437 Digital Marketers Went Head-to-Head with a Conversion-Predicting Machine — Who Reigns Supreme?
Being a digital marketer is an exciting gig. Ad platforms, best practices and tools change at warp speed, meaning you’re always learning and you’re never bored.
But it’s a tough gig, too. As competition stiffens and we spread our time and resources across more and more platforms, it’s harder to get our message seen, and we yearn for the results we once (perhaps) took for granted.
In 2004, the internet became the highest grossing channel in advertising expenditure, and it’s been on the rise ever since.
That up-and-to-the-right blue line means that acquisition has become more competitive than ever.
Couple this with a rising average cost per click and cost per conversion, and you’ve got a hefty task at hand: Cut through the search noise, create a compelling digital experience and interpret and apply tens of thousands of data points all while coming in under budget.
It’s no wonder that marketers are looking at advances in automation, machine learning and artificial intelligence and asking, “What’s in it for me?”
AI technology is already infiltrating all aspects of society, from the fleet of self-driving cars Google is testing, to the anti-snore wearable Sleep.ai you go to bed with to the Amazon Alexa smart speaker in your living room.
What these machines have in common is that they’re designed to make life better for humans so we can spend less time commuting, tossing and turning or shuffling through our record collections and more time on the things that matter.
And while some see the “rise of the machines” as a threat, Mitch Joel, President of global digital agency Mirum, sees it as an opportunity:
I see this and think, ‘Wow, this is going to be a very interesting and dynamic time for people who really want to build better relationships with customers.’
To fear or not to fear the machine, that is the question
Pop culture is rife with dystopian visions of a machine-dominated future. But we’re already seeing AI technology being used to complement human ingenuity and serve the greater good.
IBM Watson Health, for example, analyzes vast amounts of data (gathered via peer-reviewed research, doctors, family health history, even your FitBit) and leverages machine learning technology to recommend data-backed patient care plans.
From a marketing perspective, AI technology can be used to help marketers make better decisions about where to focus their efforts, and to create smarter tools that present prospects with the right offer, in the right place, at the right time. Or, as Mitch Joel put it, to “create marketing [using] assumptions and knowledge that I myself may not even be aware of as a customer.”
Mitch says marketing AI goes beyond presenting relevant offers based on past purchases, but rather presenting offers that prospects will definitely act on or — at the very least — that the technology will “leverage and [learn] from to make it better and better each time — that sort of iterative learning that makes it better with each experience.”
Carl Schmidt, Unbounce co-founder and CTO, keeps a close eye on the digital marketing landscape and how AI and machine learning can — and in some cases already is — helping the digital marketer.
Carl says AI is already being used “to automate ad purchasing, create ad copy, score leads, identify customers at risk, select ad creative, run tests and more.”
Tomorrow (or in the near future rather) AI will empower marketers “to offer highly personalized marketing experiences,” says Carl. “Digital touch points (web, mobile, chat, automated voice) will understand the visitor’s context and preferences and construct messaging that is much more relevant; almost like a digital salesperson.”
Beyond that, Carl says AI could be leveraged to “deliver digital personas; online avatars that perfectly exhibit brand values and have a complete understanding of the brand’s products and services.”
Over the past 12 months, Carl and a team of data scientists and conversion optimization experts at Unbounce have been using machine learning to analyze hundreds of thousands of landing pages built in Unbounce.
The team believed they’d hit a significant milestone when they built a machine learning model that could predict whether landing page conversion rates are above or below industry averages with 80% accuracy.
But Carl and his team were eager to know how impressive that really was. Could human marketing experts do better?
Machine vs. Marketer: The challenge
Ever since chess champion Gary Kasparov’s stunning loss to IBM supercomputer Deep Blue more than 20 years ago, “man vs. machine” matchups have been used to gauge the advancement of AI technology.
In 2011, supercomputer Watson defeated Jeopardy! Champions Ken Jennings and Brad Rutter in an epic two-day match.
And just a few months ago Google’s AlphaGo defeated world champion Ke Jie in the ancient Chinese game Go. The game, which originated some 25 centuries ago, is known as one of the most challenging games ever created — one that you can spend your whole life mastering.
While we believe the future of marketing isn’t a question of “machines vs. marketers”, but, rather, machines helping marketers, let’s face it: everyone loves a good showdown.
A few weeks ago we welcomed 1200+ digital marketers to the fourth annual Call to Action Conference in Vancouver, Canada, just a couple blocks away from Unbounce headquarters.
As Carl said during his opening remarks, it was a perfect opportunity to test our AI technology against some of the savviest marketers in the world. But it also gave us the opportunity to “gauge customer reaction [and] educate and enlighten” (a.k.a. turn AI skeptics into believers).
“We’ve encountered folks who’ll tell us they’ll ‘never trust a machine,’” Carl says. “Or they ‘will always know better than a machine.’”
What they came up with was a mobile web challenge for attendees to test their conversion-predicting abilities against the machine they’ve been working on.
Here’s how it worked:
- Using their phones, attendees were presented with one of 204 Unbounce-built landing pages.
- Analyzing only the copy, the AI technology predicted whether the page had an above or below average conversion rate, as benchmarked against thousands of landing pages built in Unbounce.
- Participants analyzed the pages at the same time and were asked to make their own predictions.
But the human marketers were no match for “The Machine”. Like its predecessors (Deep Blue, Watson, AlphaGo) “The Machine” reigned supreme:
Not even the expert marketers were able to beat “The Machine”.
On paper, 2017 CTAConf speaker Joel Klettke is the perfect opponent. His expertise is conversion copywriting, and he promises his copy will “turn skeptics into advocates and prospects into paying customers” — a promise he’s delivered on for SaaS and B2B clients like HubSpot and WP Engine.
Because the algorithm only parses copy, you’d think these two would have a pretty fair fight, but no dice.
Joel, who got 57% correct, explains why he found the challenge so tricky:
The tough thing is knowing what to look for and getting past your own biases. Some niches, offers and designs just hit me as ‘yuck.’ Even as a copywriter, it takes some serious time to get into the shoes of the people you’re trying to write for; to understand what appeals to them or not.
Joel also mentioned that he didn’t know the machine only parsed copy until he was about 65% of the way through the challenge. Like many humans (and even those trained in copy) Joel found himself “swayed by design elements or polish that makes mediocre copy look better.” A machine, on the other hand, “isn’t fooled,” says, Joel, “but we try to analyze all elements to the detriment of reason.”
While we definitely do not (I repeat DO NOT) recommend firing your marketing department and hiring robots, the results provide us with a glimpse into the future of digital marketing, and how AI-powered conversion tools and insights will amplify our marketing efforts to build truly outstanding marketing experiences and better conversion growth.
Despite losing to “The Machine,” Joel seemed genuinely excited about what the challenge might indicate for the future of marketing and AI:
I think it tells me there’s room for some AI help out there. It’s not a wholesale replacement for research, critical thinking, empathy… but it’s a good barometer for how well you’ve put those elements together, at least from an algorithmic standpoint.
I’m excited to see what will happen when the machine gets better at accounting for context, like niches or types of offers. Many of the times that I beat the machine, it was because I understood how heavy a commitment the page was asking for and knew it’d drive conversions way down.
But, ultimately it opened my eyes a bit to how AI is going to gun for all of our jobs. Until then, we stand to gain a lot by playing nice with it.
Looking ahead to AI-powered conversion optimization
Yosem Sweet is Unbounce’s Director of Business Optimization. He’s been working closely with Unbounce’s data scientists to develop applications of AI to conversion rate optimization.
I asked Yosem to explain how this was possible:
AI-powered conversion optimization leverages a computer’s capability to process large amounts of data to find patterns. These patterns are then used to help the conversion optimization process by: reducing the time needed to generate winning hypotheses, reducing the effort needed to make better pages and (hopefully) finding unexpected solutions to conversion problems.
It’s interesting, no doubt, but how it might look is the cool part.
Yosem says there are a lot of different forms AI-powered conversion optimization will take, “everything from copy suggestions and auto-layout of content to strategic recommendations on driving specific traffic sources for a campaign.”
“The sky’s the limit,” says Yosem, “but before we get to the utopian future, we’d like to start by using AI to help marketers understand where they should focus their optimization efforts. Traffic? Copy? Design? Offer? And which pages have a lot of opportunity for improvement?”
Ultimately Yosem wants to “free marketers up to focus on the creative and strategic aspects of their job.”
Again, giving marketers back precious time is the key here — augmenting their toolkit to help them provide over-the-top amazing marketing experiences and, ideally, giving them superpowers.
And if you too think that, as Joel puts it, AI will be gunning for our jobs, Facebook marketing expert, author and CTAConf speaker Mari Smith begs to differ. She insists that there will always be a place for real-life marketers:
Us humans can sometimes be unpredictable and will always crave real, human-to-human connecting. Businesses that go the extra mile, that provide extraordinary customer service and excellent post-sales follow-up and that surprise and delight their best customers…these are the businesses that will stand out and thrive in the long run.
(Cue collective sigh of relief.)
What this means for digital marketers
Here’s the thing: Most marketers either don’t know what a good conversion rate is or can’t tell if a particular page can achieve its target conversion rate. This data just hasn’t been available, so even the most seasoned marketers rely on anecdotal evidence and gut instinct to determine these benchmarks.
AI technology can help bridge this gap.
I asked a handful of marketers what this type of technology might mean for them, and no matter how tiny (or robust) there team was, the reaction was the same: AI-powered conversion optimization would amplify their results and multiply their time.
Johnathan Dane is founder and CEO of Klientboost, a fast-paced PPC agency based in Irvine, California.
When he’s not speaking at CTAConf (he’s done so the last two years) he works with his clients to get them the best results for their PPC spend.
For Johnathan et al, AI-powered conversion automation would mean zooming out from the nitty gritty details and spending more time doing the things he loves.
We’d be able to strengthen our retention rate even more than it is now (one of our core focuses behind the curtains).
It would allow us to shift our resources towards other things that help grow our business, like using time to build partnerships, launch new marketing campaigns, create more of the content that has gotten us known on other platforms.
CTAConf attendee and marketing specialist Kelsey McFarlane of Orchestra Software shared a similar sentiment. The company she works for builds business software for craft brewers and distilleries.
Competition in her industry is minimal, since many of their competitors are not currently using digital marketing. But her team is small, so anything to amplify their efforts is huge.
We’re a really small marketing team — there are only three of us. So for us to gather the data to create the landing pages and then distribute [resources] to do A/B testing — we would be able to cut down on how much time is spent doing something that computers can already do for us. So it could streamline our team and make them pay attention to the more important aspects of what’s going on.
Joel on the other hand says he would spend more time not doing work.
A part of me wants to say I’d put that time into building out my other business ventures, trying to future-proof myself and make sure I’m constantly offering services that are relevant and robots can’t steal. But, honestly? I’d probably just get my projects done earlier, and then try to get outside. Life’s short, money is fleeting, and fresh air is important.
Start integrating AI insights into your marketing today
No matter how smart, no matter how scrappy your team is, if you don’t start leveraging the power of AI in your marketing efforts, you will be at a competitive disadvantage.
In the near future, AI will amplify your marketing efforts and multiply your conversions, but more importantly, it will free up your time to focus on the most creative and impactful parts of your job. .
Johnathan explains it best: “Too many people think that they need more money to grow their business (which is easy), carving time is the hard part, and that’s what AI would help us with.”
As Unbounce’s AI gets smarter and we learn more about the variables (copy, images, form fields, traffic sources, etc.) that influence conversion, we will be sharing our learnings and insights with our customers and readers.
We recently released the Unbounce Conversion Benchmark Report — an analysis of 74,551,421 visitors to 64,284 lead generation landing pages created in the Unbounce platform this year.
The report provides marketers across 10 popular industries — including real estate, higher education, legal and more — with data-backed recommendations around copy length, emotion and reading ease. More importantly, it answers the previously unanswerable question, “What is a good conversion rate for my industry?”
Start working AI technology and insights into your landing page optimization process today — download the Unbounce Conversion Benchmark Report and find out what’s a good conversion rate for your industry.