead of letting the tool
lull you into copy-pasteconfidence.
When you ask for the sourc listfirst or run a link audit,
you're not adding steps, you'retaking back control.
(15:56):
This is especially important forthose of you who are trusted as
the voice of your shop.
You're the one who drafts themessages, the one leadership
counts on to get the languageright.
Your tone is your brand andyour accuracy is your integrity.
That responsibility can feelheavy, but it also gives you
power the power to model whatthoughtful, ethical AI use looks
(16:20):
like in the real world.
When you're the person who bothembraces the efficiency and
protects the truth, you set thetone for everyone else.
Your team learns that it'spossible to be fast and right at
the same time, that quality andspeed aren't opposites, they're
partners.
You also may be thinking, keith, do I really have the time to
(16:43):
train my whole team in this?
If you're stretched, I hearthat, but here's the shift.
You don't need to become the AIgatekeeper.
You need to become the culturesetter.
If you build these habits intoyour prompts, your workflows,
your templates, and share thatwith your teams, others will
follow your lead.
You can even start withsomething as small as a shared
(17:04):
checklist.
Add placeholders for anysensitive data, ask for sources
before writing, Confirm everycitation, customize tone after
verifying content.
That's it.
That's your culture of accuracybuilt into behavior.
And as this mindset starts tosink in, you'll notice something
unexpected your confidencegrows.
(17:25):
You're no longer wondering ifthat paragraph you copied from
ChatGPT will hold up underscrutiny.
You know it will, because youguided it, you checked it and
you owned it.
That's the real reward of thismindset shift.
It's not just about reducingerrors or preventing
embarrassment.
It's about showing up withclarity and confidence.
(17:45):
It's about knowing that everymessage you send reflects your
integrity, not just yourproductivity.
So pause and ask yourself whatwould change if I trusted myself
more than I trusted the tool.
What would shift if accuracywasn't just a back-end
correction but a front-endfilter for everything I write.
(18:06):
Let that land, because whenyou're rooted in stewardship of
time, of truth, of trust, youwrite differently, you lead
differently and your donors feelthe difference.
So today we tackled one of themost important questions in
AI-assisted fundraising Can Itrust the output?
(18:28):
We explored why large languagemodels like ChatGPT sometimes
hallucinate, how those errorshappen and why trust with donors
depends on your ability toseparate confident-sounding
fiction from verifiable fact.
We covered thesource-then-write loop, a simple
, repeatable workflow that asksfor citations before drafting
(18:49):
copy, and the five-word safetycheck Cite sources for every
claim.
These aren't just techniques.
They're habits that protect theintegrity of your message, your
brand and your donorrelationships.
And we talked mindset.
Accuracy isn't the enemy ofspeed.
It's what gives speed realstaying power.
(19:09):
When you verify before youwrite, you send messages with
confidence instead of crossingyour fingers.
Now, if this episode helped youfeel more confident about using
ChatGPT in your donorcommunications, there's one
small thing you can do thatmakes a big difference.
Would you please leave a ratingor review?
Wherever you listen, it helpsmore fundraisers discover the
(19:31):
show and join you in buildingdonor trust with intention and
integrity.
Next week, we'll startexploring workflow wins and time
savers.
We'll start with the inbox howto train ChatGPT to reply to
routine donor emails and giveyou back your time.
Until then, stay curious, stayclear and keep fundraising with
(19:56):
heart.
I'll see you real soon.
A major donor emails she's ready to move
forward with a six-figure gift,but asks for one last data point
before wiring the funds.
You type the question into chat, gpt, hit, enter and wait.
A polished answer appears inseconds.
But then it hits you when didthat number come from?
You scan the response again nosource, no citation, just AI's
(00:22):
confidence.
But now your confidence israttled because here's the thing
Send a made-up stat and thedamage doesn't stay on your
screen.
It follows you into donorconversations, into credibility
calls, into missed renewals.
But with one five-word promptyou could have surfaced the
source and safeguarded the trustyou've worked so hard to build.
(00:43):
So let's talk fundraising.
There's a golden rule infundraising that often gets less
airtime than it should.
Accuracy is stewardship.
We talk a lot aboutrelationships, emotion,
storytelling, and rightly so.
But the foundation underneathall that warmth and connection
(01:04):
is trust, and trust hinges oncredibility.
When a donor receives a messagefrom your organization, whether
it's an impact report, acampaign update or a one-line
email with a statistic, they'renot just reading facts, they're
making decisions about whetherthey believe you, whether they
believe in you.
(01:25):
And when the details in thatmessage are even slightly off,
even if the mistake isunintentional, the donor's
belief wobbles.
That wobble might not make asound, but it does leave a mark.
This is especially true when itcomes to numbers.
Donors love stories, but theyalso crave evidence.
They want to know the gift madea difference.
(01:47):
They want to feel thetransformation, yes, but they
also want to see the proof.
So when they ask for a datapoint, something specific and
verifiable, they're handing youan opportunity to reinforce
trust.
And the moment that data pointis inaccurate or fabricated,
even if it's well-written orwell-meaning, that opportunity
(02:09):
can vanish.
So here's the challenge thatmakes this even more pressing.
Today, Tools like ChatGPT areconfident communicators.
They're designed to soundfluent, polished and persuasive,
but they're also predictionengines, not search engines.
That means they're pulling frompatterns and probabilities, not
always from hard verifiablefacts.
(02:31):
And when that nuance goesunacknowledged, it opens the
door for something we callhallucinations, confidently
delivered falsehoods.
Let's be clear Hallucinationsaren't intentional deception.
They're not lies in the way aperson might lie.
They're simply the modelfilling in gaps with its best
guess.
The trouble is that guess cansound so smooth and so
(02:55):
authoritative we forget toquestion it.
You've probably seen thisyourself.
Maybe you asked ChatGPT tosummarize a research study.
It named a journal that soundedfamiliar but didn't actually
exist.
Or maybe you asked for thehistory of a donor program and
it gave you a beautifullywritten paragraph, none of which
matched your records.
(03:16):
These aren't rare glitches.
They're built into how largelanguage models operate.
This is why the principlematters so much.
Language models operate.
This is why the principlematters so much.
If stewardship is about honoringyour donor's trust, then
verifying your facts, especiallyAI-generated ones, isn't
optional.
It's part of the job, and thegap between that sounds right
(03:37):
and that is right can beenormous.
But here's what's mostencouraging Most errors don't
happen because fundraisers arecareless.
They happen because we'removing fast.
We're under pressure.
There's an email to send aproposal to finalize, a board
member waiting for a statisticand in the flurry of activity,
we assume the information issound, especially when it looks
(04:01):
polished.
This is where a simple practicecan change everything.
Imagine if, before askingChatGPT to write your next donor
message, you took a beat andran one short check, cite
sources for every claim, fivewords, that's it.
Suddenly, the model shifts fromjust drafting content to
(04:22):
backing it up, and if it can't,that's your cue to pause, check
and revise before anything goeslive.
So let's be honest.
You probably alreadyfact-checked your own work, but
AI output can feel deceptivelydone, which makes skipping the
step even more tempting, andthat's why building the check
into your prompt up front is sopowerful.
(04:44):
It prevents you from beinglulled into false confidence by
a well-written hallucination.
There's also a deeper truth hereworth naming.
When a donor receivesinformation that isn't accurate,
it doesn't matter whether themistake came from a rushed human
or an overconfident AI.
To them it feels the same andthe risk to the relationship is
(05:06):
the same.
Think about your own recentwork when the last time you sent
a message with a number or afact that you assumed was
correct but didn't double check.
Maybe it was a stat from a pastreport, maybe it was a program
outcome you remembered from lastyear.
Nothing sinister, just a quickassumption.
Most of us have done it andmost of the time it goes
(05:27):
unnoticed.
But that, most of the time, isexactly the issue, because it
only takes one time for a donorto notice and once that trust is
questioned, it's incrediblyhard to rebuild.
In a profession where trust isthe foundation of every major
gift, every renewal, everylegacy commitment, that's not a
gamble worth taking.
(05:49):
The beauty of a proactiveapproach is that it doesn't
require you to become afact-checking machine or a data
analyst.
It simply invites you to leadwith curiosity, to build a
culture of asking where did thiscome from?
That single question can saverelationships and, even better,
it can model for your team, yourinterns, your colleagues that
(06:09):
accuracy isn't aboutperfectionism, it's about care,
because when you care enough tocheck, your donors feel it and
that kind of trust isn't justpreserved, it's deepened.
So here's the fundraisingprinciple we're standing on
today.
Accuracy is stewardship, and AIis only as accurate as the
(06:29):
prompts and follow-up questionsthat you give it.
When you bring that mindsetinto your writing process,
everything shifts.
Your messages get sharper, yourstats get cleaner, your
confidence gets stronger and,most importantly, your donors
stay rooted in trust.
Take 10 seconds now.
Think about the last stat ordata point you included in a
(06:51):
donor communication.
Did you verify the source?
Did you cross-check it or didit just sound right?
No shame here, just awareness.
And with awareness comes theopportunity to lead with
integrity, even when you'reworking fast.
In a moment we'll break downexactly how to build this
verification into your chat GPTworkflow, but for now, remember
(07:14):
this Stewardship isn't justabout stories and smiles, it's
also about precision, and, inthe age of AI, the fundraisers
who pair speed with accuracywill be the ones who earn and
keep the deepest donor trust.
Let's turn now from principle topractice.
You know that accuracy isnon-negotiable and that AI,
(07:36):
powerful as it is, doesn'tguarantee truth.
But here's the good news Withthe right workflow, chatgpt can
actually make your fact-checkingfaster, more structured and
less stressful.
The key is what I call thesource-then-write loop.
It's a two-step process thatflips the usual sequence
fundraisers follow when using AI.
(07:59):
Instead of asking ChatGPT towrite the whole paragraph and
then hoping it got the detailsright, you start by asking for
the sources first.
Here's how it works Beforedrafting anything, you prompt
ChatGPT like this Cite sourcesfor every claim.
That's the five-word safetycheck.
It's simple, but it's powerful.
(08:20):
By asking for sources up front,you signal that you're not just
looking for smooth sentences.
By asking for sources up front,you signal that you're not just
looking for smooth sentences,you're looking for verifiable
facts.
You're not delegating judgment,you're guiding it.
So let's say you're preparing adonor update about a $25,000
endowed gift that supportedunderrepresented engineering
students last year.
Normally you might say chat GPTwrite a 150-word impact
(08:45):
paragraph thanking the donor anddescribing how their gift made
a difference.
But in the source, then writeloop.
Your prompt would sound morelike this Before writing, list
three verifiable sources thatprove how last year's $25,000
endowed gift supportedunderrepresented engineering
students.
Cite program statistics fromthe attached files or my
(09:07):
organization's website.
After I confirm them, draft a150-word impact paragraph in a
warm, donor-centric tone.
In response, chatgpt might lista program report from your own
organization's website, a pressrelease or a blog post featuring
a student spotlight, a fundingupdate published in last year's
(09:28):
annual report.
Then, and only then, do youmove forward to the writing step
.
Once you've confirmed that thesources are real, accurate and
appropriately cited, you inviteChatGPT to create the donor
message.
This workflow has a few bigadvantages.
First, it prevents you fromrelying on confident sounding
(09:49):
guesses.
When you ask for sources first,you're much more likely to
notice if something feels fishyor vague.
You won't be seduced by smoothlanguage.
You'll be focused on truth.
Second, it saves time, and thatmight sound counterintuitive.
Doesn't adding a step makethings slower?
Not really, because it's muchfaster to validate three sources
(10:11):
before you write than torewrite a message after
realizing something's wrong.
In most cases, you'll shave offrevision cycles, clarification
emails and back and forth withcolleagues.
Third, it builds muscle memory.
After a few repetitions, thisprocess becomes second nature.
You start spotting red flagsfaster, you trust your draft
(10:32):
more fully and, best of all, youstart sending AI-assisted
content with real confidence,not quiet anxiety.
Now let's talk about the secondpart of this process, the
one-click reference audit.
After ChatGPT gives you itssources, usually as URLs or
citations, take a minute toverify them.
Paste the links into yourbrowser.
(10:54):
See if they actually lead tothe source that's described.
If one fails, don't panic.
Just flag the sentence andeither adjust it or find a
replacement.
This isn't about being perfect.
It's about protecting yourcredibility.
You don't need to become afull-time fact checker.
You just need to stay in thedriver's seat One of the
simplest ways to spot trouble.
(11:16):
Watch out for citations thatsound almost right but don't
quite exist.
If a URL looks plausible butdoesn't lead anywhere, that's
your cue to dig deeper.
If ChatGPT references a statbut can't tell you where it came
from, treat it like aplaceholder until you can
confirm it.
Here's another real-world tip.
(11:36):
If you're pulling from your ownorganization's materials, like
annual reports, campaign updatesor board decks, drop those into
your ChatGPT prompt as part ofthe source list.
That way, the model hasgrounded content to draw from
and is less likely tohallucinate.
You're narrowing the focus,giving it guardrails and
dramatically increasing the oddsof a clean and accurate draft.
(12:00):
So let's recap the practicalprompts that power this approach
.
First, source first promptBefore writing list three
verifiable sources that showimpact statement and include
URLs.
Second, after you verify, nowwrite a 150-word donor update
(12:20):
based on the verified sourcesabove.
Keep it warm, professional andspecific, avoid generalities.
This structure worksbeautifully for a wide range of
communications, whether that'sdonor updates, a grant report,
board briefings, maybe afundraising appeal or even
year-end recaps.
And the best part Once you'veconfirmed the sources, once you
(12:41):
can often reuse the languageacross multiple messages or
segments with small tweaksinstead of the full rewrites,
you don't just save time, youreduce the cognitive load, you
stop wondering is this right?
And you start knowing yes, it'ssound, it's supported and it's
safe to send.
And when you're writing underpressure which, let's be honest,
(13:03):
is most of the time having thatconfidence is priceless.
So as you move into your nextdonor message, pause for a
second.
Ask yourself am I trusting theoutput because it looks good or
because I know it's grounded.
If it's the first, take theextra step, because accuracy
isn't a bonus, it's the baseline.
(13:24):
And when you pair thatcommitment with a tool like
ChatGPT, you don't just writewell, you lead well, you
fundraise with integrity andyour donors feel the difference.
Let's take a breath here,because this section isn't about
prompts or protocols.
It's about you and what mightbe running through your mind
even after learning the source.
(13:44):
Then write loop.
You might be thinking this allmakes sense.
I see the value, but doesn'tverifying AI output just cancel
out the time I was trying tosave in the first place?
And that question is honest andit deserves an honest answer.
Yes, adding a verification stepmeans spending another 60 to 90
seconds checking links orconfirming a statistic.
(14:06):
But let's zoom out.
What's the real cost of skippingthat check?
If a donor catches a factualerror in your message,
especially one tied to theirgift or program area, you don't
just lose face, you lose trust,you lose the sense of
reliability they felt when theyhanded you that check or they
wired those funds.
And rebuilding that trust thatdoesn't take 60 seconds.
(14:29):
That takes weeks, months, maybeeven years.
And it's rarely just one donorthe moment.
A stat is wrong in a boardreport or a fabricated ranking
slips into an appeal, the rippleeffect begins.
Donor, trust is relational,it's not transactional.
The word spreads, even subtly.
They're not buttoned up.
(14:50):
That number didn't check out,that wasn't accurate.
Those are dangerous impressionsto leave behind.
So if you're wondering whethera quick fact check slows you
down, I'd offer this it's not aslowdown, it's a safeguard, it's
speed that lasts, because thefastest route to credibility is
consistency, and that comes fromdetails that hold up.
(15:14):
Here's another shift that'soften necessary.
Many fundraisers, especiallyhigh-capacity performers, see
revisions or fact-checking as arework, as something to be
minimized.
And if you're using ChatGPT,there's a temptation to see its
fluency as final.
The writing sounds so good, sofast, that editing feels
(15:35):
optional.
But let's reframe that.
The goal isn't to cut editing,it's to edit smarter, to bring
discernment back to the surface,instead of letting the tool
lull you into copy-pasteconfidence.
When you ask for the sourc listfirst or run a link audit,
you're not adding steps, you'retaking back control.
(15:56):
This is especially important forthose of you who are trusted as
the voice of your shop.
You're the one who drafts themessages, the one leadership
counts on to get the languageright.
Your tone is your brand andyour accuracy is your integrity.
That responsibility can feelheavy, but it also gives you
power the power to model whatthoughtful, ethical AI use looks
(16:20):
like in the real world.
When you're the person who bothembraces the efficiency and
protects the truth, you set thetone for everyone else.
Your team learns that it'spossible to be fast and right at
the same time, that quality andspeed aren't opposites, they're
partners.
You also may be thinking, keith, do I really have the time to
(16:43):
train my whole team in this?
If you're stretched, I hearthat, but here's the shift.
You don't need to become the AIgatekeeper.
You need to become the culturesetter.
If you build these habits intoyour prompts, your workflows,
your templates, and share thatwith your teams, others will
follow your lead.
You can even start withsomething as small as a shared
(17:04):
checklist.
Add placeholders for anysensitive data, ask for sources
before writing, Confirm everycitation, customize tone after
verifying content.
That's it.
That's your culture of accuracybuilt into behavior.
And as this mindset starts tosink in, you'll notice something
unexpected your confidencegrows.
(17:25):
You're no longer wondering ifthat paragraph you copied from
ChatGPT will hold up underscrutiny.
You know it will, because youguided it, you checked it and
you owned it.
That's the real reward of thismindset shift.
It's not just about reducingerrors or preventing
embarrassment.
It's about showing up withclarity and confidence.
(17:45):
It's about knowing that everymessage you send reflects your
integrity, not just yourproductivity.
So pause and ask yourself whatwould change if I trusted myself
more than I trusted the tool.
What would shift if accuracywasn't just a back-end
correction but a front-endfilter for everything I write.
(18:06):
Let that land, because whenyou're rooted in stewardship of
time, of truth, of trust, youwrite differently, you lead
differently and your donors feelthe difference.
So today we tackled one of themost important questions in
AI-assisted fundraising Can Itrust the output?
(18:28):
We explored why large languagemodels like ChatGPT sometimes
hallucinate, how those errorshappen and why trust with donors
depends on your ability toseparate confident-sounding
fiction from verifiable fact.
We covered thesource-then-write loop, a simple
, repeatable workflow that asksfor citations before drafting
(18:49):
copy, and the five-word safetycheck Cite sources for every
claim.
These aren't just techniques.
They're habits that protect theintegrity of your message, your
brand and your donorrelationships.
And we talked mindset.
Accuracy isn't the enemy ofspeed.
It's what gives speed realstaying power.
(19:09):
When you verify before youwrite, you send messages with
confidence instead of crossingyour fingers.
Now, if this episode helped youfeel more confident about using
ChatGPT in your donorcommunications, there's one
small thing you can do thatmakes a big difference.
Would you please leave a ratingor review?
Wherever you listen, it helpsmore fundraisers discover the
(19:31):
show and join you in buildingdonor trust with intention and
integrity.
Next week, we'll startexploring workflow wins and time
savers.
We'll start with the inbox howto train ChatGPT to reply to
routine donor emails and giveyou back your time.
Until then, stay curious, stayclear and keep fundraising with
(19:56):
heart.
I'll see you real soon.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Special Summer Offer: Exclusively on Apple Podcasts, try our Dateline Premium subscription completely free for one month! With Dateline Premium, you get every episode ad-free plus exclusive bonus content.
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!