Generative AI for B2B content marketing is here

December 5th, 2022 will go down as the week Generative AI in content marketing passed the Gladwellian Tipping Point.

It’s all kicking off.

Heated debates on Linkedin and that other microblogging service*.

Weird experiments where people get cute little bots to say the darnedest things.

And some really cogent analysis from the best in the business.

Paul Roetzer of the Marketing AI Institute has become The Man of the Hour. After plowing a lonely furrow for years, the world is going exactly as he said it would. (And it couldn’t happen to a nicer, smarter guy).

And Scott Brinker, in his State of Martech 2022, called Generative AI (the kind that creates content) as the single most disruptive technology—and one of his 5 Big Trends for 2023. When Scott predicts, people don’t just listen, they take notes. (And buy shares).

Our own recent experiments in AI writing got some attention too. Like the blog post ‘written’ by Sudowrite. And a sad country song I prompted out of Chat-AI (Best line: “Without you, I’m just a sad country song.”).

For me, this was the week Generative AI arrived.

I feel like the uncle at the Bar Mitzvah who does a cartoon double-take when he realizes the corporate lawyer he’s chatting to is the same snot-nosed grand-nephew who (only yesterday!) choked on a Twizzler.

This shit happened fast.

One day, chat-bots are still struggling with adjective order (“Riding red little hood”?) and the next they’re cranking out thought leadership by the metric ton.

This week I’ve been playing with Jasper and GPT-3 and the penny has dropped.

With Jasper, I got a trial account and tried their many templates and recipes. Like the Blog Creation workflow—a widget that helps you turn a topic into a title into an intro into an outline into a blog post (mine was called The Power of Brand in B2B).

Was it a good blog post?

It was not.

But that’s like saying, “That poem your goldfish just recited at dinner? It sucked.”

Dude: it’s a fish. Can we just take that in for a sec?

Remarkable as it is, my Jasper-generated post on branding is not something we’d publish on the Velocity blog. Here’s the intro:

“In order to stay ahead of the curve, it’s important for B2B marketers to keep up with the latest trends in branding and voice. In this blog post, we’ll explore some of the top trends that are sure to make a big splash in the industry in 2019. From personalization to authenticity, these trends will help you take your brand to the next level. So without further ado, let’s dive in!”

Okay, it ain’t Ann Handley or Jay Acunzo.

But it’s better than the average page of Dan Brown.

It’s readable. Credible. The ideas are kind of empty and the voice bland—but I generated the entire post (‘generated’ is a way more appropriate term than ‘wrote’) in, like, five minutes. Without even suggesting a single idea about brand in B2B. (More on that below).

Generative AI may not yet be a grown-up corporate lawyer. But it’s far from the kid who almost choked to death on that Twizzler.

Cue the backlash bandwagon

You can tell just how much something is disrupting something else by the strength of the backlash.

By its volume and pitch.

Well, the anti Generative AI backlash is loud, unequivocal and dismissive.

In fact, the backlash has kicked in way before the bandwagon even got into second gear. Which tells you something.

Like most backlashes, this one seems to object mainly to an idea that few have suggested (yet): that Generative AI is ready to replace writers (and designers and photographers and illustrators).

It’s a legitimate worry. After all, it’s what this technology was invented for.

But the ‘human replacement’ debate makes it way too easy for the ‘anti’ side to invoke Never-centric arguments like, “Generative AI can never be truly creative so it can never replace a talented human.”

Putting aside the timetable issue and the certainty about the nature of true creativity (pretty big things to put aside), the Never-centric arguments stop us from seeing what is here, right now.

It makes it too easy to dismiss the disruption.

We’ve all lived through a few prematurely dismissed disruptions (like the Internet, marketing automation, SEO, social media and content marketing), so it shouldn’t be hard to recognise another one.

“Will AI replace writers?” may be the least interesting part of this disruption right now. Your money might be on Never. Mine says we’ve already passed that point—real projects for real freelancers are already going to Dieter from Demand Gen with his trial subscription to Jasper and Dall-E. But that’s not the point (yet).

The question for now is my next H2:

What is generative AI good at?

It’s too early to say for sure, but that Jasper-generated blog post about the power of brand in B2B taught me a few things:

Generative AI is good at generating sentences – Jasper spools out lots and lots of real sentences that read as if a real person wrote them. That’s not a trivial accomplishment. It’s a Holy Shit one.

Generating sentences is the superficial part of writing – A sentence is the result of a thought process kicked off with an idea. It’s the smooth, salient surface of a cognitive iceberg.

Without an idea and some thought, a sentence is an empty thing – It’s like a zombie. Looks like a person, dresses like a person, shuffles like a person, but the eyes… the eyes.

Generative AI is also getting good at structure – When I told Jasper what I wanted it to generate sentences about, it suggested this outline:

1. The rise of B2B companies and their need for branding

2. The importance of having a strong brand voice

3. Tips for creating a strong brand voice

4. Examples of successful B2B brands and their voices

5. How to keep your brand voice consistent across all channels

That’s a credible outline. A post that really delivered on this outline would be a good post.

Structure is not the same as a content – It’s the frame of the building. Once raised (Amish, barn, village), it raises our expectations. Something still has to fulfill that promise.

Here’s the thing:

If generating sentences and putting them in some kind of order is the part of writing you care about and struggle with, generative AI might already be the tool for you.

That’s not being patronizing or dismissive (maybe a little). For lots of marketers working on lots of projects, this is the task at hand.

Generative AI is as good for the dull stuff as dull humans are

Here’s a typical B2B writing brief: “You know that ebook? Crank out a landing page, ten social posts, five PPC ads and, what the heck, a blog post based on it.”

For a professional writer, that would mean reading the ebook (yeah, we kinda have to do that) and crafting copy for each of those channels. Call it a full day (two, with banjo breaks).

A marketing intern using Generative AI can now do all that in ten minutes. Then a writer might spend another ten cleaning it all up and adding a bit of voice-juice.

The intern + AI route won’t lead to better work. But it will probably get the thing over the line.

The ads and posts won’t perform quite as well as the clever stuff from the part-time novelist (ouch).

But they’ll do okay. And they’ll be done before lunch. And cost close to Nada.

Now here’s the thing that I wish-wish-wish wasn’t true but have to admit is:

The fucking bots are already as good as most B2B content writing.

Most B2B content writing is a bunch of bland, empty sentences, tangentially related to a topic, arranged in a kind of order, under a headline.

Like generative AI, it is not good.

As Truman Capote would say, It’s not writing, it’s typing.

There are are no new ideas in it and little-if-any thought (much less leadership).

Like what generative AI can do today.

But, unlike generative AI, it’s slow, it’s expensive, and it’s sometimes written by difficult, defensive people who make the marketers who commission it feel bad.

I’m going all H1 here:

The point is not whether Generative AI is good enough today. The point is that most B2B writing isn’t.

Maybe we should fix that before we waste energy joining the AI backlash-wagon.

If we don’t raise our collective game, we not only risk being replaced by bots; we should be replaced by bots.


* The social media platform I removed from my phone until the scandal-monger-attention-junkie-in-chief moves on to other things. (No, not Truth Social. The other one.)

Comments

I really admire the effort and research you put into this, Doug. I’ve been following your process (I particularly liked the sad country song).
My observation based on yours above, is that AI is learning from bad B2B content. How can anyone objectively judge the AI’s output against something that only exists in their head? AI needs to be trained, if it is learning from similar B2B posts it finds online, then by that yardstick it gets 10/10.
My suggestion is that whoever is behind the technology needs to find the best copywriters VC money can hire in order to train the AI before unleashing it.

    Thanks, Phil. That’s a great insight. I bet as AI progresses, we’ll be able to ask for different engines, each trained on different sources. For now, they’ve gone for the huge oceans of content. And that means scraping and learning from the whole bell curve of writing.

    I wonder how we could point AI only to the best stuff…

Leave a comment