Ep 296 | Using ChatGPT Correctly with Steven Lewis - Digital Marketing Therapy

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Sami Bedell-Mulhern (00:00):
Listen. AI is everywhere. Everybody's
talking about chat GPT or yourfavorite AI tool. And so I
thought it was time that we havea conversation about how you can
use chat GPT to be kind of thatextra arm in your business. Now,
I love this as a topic, and as Ikind of jump into it more, have
more conversations with people.I'm always fascinated by the

(00:21):
unique ways that people areusing chat GPT to streamline
their business and do thingsthat I had never even heard of.
Stephen Lewis is my guest today,and he also has some incredible
ideas for how you can use chatGPT to just become that extra
person that is on your team thatis helping you with the

(00:41):
different areas of yourbusiness, from HR to market
research to content creation,project management, all of the
things AI can be really, reallyimpactful. Stephen Lewis is a
marketer with 30 years ofexperience in copywriting and
technology. His course, makechat GPT your cmo shows business
owners how to turn chat GPT intoa 24/7 marketing consultant that

(01:02):
gives expert advice tailored totheir businesses and as a small
nonprofit or a lean team, thisis a game changer when it comes
to how do we increase, grow,build our content machine
without having to hire somebodyor without really understanding
all of the things that we needto know. And what I love about

(01:23):
Steven's experience is that,because he is a copywriter, he
knows how to really take chatGPT and turn it into the machine
that you need to be authentic,to show up genuinely, and to be
there for your audience asyou're kind of walking through
brand awareness encouraging themto engage and eventually to
donate or pay for your productsand services.

(01:44):
So that's what we're going totalk about today, and it is such
a great conversation. I'm Irecorded this a while ago, and
I'm still like thinking aboutall the things and the nuggets
that he shared, and how I canactually put them into my day to
day business. So I'm working onthis all alongside you as well.
Before we get into it, thisepisode is brought to you by do

(02:05):
good university. Do gooduniversity is a unique
membership for nonprofitfundraisers and development
professionals who really justneed some support in all things
nonprofits, fundraising ideas,event ideas, marketing support.
And what's even better is we golive every single week, myself
and my partner in crime, PatrickKirby, and we answer your

(02:27):
questions live. So it's a greatopportunity for you to come
together in community with othernonprofit organizations from
across the country that aregoing through what you're going
through. You can ask questions,get answers, or simply just
listen and get inspired by theconversations that you're
hearing from other people thatare working on their events,
that are working on theircapital campaigns, that are
working on major gifts, or justsocial media marketing, email

(02:49):
stuff in general. So if you wanttwo weeks free inside of do good
university, all you have to dois go to thefirstclick.net/dgu,
check it out. You can watch thebonus podcast episode that
Patrick and I did talking aboutcommunity and nonprofits. But on
that page, there is a link to,uh, sign up for your two week

(03:10):
trial for free. Come check usout. Like I said, we're live
every single week. You can joinus, and we'd be happy to walk
you through some of the ondemand opportunities that are
inside of DGU as well, so thatyou can take your fundraising to
the next level again. That isthefirstclick.net/dgu. Let's get
into the episode.

(03:31):
You're listening to the digitalmarketing therapy podcast. I'm
your host, Sami Bedell-Mulhern,each month we dive deep into a
digital marketing or fundraisingstrategy that you can implement
in your organization. Each week,you'll hear from guest experts,
nonprofits and myself on bestpractices, tips and resources to
help you raise more money onlineand reach your organizational

(03:54):
goals.
Hey friends, please join me inwelcoming Steven Lewis to the
podcast. Steven, thanks forbeing here today.

Steven Lewis (04:00):
It's an absolute pleasure. Thank you for having
me,

And we're talking all things chatgpt and
AI, kind of feels like thisthing that's everywhere, and
it's permeating every singlething that we do in our life.
But for a lot of us, we don'tactually know how to use it, why
to use it, what it's for. Sobefore we kind of jump into
specifics, why is this a topicthat you enjoy learning about

(04:24):
and sharing with people?

Long story short, I combine two backgrounds, so I
started building websites in1994 which, if anybody is you
know, old enough to remember,not many people even knew what a
website was, so I startedbuilding them, and I'd explain
to people what they were, andthen people say, I've heard of
this thing called email. What'sthat? So that is my level of

(04:53):
interest in technology. I liketo be there first, but I'm also
a writer, so for 30 years I'vebeen a professional. I've been a
copywriter, so my job ispersuading people to do
something. That is whatcopywriting is. It's not you
want you need to entertainpeople. You need to get people
to read what you're saying. Youneed them to feel informed and

(05:13):
educated. That's what I used todo as a journalist. But then
you're taking that extra leap ofpersuading people. And when
chatgpt came along, or evenbefore that, actually a couple
of years before chatgpt, we wereusing a tool called Jasper. They
purport to be copywriting tools.That's their whole offer that

(05:34):
I'm coming for your job. So thistool is out there. It says it's
coming for my job. The media isall about it. I thought, well, I
need to learn what this is,because it's the threat. And
then I realized it was the mostphenomenal tool to help me be
better at my job. It could notdo my job, but it could make me

(05:57):
so much better at it. And thatgot me so excited that I did
what has always been the patternof my life, which is, I learn it
and then I teach it. So when Ilearned how to build websites, I
became a journalist who wroteabout the internet, and again,
at a time when you wereexplaining, what is the
internet, where we still spelledout, www.we, still said the

(06:23):
World Wide Web out loud. Nobodysaid the web, because nobody
would have known what the webwas. So that's been my pattern,
and I'm so excited about whatchat GPT can do to help people
whatever their job, be better attheir job, but particularly if
your job involves writing andpersuading people you need to

(06:45):
know how to use chat GPTproperly, not the way that other
people use it properly,

yeah, well, I first of all love your take on
technology and your kind ofwillingness to learn. We did
another podcast episode a longtime ago. Might have been like
even a year and a half, butwe'll link it up in the show
notes, about like, how to bringyour organization into
technology, how to talk topeople on your team, about like,
embracing technology, because toyour point, it's not going

(07:13):
anywhere, and if you learn aboutit, it's just going to help you.
To your second point on if yourjob includes persuading people
or sales. I mean, I think in anyjob that is the case, whether
it's trying to figure out how toget your boss to do this new
idea, or how to get your childto clean up their room, or like,
like, it's just reallyinteresting, like that. There is

(07:34):
nobody that isn't doing that insome way, shape or form in their
job. And so I love that you saidthat, but I think anybody can
use chatgpt for all sorts ofthings, even if it's not work
related.

I think a lot of people don't realize that
everything is persuasion. So Iused to work briefly in an
internal communications team ina bank, and what my colleagues
didn't really understand wasjust because you're writing to
people who are paid by the bankto work there doesn't mean you

(08:08):
don't have to persuade them toread the stuff that you send.
Even to open the email. Justbecause the email comes from the
CEO doesn't mean anybody's goingto open it. You have to persuade
them that it's worth openingeverything is persuasion. And
everybody, nobody likes the wordselling. Everybody is selling

(08:31):
something, even if it's I'mselling myself. For you to pay
attention to when you get homein the evening, frankly, you're
selling everything through thatlens is selling and chatgpt can
help you. I've got a talk thatI'm giving at my daughter's
school in April, which is how topersuade your parents to do

(08:55):
whatever you want them to dowithout and have them be happy
about it, using chat GPT likethat's the that's that's the
talk that I'm giving. How to usechat GPT to get your parents to
bend to your will.

That's so fun. And talk about the right
title for the right targetaudience, right? But let's
before we get too far into this.For those that are listening to
this and you've maybe heard theterm chat GPT, or you've kind of
seen it places, but have nevertouched it, or don't really know
what it is. Could you give kindof just the basic rundown of
what chat GPT is?

Chat GPT is what they call a large language
model, which, again, is leakinginto the world such that you may
see even newspapers now just sayLLM, as if we all know what a
large language model is,chatgpt, it's an LLM. Large
language model just means it's amachine that has read an

(09:51):
unimaginable amount of material.So you imagine that it's read
your website and every otherwebsite in the world it's read.
Every book that they could feedinto it, every blog post, every
forum post, every academicjournal, anything that its
makers could get their hands on.OpenAI, the makers of chatgpt

(10:12):
and all these large languagemodels have fed that into the
beast. So you train the beast onthat so that the beast can then
write on pretty much any topic.So that's that's the AI part.
You say, write something, andthe Beast goes into the belly of
its knowledge and startswriting. And it looks like it's

(10:33):
super intelligent and it'sthinking, but it's actually
writing statistically, one wordat a time. So it writes the
first word, and then it says,what would the average human say
next in this context? So ifyou've said, please write me an
email to my boss explaining thatI'm not coming in for the next
week. It will say, Okay, right?What would the average human

(10:55):
right? After this word, thenit's got those words, what would
it write next? Which is why youget super bland writing, because
the AI is asking quiteliterally, what would the
average person say now? So theskill with chat GPT is being
like a potter. So you've got theclay on the wheel, and you think

(11:18):
of chat GPT is the wheel, andyou've got to be there shaping
it. So the more context you givechatgpt, the better its writing
will be. So if you say to it,for instance, my tone of voice
is this, and then you describeyour tone of voice, you've
immediately changed the contextso that it will write
differently. Then you could say,and I'm writing to somebody

(11:42):
who's six foot tall, lives inthe Midwest, drives a truck, but
also enjoys ballroom dancing.Well, immediately you're
changing all of that contextthat chat GPT is applying. So
what you've got is this huge,unimaginable library of
knowledge being dipped into bythis incredibly intelligent

(12:05):
computer controlled by you, andif you know how to control it,
you are a human being sitting ontop of the world's largest ever
library and most intelligentever computer, Getting it to
help you to do your job. But thething that is not really widely
explained is, how do you getgood at being the driver of that

(12:29):
machine? Because it isn'ttyping, please write me a
LinkedIn post. Please write me aflyer about my program. That is
not how you get the best out ofit.

Well, and I feel like when it first started
coming out, or when people firstuse it, it looks very much like
a search engine, you know, whereyou just type so it almost feels
like you should just be puttingin prompts, as if you're
searching online, but really,you can have a whole
conversation with it, right? Soit's not just I put in my
prompt, it spit something out,and then we're done with our our

(13:03):
day. So even if you start with asimplistic prompt, you know, we
can still kind of refine andstart to add more things into
it. I mean, you can really havea conversation with with it. And

and it gets better, because chat GPT
operates on something called thecontext window, which means when
you're in a chat with chat GPT,when you type something into
that prompt box, that prompt boxis disastrous, because that
prompt box is this skinny littlething at the bottom, yeah, that
implies is short, yes, oh, Ishould type in one sentence.

(13:36):
It's like if you had a robotsurgeon, right? That you had a
button that said, push here fortaking your appendix out. And
you're like, Oh yeah, great.I'll just put my kid on the
table and I'll hit that button.You're like, No, do not touch
the button unless you know whatyou're doing, right? It's not
that simple. So the prompt boxis that thin, but I write

(13:56):
chatgpt prompts that could be300 words long. Now I don't
write a 300 word long promptnecessarily. Every time I have a
library of them, I keep them. Iwork on a good prompt and I keep
them. But to your example, let'ssay you type in that one
sentence and say, Please writeme an email about a program that
I'm running, and it writessomething. And you say, Okay, I

(14:18):
don't really like that. Thenabsolutely you can say, look, I
like where you're going withthat, just as you would to a
colleague. You know, little bitof encouragement goes a long
way. Oh, you know, I like whatyou did with that. I see where
you were going actually. I hateit, you know, like, I'm English,
right? So that's what we do allthe time. We go, oh, that's an
interesting approach, whichmeans, to an English person, I

(14:40):
absolutely loathe what you'vedone. Oh, interesting approach,
so you give it a little bit ofprice. Oh, you know, that's
nice. I really see where you'regoing. Thanks for doing that for
me. Because one chat GPTresponds to flattery. I'm not
joking like there's, there's atip that's that's been proved in
academic research. They don'tknow why, but large language.
Models like to be flattered.They work harder for you,

(15:03):
interesting, flatter them, actgenuine. Fact, you I don't like
that. But paragraph one, couldit be a bit shorter? Would you
make it a little bit longer?Like, actually, you've got a
small detail of the programwrong? It's actually this. So
first is a mindset shift ofyou're not it's not one and

(15:23):
done. It's not like you know,it's not like you load the gun,
you fire and the bullet eitherhits the target or it doesn't as
you say. It's a conversation,and you will have a much more
fruitful conversation with chatGPT, if you think of it as a
very intelligent but juniorcolleague who has only just

(15:45):
joined the organization. So ifyou have somebody starting your
business on Monday, you wouldn'tjust say, Sarah. Write, write,
write me a flyer about thisprogram, and then swivel in your
chair and leave Sarah to do herown thing while you go off to
the kitchen and make yourself acup of coffee, because Sarah's

(16:07):
got it. No, she doesn't. Sheneeds some briefing, brief chat.
GPT,

well, and even to take that analogy a step
further, if you were writing iton your own, even as a seasoned
copywriter, you don't writethings one time and be done
with. You don't just write thefirst draft and say, This is
brilliant. We're going to goahead and send it off to the
press like I'm done. So how canyou expect AI to do that for you
as well? Because that's just nothow you would work.

That's absolutely true. And also this takes you to
a really interesting point. Isthat I said that chat GPT has
read all of this stuff. It'sunderstood all of this stuff,
and that, I think doesn't reallycome across. I think people
think I can get into chat GPTand I can say, write me a write
me a flyer or an email or aLinkedIn post or whatever it

(16:54):
might be. But even if you never,ever like I understand there are
people who are writersprofessionally. You know, maybe
they've been writing fundraisingcampaigns for a really long
time, and they're like, I knowmy organization. I'm a
fundraising writer. I know howto do this. I will never
diminish myself by sendingsomething out that was written
by a robot. And I get that. Iget that point of view. I've

(17:17):
been a professional writer for30 years. It's how I've made my
living. So I understand that,even if you don't like that thin
AI, as I say, is this librarythat you can tap into. So you
can think, Okay, I'm never, I'mnever going to let you write for
me, robot. But you know somethings, why don't I work on
that? And if you know how to getinto chat, GPT is understanding

(17:42):
say of people, then you couldhelp your writing to be much
more persuasive. So, I mean, Ido all sorts of things like
develop avatars of personasusing what chat GPT knows about
people. But if you want say asimple example, for instance,
you could say to chat GPT, chatGPT. My organization does this.

(18:03):
Our typical reader of ourmaterial is that, what can you
tell me about their psychologywhen it comes to material from
us? Now that is an incrediblysimplistic prompt, but you will
get depth from that. Then youcan use that in what you're
writing. So for instance, if youthen did want to write and we

(18:25):
use our please write about myprogram. Example, you can say,
please write about my program tothese people that chat GPT has
just told you about. So exactlyas you're saying. As a
copywriter, I wouldn't juststart give a prompt and leave
it. But equally, I wouldn'tstart with no brief. My brief
would be, this is myorganization. These are my

(18:47):
goals. This is who I'm writingto. This is why they care about
us. This is why they mightignore us and give money to
somebody else. You know, all ofthose things are. As a
copywriter, if I came to yourorganization, I'd want to know
that. What do you know about theperson reading what I'm writing?
And chat GPT knows, and if youknow how to ask it, there's

(19:10):
gold. I mean, there's absolutegold. Like you wouldn't believe
what you could get chat GPT todo for you well. And I

really want to double down on this real
quick, because I've heard thisfrom a couple other people in
the chat GPT space as well, andI want to make sure that people
hear what you said, because Ithink it's critically important,
prompts matter, and you don'thave to come up with them on
your own. I love that you'retalking about letting chat GPT
help you build those prompts andbuild those responses and help

(19:39):
you understand how to use themachine, just by asking
questions and going from there,kind of taking the different
layers of okay, well, if I'mjust at the part where I'm
trying to figure out who mypersonas are, start with the
basic questions and kind ofbuild from there. I think that's
critically important and such agreat way for people to use this
tool, especially if you're not acopywriter, especially. If

(20:00):
you're in a small team,especially if you can't afford a
huge consultant. And so I justwant to make sure we don't gloss
over that, because I think, yes,it's good for content generation
and things of that, but it'salso good for helping you figure
out how you want to create andcraft the prompts like, let it
help you do that. I

think one of the this little prompt window that
we're talking about being thesort of dangerous is because
people then think, I'm I'm afool if I can't work out how to
use it, because obviously itmust be so simple, because all
they give you is this littlewindow so they must think it's

(20:39):
simple. You know, like whenApple sends you a product and
there's no instruction manualanymore, and and my mother says
to me, when she gets a newiPhone, she's like, how am I
supposed to work this out? Andthen you're like, well, Apple
just thinks you will, or you'reGoogle, like, they don't. They
don't see it as their problemanymore. They've sent you a
thing in a nice box, and that'schat GPT. It's a thing in a nice
box with this window. But lasttime I was in in Vegas, the

(21:04):
thing that blew me away that I'dnot noticed, I hadn't been for a
long time and hadn't noticed,was, you've got the strip, and
then one block over is nothinglike, it's like, you know, like
a bail bondsman and a, you know,like a, you know, lawyers for

(21:25):
people who've been locked up,and the bail bondsman next door,
like so you've got the strip,which is world famous.
Everybody's seen that, droneshots, helicopter shots. We've
all watched honeymoon, you know,we've all seen the strip. And
then one block over, it'snothing. It's a barely
illuminated street full of bailbondsmen. If you are just typing

(21:47):
into chat GPT things that youcan think of, you're probably
walking down the bail bondsmanpart of Vegas thinking. Everyone
talks about Vegas, but it's abit rubbish. Like, there's
nothing, but bails bondsman andcrappy motels. Like, what's
happening? And you're like, No,what block over the Strip? The

(22:10):
strip? People like, what's thisstrip you're talking about? I
keep reading about the strip inthe media, and this is why I
think, I mean, obviously I runcourses on chat, GPT, so I would
say this, but to go and findsomebody to show you what it can
do is, is like somebody takingyou from the back of the strip,

(22:31):
going, I don't understand this.It's just rubbish bins and
service entrances to the stripitself again. Oh, my God, I get
it, and my life has changedforever because I teach people
how to use chat GPT as aconsultant. So as I say, whether
they ever write anything or not,your team, and a lot of my
clients, are teams of one. It isa very lonely team. They are

(22:54):
everything, and even if theywork in an organization, they're
the one person in theirorganization who does what they
do. So it's lonely. They've gotno one to bounce ideas off, even
if they work in a small team,nobody has time for you to
swivel in your chair and say,Hey, I'm working on something at
the moment. Could you dropeverything you're doing and

(23:15):
spend a couple of hours with medoing what I'm doing? No, I
can't. I've got my own job, butif you know how to use chat GPT
as a consultant, you have thatincredible thought partner,
where it can't do what you do.You can't do what it does, but
together, it's like Iron Man,right? You've stepped into the

(23:36):
suit. It's the whole it's thewhole thing,

yeah? Well, so this leads me to, you know, I
think another reason why somepeople stay away from chat GPT
or AI are security concerns. Iknow it's been interesting. When
I look at my analytics, Iconsistently have strong
referrals from chat GPT to mywebsite, which I don't you know

(24:01):
that it's showing up in people'sresults and whatnot. That's not
something I'm controlling ortrying to do. So kind of, what
does not, I know it's a like, wecould spend a whole episode on
security and AI, but just kindof top level, you know, as we
think about a paid versus a freeaccount, kind of proprietary
information, like as we're kindof dipping our toe into this and

(24:23):
having conversations or playingwith with the conversations
inside of chat GPT, what mightwe want to consider at a very
basic level before kind of goingall in

chat GPT will tell you at various levels of account
that it doesn't use your whatyou put in doesn't go into the
belly of the beast. It juststays in your conversation. Big
technology. Big tech does nothave what you'd call a super

(24:55):
sunny track record at honoringwhat it tells you it's. Going to
do. So I'm not saying that openAI doesn't honor that, but even
if open AI does honor that,which one has to suppose they
do, it only takes one engineertinkering around under the hood

(25:17):
to flick a switch accidentally,and your stuff falls through the
trap door into the belly of thebees. So you do have to be aware
of that as a risk. So I wouldn'tbe putting my DNA profile into
it, for example. But on theother hand, most people's, and

(25:40):
I'm not talking about your donordetails, for instance, you know
their addresses and their socialsecurity numbers and so on. But
most people's quote, unquoteproprietary information isn't
really all that proprietary,right? Like it's not the formula
for Coca Cola. So you might inyour office to make yourselves

(26:02):
all feel very good, be talkingabout your proprietary processes
and data, and you know, yourspecial source. And I don't mean
to diminish that, because I'vegot my own special source, so I
understand that. But does anyother copywriter do kind of what
I do in the order in which I doit with a yes, they do. So I

(26:22):
would be looking for a sensible,rational balance between what is
truly holy shit if weaccidentally lost that, like our
donor list or somebody's socialsecurity number or an employee's
health information, you know,like no that that's not that's
not cool. But if you're writingfundraising material, for

(26:43):
instance, that is not going tobe super proprietary. And most
of you know, most of my clients,for instance, work in marketing.
So you know, fundraising being aform of marketing, obviously
what we're writing, what we'reworking on, is intended
ultimately to get out to thepublic. So the inputs, generally
speaking, aren't things thatwould be disastrous to us if

(27:06):
they came out to the public,like, Oh, we don't really give
the money to the children. Youknow? It's not that, it's not
that kind of stuff, right? It'sinputs in. Like, we did really
good work. We're putting thatin. We got great results. We're
putting that in. We're having anevent on this day we're putting
that in, so I would be lookingto form a sensible policy on

(27:28):
what is really super secret.What would we be devastated if
the newspaper got hold of thefact that we'd drop that into
the belly of chatgpt, and whatis actually not disastrous if it
gets out there and using that sothinking, and once you and then
I think you have to be careful,because once you start using

(27:49):
chat GPT as a colleague, likereally understanding its ability
to be your colleague, you willwant to put everything into it,
because you will not want to runanything through chat, you know,
through your world, withoutasking chat GPT for its opinion,
which I'm I will shut up in onesecond. Just today, we haven't
even talked about putting thingsinto chat GPT and asking it to

(28:10):
give you an opinion back, evenif it doesn't write say, I bring
this. What do you think of that?

Yeah, well, I think there was a whole trend.
Wasn't there, like, at the endof the year, where people were
putting things into chat, GPT,about, like, tell me, tell me
who I am, or tell me what myyear is going to look like, or
tell me what 2025 is going tobring to me. Like, like, there's
all those sorts of trends, ofthings that people kind of
create and are surprised by whatit spits out. I think the

(28:39):
opportunity for it is endlesswhen you start to just play with
it. But I think that's the pointyou're making, is you just have
to kind of start playing with itand kind of staying on top of
it. And also, I love that youmentioned earlier, like saving
those prompts so that you canreuse them later if it's a
repeated task that you're doing,so you don't have to create them

(29:00):
over and over. And so I thinkkind of one of the there's we
could we might have to have youback for another episode,
because I feel like there's somany other things that we can
talk about. But the other thingthat I think people get stuck on
is my tone of voice. I don'tknow how to explain my tone of
voice, like, I feel like that'sanother buzzword that we use all

(29:20):
the time, but don't reallyexplain. So could you kind of
maybe give people just a quickrundown of if I'm trying to
figure out what that is, I knowyou already mentioned, just
starting to have conversations,you know, with with the search,
with the engine, and see whathappens. But kind of, how do we
start to develop that tone ofvoice if we don't have one
already, tone

of voice is a fantastic is a fantastic place
to start, because, as I said,chatgpt writes average stuff in
a context. So if you give it areally good tone of voice, and
by good tone of voice, I mean adescription of your tone of
voice that's a couple of 100words long. I don't mean I'm

(29:58):
friendly, thank you. That's mytone. Voice, I'm professional
and approachable, so there are acouple of ways to train it, and
I actually have a free minicourse. It's a five minute
video. I call it a mini coursebecause that sounds so
impressive, but it's a fiveminute video that will show you
how to create a one way tocreate a tone of voice, and that

(30:23):
way is to take the things thatyou've already written that you
like and feed them through aprompt that asks chat GPT to
divine your tone of voice. Andwhat you could do is take this
short course and test. It's agood test of the little write me
a tone of voice based on thisversus Stephen's prompt,

(30:44):
generating a tone of voice, andyou'll see the difference
between a really good, detailedprompt and a one line, tippity
tap prompt. So what I would doas a starting point is gather
material you like that iswritten the way that you want it
to sound, and then use myprompt. Or if you just want to

(31:07):
type, tell me the tone of voiceof this that that would do it,
but it's not going to be as goodas my prompt, and it will tell
you your tone of voice. And ifyou don't have anything that's
written the way you really wantit to sound. Find somebody who
writes the way that you wantthem to write. So let's be

(31:28):
honest. In your game, thereprobably is somebody who does
what you do like you know, noneof us is a unicorn, right? So
find the organization that youthink that your boss is always
saying, I really like the waythey do it. Oh, they're so good.
Their stuff is amazing. Theysound incredible. Go and find

(31:50):
that and feed that in and say,what is that tone of voice?
Because there's no copyright.And I understand some writers
will be thinking, oh no, but I'mplagiarizing. I don't want to
plagiarize the otherorganization. No, you're going
to say things in this tone ofvoice that are your things, your
original things. There's nocopyright in a tone of voice. If

(32:11):
I wanted to go out and soundlike Coca Cola sounds or
Microsoft sounds or Applesounds, I can absolutely go out
and do that. I can't say Iinvented the iPhone. That would
be a lie. But I could writeabout my chat GBT courses in the
tone of voice of Apple, thatwould be perfectly legitimate.
So you don't even have to findanother organization that does

(32:34):
what you do. You might think,you know what I really like the
way Apple describes itself, or,you know, I don't a politician
that you like, for example, whoyou think you know they speak
really clearly andauthoritatively, and I like the
way they come across. Will goand find their material, their

(32:54):
speeches, and feed those in. Andremember that what you get at
the end of this is editable. Soyou can say, Yeah, okay, you've
described that politician, forexample, tone of voice in this
way. I like that. That's a gooddescription of that person. But
we actually want to be softer,or we want to be funnier, or we

(33:15):
want to be lighter. You can typethat in. You don't have to go,
oh God, chat. GPT has given methis. I'm stuck with it. Put it
in a Word document and then editit, and then you've got a tone
of voice prompt that you can useover and over again. So every
time you want to write somethingon behalf of the organization,

(33:36):
you just say, hey, chat, GPT, Iwant you to write this. And this
is the tone of voice that I wantyou to use, and it's game
changing for people, becausethey suddenly see that it's not
set in stone how chatgpt writeswhich, as we know, is bland, as
we know, chatgpt loves an emoji.Chatgpt has never met an emoji

(33:57):
it doesn't want to use. Well, ifyou're fundraising for people in
distress or a worthwhile cause,you probably don't want a lot of
thumbs up or explosion emojis.So you can say in your tone of
voice, please don't use emojis,right? You know, it's that sort
of stuff that you can you canweave in. So that's what I would

(34:18):
say. Feed things in that youlike, and get chatgpt to
describe the tone of voice. Itis much faster than you trying
to describe the tone of voiceyourself.

I love how easy that is. I will challenge
listeners to remove the termwhat you like. I'm going to just
push back on that wordspecifically, just for a second,
because I think a lot of timesas copywriters, content
creators, or when we live in ourworld, there's things that we
like that maybe don'tnecessarily resonate with our
audiences as much. So take ifyou're looking for words that

(34:53):
you want to pop in, or languagethat you've already written, I
would take stuff that performswell, what emails got open. The
most and clicked on the most,which event email actually got
the most clicks through to yourwebsite, to where people
actually made a purchase? Whatlanding pages on your website
work the best, what social mediaposts resonated the most? So

(35:15):
like your statement was perfectand amazing and like super
simplified, I just know so manynonprofits tend to make
marketing decisions based off ofthat one board member that's the
loudest in the room, or the oneperson on the team who is like,
you know, oh, well, this is howI use things, so therefore
that's how our entire audienceuses things. So I like as a

(35:37):
copywriter, for sure. You knowwhat's working well. But I just
want to challenge you aslisteners, don't just go with
what you like, take what Stevensaid perfectly, but pay
attention to the data and usewhat performs well with your
audience so that you don't hitthat mark. So I just wanted to
kind of throw that well there,so we don't get lost in our
personal preferences. I

mean, one thing that I teach people to do that I
think would be really useful inthis circumstance, because I
think we've all met that boardmember, right? And it could be a
board member could be acolleague. There's always
somebody who, as you say, isvery comfortable saying things
like most people, and you'rethinking, No, you mean you like
you have no data. Your data isyou, most people that, to me, is

(36:23):
the reddest of all red flagswhen somebody tells me what most
people think, but what you cando, and this is super advanced
and what I'm saying, I just wantto give people an idea of the
destination to which you canget. If somebody can show you
how to get there, is you cancreate a very detailed picture
of your reader using chat GPT,because it knows them then, and

(36:47):
this blows people's mind when Iwhen I show them how to do it.
You can run a focus group withthose readers. So as an
organization, you might have anideal donor who is this right,
but you might have a secondaryideal donor who's maybe not as
much, that maybe they don't haveas much money, or they're not as
philanthropic, or whatever itmight be. You can have a panel

(37:09):
of them, say, three of them in aroom, and you can talk to them,
which sounds mad that you canhave three avatars in a room,
and you can say, Hey, I'vewritten this email. What do you
guys think? I don't writeanything now I'm right now. I'm
writing a landing page. That'smy big project at the moment, a
sales page for a product. I'mwriting the whole thing in

(37:30):
consultation with two idealreaders. So every time I write a
paragraph, I put that to them,and they come back to me, and
they're different. So they havedifferent agendas, and they will
give me feedback, but to yourboard member, the beauty of a
process like that is to be ableto say to your board member, I
absolutely get that in yourresearch. You've discovered that

(37:52):
most people, but we ran a focusgroup with our avatars, and we
put this to them, and this ishow they responded. And I know
this sounds mad, but this isacademically verified that large
language models, which you nowhaving listened to this, will
refer to immediately as an LLM.You'll be like, don't say large
language models. That's likesaying World Wide Web, it's an

(38:16):
LLM. No llms produce valid,insightful and deep answers that
are equivalent to the answersthat equivalent humans would
give. It, as I say, validated byacademic research. So you then
have, when you're in that boardmeeting, or you're dealing with
that difficult colleague whoknows what everybody thinks,

(38:37):
you've actually got data thatyou can go into. So it's not you
saying, I think this, and themsaying, Well, I think that. And
most people think the same asme, you've actually got data.
These are the kinds of thingsthat if you learn how to use
chat GPT properly, you canachieve. And a paid chat GPT
account is $20 a month. So youcan get all these tools that

(39:00):
want to, you know, charge you$99 a month to do like, one bit
of the puzzle. But if you knowhow to drive chat GPT, for $20 a
month, you can do it all. I

want to be explicitly clear here, so you're
saying that you have you're notdoing this focus group with real
humans. You've just got theseavatars that are created inside
of chat, chat GPT that areconsistent, that you have really
honed in on. This is exactly whowe're talking to, so that you
can feed at your content and getreal results. Yeah, so it takes
you a matter of seconds whenyou're trying to edit your copy.

(39:34):
Oh,

it is absolutely phenomenal. So yes, you are
right. The academic term forthese avatars, so people have
now absorbed large language.Model is LLM. Avatar is a
synthetic persona, and they talkabout synthetic personas and
synthetic data. And yes, you aretalking about inside your
computer, inside chat, GPT. Youare hosting a focus group where

(39:59):
you can ask. Thing you want.Because if you got your best
donor in the room, you probablywouldn't want to say we're
thinking about doing, you know,cutting this program and doing
this thing. In case the donorwent, That's horrendous. Don't
tell me that. Or I know from notfor profit clients that people
get terribly involved in howmuch your admin costs. They

(40:20):
always want to know, where is mymoney going on to the cause or
to the paper clip. And you just,I'll be honest with you, because
I was on the committee of theRSPCA when I lived in Hong Kong,
so animals. And we got thatquestion all the time, and I
did. I'm not a violent person,so I'm only meaning
metaphorically. I did want togive the person a little tap,

(40:40):
tap, tap, tap, tap, because weemployed vets, and they would
say that our vets were paid toomuch. And I would be thinking,
so our vets are supposed to takea big, fat pay cut, so
essentially, donate 1000s ofdollars a year to the to the
cause by taking a pay cut,because you think they should be
I say, how much of your salaryAre you donating every year?

(41:01):
Like, do you take a 20% pay cutso you can give? No, you don't
tip. Tap, tap, Tip, tap, Tip,tap, I don't like it. I don't
like the argument. But youwouldn't have wanted that donor
to come into a room and for youto say, listen, we're about to
give our vets a 10% pay rise.What do you think? And for the
guy to say, you know, you don'twant to you don't want to

(41:21):
discuss it. The beauty of havinga synthetic persona is you can
ask it anything you want and itwill give you a valid answer,
but you're also not affectingthe genuine relationship with a
genuine donor. So you can do allkinds of research with these
synthetic personas withoutworrying like, let's say you

(41:41):
wanted to change the name ofyour organization, where you
might not decide to do it, soyou might not want to discuss
with donors that we're thinkingabout grabbing ourselves this,
and the donors don't like it, orthey do like it, and then you
ultimately decide you don't wantto do it, and they're like, Oh,
but I told you, most peoplewould love it. So you can do all
kinds of things with thesesynthetic personas, with safety,

(42:05):
because when you know,particularly, I think, in a not
for profit, life is not safe alot of the time, so the more
safe you can and you know youdon't necessarily want to be
bold, unless you're sure thatit's going to work, or to change
things, unless You're sure thatit's going to be accepted. And
this can give you a level ofconfidence, because chat GPT

(42:26):
again when you know how to doit, can run the world. Can run
these things for you in anincredible way, and, as I say,
completely validated.

There's such a I would have never
thought to use it that way, butthat's so smart, especially for
smaller teams. As you'rebrainstorming and working
through things before you kindof have to make those decisions
or do things that are going toreflect on your donors or make
them scared or skittish or yourboard members. So I think that's
wonderful and such a great placeto kind of wrap this up. I mean,

(42:58):
we could talk. I mean, we'vebarely scratched the surface of
what can happen within chatgpt,but I hope this episode has
really encouraged you to jump inand try it and see what happens,
and see how it can support youin all of these different ways.
Steven, I know you mentionedyou've got a free resource, so
we'll make sure to get thatlinked up in the show notes
here. But if people want toconnect with you and learn more

(43:19):
about how you supportorganizations and how to connect
with you. How can they do

that? The agency that I run is called Taleist.
Tale as in telling tales,because we're storytellers,
ultimately, so it's T, a, l, e,i s, t, dot agency, and there's
lots of details about how wework there, and also my email
address, or there are emailaddresses all over the site. So

(43:45):
you can get in touch and askwhat, what we can do with chat
GPT, or what courses we havethat would level you up in I
mean, just a little plug for thecourse. I get an email every day
from people who take we have acourse called make chat GPT your
CMO. I've run a lot of coursesin my life. People have never
used the words life changing asoften as they have for this. And

(44:10):
the reason people find it lifechanging, and so I'm not
plugging the course. I'mplugging learning how to use
chat. GPT, the reason I find itlife changing is because people
are tired and having an alwayson energetic, intelligent
colleague to bounce ideas offand to give you insights and to

(44:30):
show you new ideas, or whysomething might not have worked
as well as you hoped. It's justbreathing new life into lots of
people I know who are beatendown by circumstances externally
or how tough their job is.Suddenly, they've got a
colleague to work with for $20 amonth. It's an incredible boost.

I love that. Well, we will have all of
that linked up in the show notesat thefirstclick.net/296,
Thanks, Steven. Thank you somuch for all your time and your
insights and strategies today.It

was amazing. It's been a real pleasure. Thank you
for having me.

Okay, so if you haven't, in the middle of
this episode, already gone tomake sure you have a chat GPT
account or checked out yourtools with your favorite AI
elements, I don't know whatyou're waiting for. I loved this
episode so much, and it was sucha great way to really think
about how we are using thesetools to supplement our

(45:29):
business, not to take over, butjust to help us get more done
with less time. Now, again, ifyou want to check out the show
notes and additional resourcesthat Steven talked about and
check out the freebie that hementioned so that you can get
your prompts going for yourideal customer, you can go to
thefirstclick.net/296, for now,I hope you will subscribe

(45:49):
wherever you're listening, andif you're on YouTube, leave us a
comment about your favorite keytakeaway from this episode.
Otherwise, if you are on Applepodcast, we would love a review
that really helps us get thispodcast in front of more people.
Episodes come out every singleweek, and we are so excited to
be here. And thank you fortaking the time to listen to the
digital marketing therapypodcast. I will see you in the

(46:09):
next one.

All Episodes

Ep 296 | Using ChatGPT Correctly with Steven Lewis

Episode Transcript

Popular Podcasts

Stuff You Should Know

The Joe Rogan Experience

24/7 News: The Latest

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}Ep 296 | Using ChatGPT Correctly with Steven Lewis