john higgins: leveraging chatgpt for advisory | gear up for growth

click to subscribe anywhere: apple, google/youtube, spotify, iheart, deezer, amazon music and audible, player fm, audacy, gaana (india), boomplay (africa), or rss. 

gear up for growth
with jean caragher
for 卡塔尔世界杯常规比赛时间

“ai and chatgpt-type tools can become your advisory services assistant,” says john higgins, founder and ceo of higgins advisory. “it can help cpas communicate better as an advisor and help identify the things cpas should focus on for the type of client they’re advising.” 

higgins was a guest on gear up for growth, hosted by jean caragher, president of capstone marketing, and powered by 卡塔尔世界杯常规比赛时间. higgins shares specific examples of how chatgpt can help cpas become more efficient by providing quick research assistance, summarizing complex information for client presentations, and creating tools like excel templates for common tasks. how does efficiency impact a firm’s pricing model? 

jean:  we’re going to chat about chatgpt. and i know you’re the guy who’s going to help answer some questions for our cpas and accountants listening and watching. so let’s start off with, why should cpas use chatgpt? what benefits are they going to experience?

john: well, there’s lots of benefits at a macro level, whether you’re a sole practitioner running a small firm or a larger firm. we have a couple of big challenges we’re always trying to focus on is, how do we increase our productivity to get more done with less time, and how do we maximize our earnings potential? and so tools like chatgpt…and of course, chatgpt is a specific tool in a broader category that is referred to as generative ai. and so those are the two ways that this tool can help you. there’s lots of opportunities in any practice to apply generative ai, chatgpt, to just drive out inefficiencies, to make the process more efficient. and we can talk about some examples along the way here.

on the other hand, you also have many opportunities to use it to expand your revenue. and so i know, jean, if you’re like me, you go to a lot of accounting conferences and such, and i can’t tell you for how many years i’ve been hearing the message that you have to become more advisory, you have to become more advisory in your services. it’s like, you know, it’s really been drilled into our heads, but i always come back with the question…

jean: john, how many decades have we talked about the need for cpas to be more advisory? and we know that some are doing an excellent job at that, but there are many others that feel like if they answer the phone and answer the question correctly, that they’ve been advisory. and we both know that that’s not really what we intend by advisory.

john: right, right. and you know, so we hit this message that we need to become more advisory, but then the question becomes, okay, well, how do i do that? and for many of the cpas, they didn’t really want to be advisors initially. that’s not necessarily why they came into the profession. but i mean, i agree with the reality. we do need to become more advisory. and so the opportunity to use ai, chatgpt-type tools to help you become an advisor, it can become your advisory services assistant. i do a full presentation on how to integrate chatgpt into your client advisory services because it can help you communicate better as an advisor, it can help you identify the things you should focus on for the type of client you’re advising, and it’s just lots of different incremental ways, which i think most cpas are comfortable with moving into advisory incrementally. with generative ai, you can do that on an incremental basis.

a simple example that i like to use is let’s say you are preparing financial statements for a nonprofit organization, and you, as the cpa, need to give a presentation to the board. the board may not have any financial expertise, right? they may be other volunteers that just want to do good. maybe it’s for, say, a food bank. but you, as a cpa, you prepared the financial statements and you need to kind of present them to the board. chatgpt can help you look at those financial statements and identify what are the key things that would be of interest to the board members that i should focus on. you can literally upload the financial statement, and there’s things you can get into about how to do that securely and confidentially, but basically upload the financial statement and say, “hey, chatgpt, give me an outline of a 20-minute or 30-minute presentation for this board of directors that have no financial expertise or little of this food bank in tampa, florida.” and it’ll come up with a very nice outline for you to follow. so it’ll help augment you and assist you as your advisor, help you to provide those advisory services, as one example.

jean: right. now, because i’ve used chatgpt in different areas, of course, like everybody else, and it is very helpful, and it does make you more efficient. i think something we need to point out is the need to really check on what is being created for you through chatgpt, right, that we shouldn’t necessarily take everything that is perfect. am i asking that correctly? because, isn’t there a point that you need to verify the information that it creates for you and not to just take what it spits out?

john: right. and that’s the way that we should, as cpas, always be looking at this technology. it augments us. it does not replace us. it does not replace our judgment and our overall broader scope of knowledge. and that’s why microsoft calls their version of chatgpt microsoft copilot. it’s not autopilot, it’s copilot. now, as i’ve gotten more into this, i’ve kind of immersed myself into this technology the last couple of years because it is such a fast-moving and such a powerful technology, but one of the biggest concerns that i’ve developed with this technology is that i worry we’d get intellectually lazy, that, you know, it can do 80% of the work, 90% of the work, but we still have to do our 10% or our 20% of it. but i can see a lot of situations where people say, “yeah, that’s good enough. let it go. let’s just go with it.”

and that’s the concern i have, is that we get intellectually lazy. because it really should be a ladder. it should be a step on a ladder for us to move up higher into the value chain, as they say, and not just kind of let it do everything for us and make sure we’re using our professional skills and judgment to work with the information that it generates.

jean: right. so we’re going to get into a little bit more specifically about chatgpt and maybe some different examples. but in this initial conversation, what impact does chatgpt have on learning, i think you just started that conversation, or pricing, or risk, or other operations within a firm?

john: so it’s like electricity. as andrew ng from google would say, it has application everywhere. so it’s going to impact every aspect of a firm inside the firm and then externally with the client services. on the learning side, it’s the best learning tool i’ve ever come across in my career. so i give you a good example. i develop a presentation on cryptocurrency and bitcoin. and so i’ve learned a lot about that as i’ve developed that, you know, base of knowledge from the technology side, but i wasn’t really a financial guy in terms of the financial markets and so on.

so earlier this year, the sec approved etfs, exchange-traded funds for bitcoin. and i said, “okay, i have an understanding of the concept of an etf.” but i wanted to know, so i asked bitcoin, i said, “explain to me at the ninth-grade level, what is an etf in the context of bitcoin and cryptocurrency?” and boom, within 60 seconds, i have a nice 2-page, you know, dissertation at the 9th-grade level so i can understand at the 9th-grade level of what an etf is and how it impacts bitcoin. this is one example. it is my go-to research tool today.

prior to that, it used to be google search, and google search, yes, by comparison, not to knock google, because google has gemini, which can do the same things as chatgpt, but in terms of a google search, well, first of all, we all know the first 20 hits are all sponsored ads, right? a lot of it is, like, you know, content that is based on self-interest of whoever is providing that content. i think that’s kind of fair to say. whereas i see the generated output from chatgpt as being more pure. so it’s a good learning tool. it’s a really good learning tool.

when we get into pricing, so interestingly, about a month ago, i was giving a presentation on this topic at the…auburn university had an accounting and auditing summit, and so i was, like, third or fourth speaker in the day. so i went for the first speaker who was a partner with pwc. and, jean, i was blown away. now, i’ve immersed myself in this, and i’ve been talking about it and educating cpas about this technology. but when i saw what pwc is doing in the way that they’re using this technology, he talks specifically from the audit function, it was mind-blowing for me. and so then i asked the question of him, i said, okay, you know, i forget his name, tom, i’m going to say, “as a firm, how is this impacting your pricing, your fees?” he said, “we talk about that every day.” because the issue is, and i’ve seen this with cpas over the years, you now have a tool that can increase your productivity substantially. i mean, substantially. twenty percent is, like, the baseline productivity. you can get a lot more productivity if you really, really dig into it.

but then, how do we, as cpas, transform that into increasing our revenue? because if you’ve got the good old hours times rate and now you get the work done in half the time, are you going to double your rate? well, if you do that, you know what, people are going to get upset because it’s like, “you’re not worth $1,000 an hour. get out of here,” right? so we have to…definitely, firms have to be looking at this as, now, another big push into why we have to go to value pricing or the subscription model that a lot of folks are talking about these days, because clients want value, and now, with these tools, you can give them more value, just don’t give away your time. that’s the thing i do worry about that firms, larger and smaller, are going to have that problem.

jean: exactly. because chatgpt is making you more efficient, so it’s not taking you as long to provide the services for your client. and if you’re doing hours times rate, you’re losing money. so you’re still providing the same value, but you’re losing money. and we know that our friend, michelle river, and her advanced pricing methods is going to say, cpas need to charge for their worth, and clients really don’t care how long it takes them to do something. what is it worth to the clients? and it’s not whether it took two hours or five hours or somewhere in between.

john: yeah. and my experience is i’ve been more in the client advisory of technology services for the last many years, and i can’t remember the last time i generated a proposal or an invoice that was based on hours and rate. i’ve always done value pricing. and the clients love it because then they have more certainty. you explain to them what they’re going to get. what are your deliverables for whatever the project is? and here’s what the costs are. and here’s what the price is. and they like that because, like i say, it’s certainty. if they don’t like the price, they say, “i don’t like the price.” okay, well, then we’ll move on. that’s fine. let’s figure that out now before we do the engagement as opposed to after we do the engagement, and then we have a bunch of write-offs.

jean: yes, exactly. and then we have to negotiate, you know, exactly how much the client is going to pay. okay. so we could go on and on about the pricing issue, right? let’s move back into how cpas are using chatgpt. and you gave me an excellent slide, which i am now going to share. and let’s get the full screen. yeah. okay. so talk to us about this a bit, john. so, are you guys part of the top of the research and the learning piece that you touched on a little bit earlier?

john: yeah, we’ll just kind of walk through these. so i use this in my, what i call, chatgpt boot camp for cpas presentation. and so what i really try to focus on is, one is demystifying the technology. first of all, demystify it, help them understand the fundamentals. but then, okay, how do i use this? how can i start to use this in my practice? so i’ve kind of evolved as i’ve learned and used the tool quite a bit over the last couple of years to link these four buckets. you know, just look at them as four buckets of different types of applications. so i already kind of talked about the research and learning. i gave you the example with bitcoin and etfs.

another good example is with excel, right? now, i’ve done a lot of excel training over my career, and it’s still one of the most popular topics to provide education on because there’s so much to it, and people are always trying to…cpas are always trying to find better ways to use it. well, now, let’s say you bring out a new staff person, and who knows what kind of education that person has had relative to how to use the nuts and bolts of excel in college. it doesn’t matter because you haven’t given them two monitors and you say, “okay, here’s what i want you to do. i want you to build this carryover schedule, this tax carryover schedule for this client.”

and so then the staff accountant goes to his or her desk, and on the other screen, they have chatgpt open. and they say, “what’s operating loss carryover schedule? what is it?” okay, i’m just, you know, simplifying it because maybe they don’t really know. so we can educate them on what the schedule is and why it’s important. then you can say, “help me or create for me an excel spreadsheet that will track this carryover schedule.” and the more information the staff accountant can give it, it’s going to come back with very specific details, keystroke by keystroke, of how to build that spreadsheet.

so you talk about on-demand learning, this, like, right-now learning. “i need to know right now. i have it on my other monitor. i just ask chatgpt how to do this in excel, and immediately, it comes back.” and like i say, it’ll go as detailed as mouse-click and keystroke level of detail. so that’s what we need. and of these four buckets, jean, this is the one that is the least risky, right? because there’s risks associated as we go through these, and we’ll talk about the level of risk, because when i give my live in-person presentations, that’s the number one concern, is what about the risk to my sensitive confidential client data? well, here, you’re not putting any client data in there. you don’t need to to learn and to do research. you can anonymize whatever you’re asking about.

then the second bucket is communications and content generation. this is one of the more…probably the most frequently used application across businesses, right? marketing communications people use this a lot. i use it a lot. so i just started higgins advisory. it’s a little over a year old now. so i said, “well, i better get a website,” right? so i went through and i went into chatgpt and i explained, “here’s the nature of my business. generate for me the content that i should have on the website and recommend the images that i should use and where on the website.” within literally two minutes, it gave me a nice document with the suggested images, and then i found a freelance web developer and i reached out to him, i said, “can you create a website for me?” and she charged me, i don’t know. she undercharged me, which is good for me, but bad for her. it’s like 300 bucks, okay? i don’t know, maybe 500 bucks. you probably don’t want to hear that.

jean: you were quite undercharged.

john: yeah. because she just did it as a gig. anyways, so i just sent her the content that i generated from chatgpt. i said, “here you go, build the website off of this.” so i might have spent maybe an hour, two hours in total on the website. chatgpt gave me the content, and she put the content into a website format. so other types of communications. i gave the example of if you have to give a presentation to the board at the nonprofit organization. whenever you have to give a presentation, chatgpt can be a great resource to help you develop the outline for that presentation. i’ve used it many times in my courses because, a lot of times, i have 75 minutes or a 100-minute presentation. i say, “give me an outline to present the fundamentals of chatgpt to a group of cfos. i have 100 minutes and give me time allotments for each segment of the presentation.” and again, within 60 seconds, i’ve got a nice outline of what it recommends as the sequence of topics and kind of bullet points maybe of what i should cover in each topic. so those are all examples of content generation. hosting an ad for a staff accountant position is another good example.

jean: right. let me interrupt there for a second because i would be remiss as a marketing consultant to add, not to forget about the expertise of specialists in marketing and design and programming. and the example you gave about your website is fantastic. but there are added benefits to really building a brand through the website, and that takes more than chatgpt.

john: absolutely. yes, totally. i just, you know…

jean: it’s an assistant, just like you were talking about in doing research and answering questions and all that. it’s not all chatgpt. it’s another resource that we could use and to use our human brains and creativity added to that.

john: yes. i can give you a good example. that’s a great point that you bring out, and i can give you a good analogy to that in the cpa world. so i sold my previous business four or five years ago. and so there’s some tax complexity with that. so i had an old acquaintance when i worked at bdo. he’s still in practice, and he did, you know, what i would call higher-level tax advising, tax planning for me. and then i also had a sole practitioner who actually prepared my tax returns, right? so the sole practitioner prepared the tax returns. it was very, you know, inexpensive. again, underbilled me. i would, you know, tell him that, he’s a friend, actually, “you know, i would have thought you’d bill me more, but you didn’t.” so that’s next time, maybe, if there’s a next time. whereas the higher-level tax advisor was bringing much more value to me because he’s telling me how to structure, you know, the terms of the sale, things like that. and so that goes back to your example from a marketing perspective.

jean: exactly.

john: sometimes you just need a simple web page, but there’s more to it when you really want to have, like, a digital marketing campaign set up.

jean: exactly. exactly. so tell us a bit about data analysis.

john: okay. so the first two are kind of the low-risk or no-risk, you know, research, learning, communication, content generation. now, you get into data analysis, and here, you can get into some risks because…well, let’s talk about the risk next, but let’s talk about the applications. so where a lot of auditors are using this now, for example, is they have to go out to a client and they have to look at, you know, very lengthy detailed contracts, whatever they may be, sale of a building, sale of a business, purchase of an asset, lease of an asset. well, they used to have to spend a lot of time reading those documents word for word. now, they’re uploading those documents and asking chatgpt to analyze the document, telling it specifically what they’re looking for, and it will find, you know, what they’re looking for or tell them it can’t find what they’re looking for.

another example of data analysis is you can upload an excel spreadsheet, and rather than having to create a pivot table in excel, which many cpas do, you can just ask the question right in chatgpt. you upload the spreadsheet of, say, transaction data and say, “give me a summary of or give me the average gross profit by product category or give me the average sales by territory,” whatever the data is. and so it can generate. it can look at that detailed data and summarize it for you without having to build complex formulas and such, say, in excel or building a pivot table in excel.

so one example i like to show is i pulled down from the web, i pulled down nvidia’s 2023 annual report. it’s like a 65-page document, and inside of all the glossy images and such, they have some data. they have their financial data. so i went in there and i just show that, i say, “chatgpt, take this pdf of the annual report and calculate these three financial ratios for me.” and it goes and does it. it pulls out the data, finds the data in the document, pulls it out, does the calculation, and explains the calculation to me. that’s the beauty of it.

the downside is the risk is i have to go back and verify those numbers, because if i’m using chatgpt or any generative ai tool to generate, in the output, if it’s generating any facts or figures, then i feel that i need to verify those numbers or those facts through an alternative method relative to the extent to which i’m going to depend on it or i’m going to have a client depend on it. you follow what i’m saying? sometimes it may be that it’s not that big of a deal. you’re just kind of looking for some information, and it’s not critical if it’s off a little bit. but if it’s specific data that i’m going to hand off to a client, then i need to make sure that i verify that information.

jean: right. i mean, if they’re going to be making important strategic decisions based upon this data you’re giving them, you really need to be confident about it.

john: exactly. and cpas have been using excel for years to do projections and things like that. i can assure you there’s a lot of errors in many of those projections. it’s just the nature. a professor at the university of hawaii did a study with one of his graduate classes. the amount of errors they found in spreadsheets was pretty phenomenal. so we’ve always had errors. we’ve always had that risk. but in these early stages of generative ai, that risk of the data being wrong or inaccurate is high. it’s not real high. it’s high.

but i would also say, as soon as i’m saying that, that over the next couple of releases of chatgpt 5, well, chatgpt will be 5 and then 6 or whatever they decided to call it, if they’re getting smarter, as sam altman, the ceo of openai and co-founder would say, it’s just going to get smarter and smarter and smarter. and so the risk of bad data in the facts and figures will become much, much lower. it’ll become lower. it may already be lower than, say, a staff accountant. not putting down staff accountants, but we all make mistakes.

jean: right.

john: in one of the stats for the newest release of chatgpt, for example, what’s it called? the ame, i think it is. it’s like a high school high-level math exam that’s given to high school students to really find out who the top mathematicians or mathematic brains there are in high school. chatgpt o1 or whatever it is, the newest version, scored 95 percentile.

jean: okay. so then this last piece is the workflow automation and personal productivity.

john: yeah. so this is where you get into nitty-gritty details. one of the best examples is…i keep using excel because that’s such a commonly used tool by cpas, right? but i show examples of where i can upload an excel spreadsheet, and if you’ve created a macro, if you understand the concept of a macro where you’re basically automating a sequence of keystrokes and mouse clicks, well, you can upload an excel spreadsheet and tell chatgpt to create that macro for you.

so as an example, i upload a 30-year loan amortization schedule. all right. so let’s say i create that for a client. well, if i’m going to print that to a pdf for the client, they don’t want to see all 30 years because we’re only 2 years into the loan. so in that example, i said, “chatgpt, please create a,” and i always say please, because we never know when these robots are going to take over, right?

jean: do you get a better answer when you say please?

john: so i say, “chatgpt, create a macro so that the user can put in a date for which they want to print the amortization schedule through.” and within two minutes, it gives me the macro. i just paste it in excel, and i get my macro. that’s workflow automation. that’s one simple example.

jean: right. this is all just amazing to me and perhaps a bit scary at the same time of what this technology is able to provide to us. so in your experience, what is the best way for cpas to get started with chatgpt?

john: very simply, use it. just use it. you can go out to the openai site, and you can use it for free and just use it for personal stuff. just to give you an example, you can take a picture of ingredients in your pantry and say, “what can i make with this?” or, “what can i make that’s…what kind of cookie could i make from this?” this is one example. one of my favorite examples is, if you’re a parent with young children, or even better yet, if you’re a grandparent with young children, you want to be the hero, before you go over to the house to see the kids, you go into chatgpt and say, “chatgpt, create a bedtime story for my grandchildren, ryan and nicole. ryan’s superhero is superman. nicole’s superhero is superwoman. put them into the story and make sure that it has a happy ending. and make it 500 words or less because i don’t want to stay up all night reading to them.” it gives you a beautiful story, and you can have a different story every night.

jean: so the limit of what this technology can do is limitless, actually. i mean, i would never have thought of an example like that of a bedtime story, but literally, anything that might pop in your head, you can put into this technology and have it spit something out at you.

john: that’s a great way to look at it, jean, for sure. i mean, think of cpas. we’re knowledge workers, right? we’re working with information that we’re generating. generative ai generates information.

jean: right. because i’m thinking, from a marketing and sales standpoint, and i know there are services that provide this, which again is another reason why this is all a little scary, but if a cpa is going on a prospect meeting to a company in a particular industry, i’m sure that you could ask chatgpt, “please, provide me some important questions that i need to ask the owner of a construction company in tennessee.”

john: exactly. and that’s how you become more…that goes back to that thing we talked about becoming more advisory.

jean: right.

john: it’s your assistant that’s going to help prepare you for the meeting.

jean: exactly. right. so now, we’ve talked about risk and having to verify information that we get through chatgpt. is there anything else that cpas need to be careful about when using this technology?

john: yeah. so we talked about chatgpt. now, i also want to take a minute to talk about microsoft copilot, because when i give presentations, most of the people in the room, the vast majority is using microsoft 365. so microsoft 365 copilot is chatgpt kind of rebranded, because microsoft has a big ownership interest in openai. so microsoft, for example, is fully committed to using your data, your client data, whatever data you have in microsoft 365, to generate an output, and they’re fully committed to make sure that your data never leaves the four virtual walls of your organization. in the case of chatgpt, you have some capabilities to make sure your data is not being passed out to the large language model as well.

but the best step to take is just, if you’re going to upload content as part of your prompt, as we call it, then just anonymize it, you know. like, if it’s a tax return, take off all the taxpayer information, the address, account numbers, you know, which could be a little bit of work. that’s why more advisors do it in microsoft copilot because i’m not going to be as worried about it or concerned in that regard. but if you want to be extra safe, which i know cpas want to be extra safe, just make sure you anonymize any content that you’re uploading.

jean: right, right. that’s excellent information, because we know, whenever i read or hear about identity theft and information being stolen, i always think to myself, “why don’t those people use their ingenuity for the good of the world instead of trying to scam somebody?” i just don’t understand it. because, obviously, these people are smart. i hate to admit it, but they’ve got to be smart to figure out how to steal all this information. they could be doing that for so much good. now, i know you just mentioned copilot. is there any other exciting new technology that you see coming ahead for us?

john: well, i mean, so this whole path of generative ai is just really beginning. like, chatgpt was released just about two years ago as we speak, and it’s moving. you know, jean, i’ve been involved in cpa technologies for a long time, and i’ve never seen anything moving as fast as this is moving. and so one thing i would say for cpas listening to us here today is don’t sit back and wait. you’ve got to learn how to ride the tricycle, then you go to the bicycle with the training wheels, then you let off the training wheels. otherwise, you’re going to find yourself just getting on the bike with no training wheels and you’re going to fall down. it’s going to be hard.

but back to your main question, the exciting technology, which they’re really starting to talk a lot about, is this concept of agents, ai agents. so, what’s an ai agent? well, think about generative ai, like chatgpt, does our…i won’t say it does our thinking for us, but it creates information for us, okay? that’s what it does. it creates information for us. and so it kind of thinks for us, but it definitely creates the information that we would otherwise have to create. agents act on that information.

so an example, let me just give you an example. i could go into chatgpt and i could say, “create an agenda for a five-day trip to the southwest, okay?” i could get more specific and i could say, “i’m flying from, say, detroit and create the agenda, create a budget, blah, blah, blah.” and it’ll come back with an agenda and all this. and i could say, “provide recommendations for hotels, airlines, blah, blah, blah.” an agent, i could then, with agent technology, i could go in and say, “okay, now, book everything for me.” it will go to delta and book the flight. it’ll go to hertz and book the car rental. it’ll go to the hotels and book the hotels for us. that’s where this is happening.

now, it’s very early on, so i don’t want to be too hyped about it, but you always have to be careful with emerging technologies about being too hyped up about it. but in this case, i think it’s legitimate because, if you just step back and look at what’s happening in the marketplace, you look at nvidia, right, everybody knows about nvidia stock just going through the roof. well, nvidia is the baseline. nvidia is building the core raw ingredient for ai systems. it’s going to be two, three years before that really gets out there at that level. when all those chips get out there, those data centers get built, there’s going to be so much compute power that it’s just going to be, i think, mind-blowing, frankly, to the things we’re going to be able to do with this technology.

jean: boy, this is just amazing.