Here’s the Ad Age headline that caught my eye a few months ago, “Chase Commits to AI After Machines Outperform Humans in Copywriting Trials”.
I follow the subject of AI closely enough to know that someday, this headline could be dismissed: “Obviously“.
But I also know that such a day is probably really far off. Like The Really Big One that will hit the Pacific Northwest, a wise, self-aware AI entity, with human level intelligence, might manifest in 15 years – or not for another 150.
This meant there had to be a hole in this headline, because copywriting is just writing.
And writing is not chess and it’s not a self-driving car. It’s not a science, at all – no matter how scientific your measurement of its impacts.
In fact, performance marketing itself really only exists in theory. You can never, ever fully measure the impact of marketing. All we know is that it works.
That’s partly because a machine (and people struggle with this too) cannot resolve the essential tension that copywriting usually (not always) deals with: long-term vs short-term interests.
This tension calls for protecting an organization’s long-term, strategic interests using “brand messaging” while maximizing impact in the present using “conversion copywriting”.
By the way, can you guess which of these two Chase digital ads got more clicks?
- Access cash from the equity in your home.
- It’s true—You can unlock cash from the equity in your home
A human copywriter, we are told, wrote #1, while #2 was written by a machine learning algorithm provided by a machine learning software company called Persado, with whom Chase has contracted.
Of course it was #2 – a rigged test? Who knows.
But despite my skepticism, I was interested in this story because I know that we’re all of a sudden deep into the era where AI (or Machine Learning, actually) is our extremely useful assistant, in so many walks of life, banking, driving, healthcare, and on and on.
It’s already in use in Gmail, LinkedIn, and – quite a bit more interestingly – on Crystal app. And two of those tools are used at enormous volume around the world in marketing and in business.
Machine Learning for Ideation
Reading deeper into the article, the implication of the headline (that machines write better copy than human copywriters at Chase) is contradicted. As it turns out, the machine language software in questions hasn’t replaced anyone’s jobs.
But it has served as a useful ideation tool.
Chase plans to use Persado for the ideation stage of creating marketing copy on display ads, Facebook ads and in direct mail, according to Yuval Efrati, chief customer officer at seven-year-old Persado. He says that the AI company works alongside Chase’s marketing team and its agencies.
When asked if the new relationship will impact staffing at Chase in terms of downsizing or changes, a spokesman for the bank said, “Our relationship with Persado hasn’t had an impact on our structure.”
With that context, the article makes a lot more sense. Machine learning software as an ideation tool – a tool for seeding the human mind with the thousands of little ideas to make marketing work over an extended period of time.
This is similar to what SEO/AdWords keyword concatenation tools do, such as https://kombinator.org. Like most “AI Software” firms, Persado claims to do a lot more than concatenate; it uses NLP to evaluate and predict ideal word usage.
I’m sure there’s something to that; it’s something any business owner should be aware of.
What’s Next?
So how do you harness machine learning for your marketing ideation if don’t have Chase Bank’s R&D budget or marketing technology budget? Persado is not open source.
Not a rhetorical question – I really have no idea. I do think you have an enormous opportunity to leverage analog ideation approaches.
But there’s no point in ruling out software wherever it can help.
Maybe we just wait until machine learning starts showing up within accessible marketing technology like MailChimp and WordPress and Google Docs; it’s already part of Google Ads, which writes ad copy for you, if you want, in much the same way that Persado describes its products behavior: through text analysis and acquired learning.
So does another Google product – Gmail, though I’m not sure it’s solving for inspiring action in your favor, as Google Ads supposedly is. Maybe Gmail thinks what you want is for people to be less – not more! – likely to respond to your emails.
Meanwhile, most of the ideation-for-ideation’s-sake software (IdeaScale, Spigit, Qmarkets) seems to offer a digitalized version of the office-hallway “idea box” that employees drop anonymous notes into.
That approach is probably not leveraging machine learning in a significant way. And really not doing much more than facilitating a large, low-energy brainstorming effect.
I haven’t done an exhaustive analysis here, though, so correct me if I am wrong, but “Idea Management” doesn’t really seem like ideation software at all.
It certainly doesn’t square with my concept of ideation for marketing and business development: condition your individual mind to produce as many new, valuable ideas as it can within given project constraints.
That’s pretty much the straightest path to creating a marketing strategy that will yield more revenue than whatever somebody else is doing.
But let me know if you have something better up your sleeve..?
Best,
Rowan