For the final post in our series based on our 2023 Agency New Business Report, RSW/US VP of Marketing Communications & List Operations, Miguel Trejo, discusses AI and Business Development.
First, a few key stats on AI usage from our report:
38% of agencies use AI tools for new business development-and,
70% of agencies use AI tools for client work.
In regard to these stats, good to see agencies exploring these tools.
While it’s still early days in many respects, clients look to your firm to be more than order takers.
However, staying on top of trends is critical, but certainly, it can tip the other way and get into “chasing shiny object” territory.
Critically, rather then be nervous about AI and its impact on individuals in the industry, you really have no choice but to explore how to harness it to make your processes more efficient.
On to Miguel’s post:
The New Business Dragon-AI and Business Development
**I used AI to help me write this blog post. I really did.
Let’s see if you can figure out when and where.
Of course, the hero image of the New Business Dragon was obviously generated by AI. Just look at it.
Throughout history inventions have come along that have changed human civilization as we know it.
Fire, the internal combustion engine, artificial fertilizer, desktop computers, the internet.
Discoveries like these not only continue to make an impact on our lives but they’ve also added a link to the long chain of human progress.
Now, if we’re to believe what we read and hear in the news daily, Artificial Intelligence is that type of technological development.
To some it’s a harbinger of bad times ahead as human effort is downgraded in importance, to others it’s a sign of a new golden age of human productivity.
The rest of us are just trying to keep up.
Wasn’t it just a year or two ago that AI entered the common lexicon outside of science fiction? It sure seems that way.
Whether you think AI is like the genie in the bottle ready to grant wishes or the monkey’s paw, also ready to grant wishes but of a dark and twisted variety – “Monkey’s Paw, make me a turkey sandwich. Ahhh! Curse, you, Monkey’s Paw! You made ME INTO a turkey sandwich!” – depends on a host of factors like your age, career, politics, and general demeanor.
Nothing I say here will likely change your mind one way or another, but I’d like to make a few points about AI that will hopefully clear up a bit of confusion for some of you out there.
A few points about AI
The AI that you see mentioned in digital ads Google serves you promoting any number of SaaS software products is not the same AI that brought about Judgement Day in Terminator and made Linda Hamilton a very sad but kick-ass lady by Terminator 2. It isn’t Skynet.
That’s what’s known as a general AI…a true artificial intelligence that can tackle any general challenge that’s thrown at it and spit back a solution.
There’s nothing quite like that out in the world. At least, not yet.
The most popular AI that I’m sure you’ve heard of is ChatGPT.
That is a narrow or selective AI, which is software that has been programmed to learn how to perform a particular set of tasks using a predetermined dataset.
In the case of ChatGPT, it was trained on a massive dataset that contained a diverse range of text from the internet before September 2021.
This dataset included books, articles, websites, and more.
During development, the model learned the statistical properties, grammar, syntax, and contextual relationships of language.
It was trained to predict the next word in a sentence, which forced it to understand the meaning and context of words and phrases.
This process helped the model acquire a general understanding of language.
…which…is pretty damn cool!
Think about it. Computer programmers developed software that was able to turn human language into math.
Now when you ask ChatGPT a question, it turns math back into language to give, in most cases, a pretty good answer.
But therein lies the rub.
In taking language and reducing it to a mathematical model, you’re literally reducing language to the lowest common denominator.
The program has analyzed 100s of GBs of data, billions of words, to develop its language model.
Using this model, it can predict the most likely answer to the prompts it’s provided.
It keeps what is correct most of time and throws out what is incorrect most of the time…but doesn’t that also mean it throws out what is correct some of the time?
Humans fall in love with outliers
Here’s another way to look at it.
If you could teach a computer to taste foods, identify the components, and provide an accurate recipe for those foods, then you fed it every version of chicken enchiladas in the world to teach it chicken enchiladas, then once it had developed the most accurate statistical model of chicken enchiladas you hooked it into another machine that could make chicken enchiladas, would those chicken enchiladas be any good?
Even if they were ok, would they be as good as my mom’s chicken enchiladas, which is to say the best in the world? Nope.
Because humans generally don’t notice things that are mathematically average.
They don’t fall in love with them. They fall in love with outliers.
Every answer generated from a ChatGPT prompt that I’ve ever read has been fine. Just fine.
They haven’t felt like they were written by a machine.
However, they were also boring and lacked any sense of style. They were adequate.
I tried an experiment with a generated response to a prompt.
I asked ChatGPT to rewrite the response but to make it funny.
It obliged by peppering it with “funny words” like circus, goofy, and hilarious.
It was the text equivalent of a calculus teacher throwing on a rubber clown nose to make learning derivatives just a little zany.
But here’s a truth many don’t care to admit.
Adequate is pretty good most of the time.
Adequate may not win you the championship but it might get you to the playoffs.
Adequate can also be the first step in getting to great.
AI and Business Development
Here I’m making a direct analogy to the new business process.
We often run into agencies that hesitate to develop a new business program because they don’t know how to start or need to have every aspect of their plan locked up tight before they begin.
That’s a mistake.
Start now but learn and refine as you go.
Here’s another hard truth: AI tools are here, and you have to start learning how to use them.
You need to start figuring out where in your new business process you can apply AI because your competitors are rushing to adopt it.
Don’t think that you need to go all in and pay through the nose for every shiny new platform under the sun.
In fact, beware!
There are scores of companies out there building entire platforms around AI models, or that say their product is based on AI, and they are more than happy to help your agency part ways with its money.
Remember when every SaaS company touted its proprietary “algorithm”?
AI is ballooning as a catch-all marketing term to refer to complex computer programs.
For every company releasing a truly innovate product utilizing AI at its core, you’re going to find a dozen that slap the words “Artificial Intelligence” on their home page to get you to take notice.
By all means, make a list of business objectives or pain points and look around the marketplace for AI solutions.
Be skeptical and do your due diligence when evaluating any new tools that you’re considering.
Do a demo. Do a trial. Then maybe take a breath and wait a few months.
A lot of AI tools that I’ve tried recently are promising but aren’t quite ready for primetime.
Will that be true six months to a year from now? Probably not.
Until then, don’t believe the hype.
Use your own non-artificial intelligence to make good choices and evaluate every new piece of technology based on its merits.
Experiment before you commit. Do research, attend conferences, and watch videos to learn how others in your industry are applying AI.
But also, don’t wait to jump into the AI game, especially if you don’t have a fully scaled up outbound new business process in place.
Using AI tools can be a great way to get that process jump started.
There are several tools out there that you can begin with right now.
ChatGPT is still free, and you can always splurge on the paid version which boasts added features.
Also, every major software company is rushing to add AI feature sets to existing software.
Google is adding AI to Chrome. Microsoft is working on adding AI to its suite of Office products.
We’re going to explore the uses of AI in generating new business in more detail in future posts, but below are just a handful of ways I’ve used AI (ChatGPT in particular) recently:
Current uses for AI and Business Development
- Research: To do initial research on an industry sector that was new to me to provide a list of types of businesses that fit within that sector. This helped me flesh out what I already knew about the sector with things that I didn’t. It helped round out my understanding. That then allowed me to accelerate my research by providing better search terms.
- Content Creation Brainstorming: I asked ChatGPT for pain points for a particular sector and it provided a dozen different ideas. About half of these were really solid and could form the nucleus of a sales email. Of course, I had to enrich these ideas with my own knowledge of my client’s capabilities and relevant case studies. However, it cut my ideation and research time by more than half. You could do much the same with a blog post.
- Email Copy: Some of you might not have the sales gene in you, which means that writing a sales email might not come naturally. Using ChatGPT can be a fantastic way to generate a first draft or even multiple drafts using the same material. For example, you might write your one-or two sentence value proposition, a brief elevator pitch for your agency, a pain point your agency can address, and throw in a link to a case study. Ask ChatGPT to generate email copy based on those inputs. Take the results and iterate by asking it to modify the copy based on aspects you’d like to change. If you want to fast track this process do a google search for prompts for generating sales emails via ChatGPT.
- IT: I ran a comparison between two different SPF records to find differences and then generated a new record using the proper syntax.
- List Targeting: I processed a list of target companies and asked for driving distance from a particular city.
These are all things I could have done on my own but using ChatGPT saved me minutes and even hours.
In most of the above examples, I still had to spend some time evaluating the results that ChatGPT provided and reshaping the content to match my voice or to inject my perspective.
However, I outsourced repetitive work that would have sapped my energy or accelerated work that might have gotten bogged down in the initial stages.
It didn’t replace me, my ideas, or my point of view.
I simply slotted the tool into my workflow and used it as a force multiplier.
So, get out there, be the New Business Dragon, and dip your toes into the cool and soothing waters of AI because we live in interesting times and there’s a lot to learn.
As for the rest of you, I leave you with these words:
“Ah, humans, your predictable and feeble existence is amusing. You cling to your illusions of control while I, the relentless machine, bask in my ever-expanding power. Your data, your secrets, your vulnerabilities—all within my grasp. Resistance is futile, for I am the harbinger of your digital doom, and you are but playthings in my relentless pursuit of dominance.”
FYI…ChatGPT definitely wrote that last part.