Late last month, Peyman Nilforoush and I were lucky enough to do an artificial intelligence (AI) presentation at J.D. Power’s Automotive Marketing Roundtable in Las Vegas. We’ve been fortunate enough to partner with many auto brands over the last 18 or so months on their content amplification across over 40 social and native advertising networks.
This partnership has produced millions of fascinating data points to dissect and analyze. As a result, for the first time we presented this industry benchmark data to a multitude of auto executives and their agency counterparts. The benchmarks included click-through rate (CTR), engagement rate, cost per engagement (CPE), and average engaged time in seconds.
The live-action stage presentation can be viewed here. Below is the same presentation, but given exclusively over the slides.
The Full J.D. Power AMR Speech
The Full J.D. Power AMR Slides
Never has there been a time in human history that man has been able to teach machines to learn on their own. That said, AI is still just in its infancy. AlphaZero AI, a sibling of Google’s DeepMind, taught itself how to play chess without the aid of human knowledge in under four hours and beat the world champion chess program.
What is AI and how does it work?
Former Walmart CMO, Julie Lyle, famously once said, “she who learns fastest wins.” She was right. In business today, the person that learns the fastest does win. AI can be our competitive advantage in marketing when used properly. Unfortunately, there’s a multitude of AI solutions that are merely marketing ploys and don’t really provide the value add that we seek.
For example, you know when you upload a picture of family and/or friends to Facebook and it puts a box on their faces as a prompt to tag? That’s AI. Does that feature provide real business value? Would you pay for it?
My good friend, Paul Roetzer, CEO and founder of PR 20/20 and the Marketing AI Institute likes to quote Demis Hassabis, Co-Founder and CEO of Deepmind, when defining what AI is.
The science of making machines smart.
I like this definition because I’m a “keep it simple stupid” (KISS) type of guy. Paul further goes on to say that this is important because it augments human knowledge and capabilities. This is where our potential competitive advantage comes from.
We could get into buzzword fan fair at this point, but it’s not necessary. KISS… Don’t let the buzzwords distract or confuse. AI is simply the science of making machines smart. Making them learn on their own without the aid of human knowledge.
Content distribution and the buyer’s journey
According to the Havas Group 84% of consumers expect brands to create content. For auto buyers in particular, J.D. Power found out that the average consumer spends over 13 hours using online content for research purposes before making a buying decision. For that same consumer, according to Forrester, read and/or watch 11 different online content assets. This is good information to have when putting together a content strategy.
There’re two major problems, however.
Last decade it was easy for most brands to create lots of online content and have organic search and social drive all the traffic they needed. Fast forward to today – that cat has been let out of the bag. Most brands realize they need to be creating online content. And as a result, they are. That means the web has more information on it now than at any other point in its short history.
And it’s only growing as time progresses.
The above screen shot represents a 24-hour clock taken just before 5:00 PM EST. Three years ago, there were just over 2.8 million blog post published every day. With over seven hours to go it’s almost at 3.6 million blog posts published today.
In this scenario it’s difficult for many brands to get the organic visibility they were used to last decade. There’re only 10 spots on the first page of Google and social media has universally and massively reduced organic visibility for brands. That leaves earned media and paid media as the only other means to drive visibility.
This brings us to the second problem – interruptive ad experiences. It’s no secret that consumers dislike interruptive advertising. They’ve turned to ad blockers or simply ignore it. There’s so much disdain for interruptive advertising that South Park even made an episode reviling them.
It’s these two fundamental problems that helped give rise to native advertising as we know it today – manual buys on each individual networks chosen by the advertiser. This can be problematic because it doesn’t easily scale, there’s a new and separate dashboard for each, and measuring across platforms isn’t universal.
It’s also necessary to manually choose (educated guess) the demographics and the targeting on each one of the above channels. Advertising has long been an exercise in trials and testing, but it can cost millions of dollars to find the right audience that’s optimized toward content engagement.
The above scenario is also built on the click-economy. That means the incentives for the exchanges and the advertisers are concentrated in and around the click and not real content engagement. Based on the statistics mentioned above concerning the buyer’s journey and content consumption shouldn’t the incentive be on engagement and not clicks?
We’ve found zero correlation between CTRs and engagement rates and in some cases, we’ve found a negative correlation. This is caused by bots faking clicks. The click-based economy described a above is rife with waste. According to Chartbeat Analytics, two out of every three native advertising clicks bounces off the content within 15 seconds.
This is important to note because at 15 seconds 70% of folks consume 80% or more of the content. That means that 66% of budgets spent on these native advertising networks is wasted. By 2020, according to Adyoulike, $85 billion will be spent on native advertising globally. That amount of waste is unsustainable.
Artificial intelligence is the solution
As mentioned above, AI helps augment human capabilities and knowledge. When it comes to large volumes of data, AI can manage it better, faster and stronger than people. So rather than manually choosing networks, demographics and optimizing towards clicks, why not let AI test and choose the networks and audiences while optimizing toward content engagement? AI can test and learn better, faster and stronger than us.
Just promoting and testing a handful of content assets online can lead to thousands of ad unit variations and permutations when you consider headlines, images, copy, gender, devices, interests, networks, etc. AI can test and target better and more efficiently. In fact, we’ve seen improvements in efficiency up to 80%.
What the numbers say – with and without AI
The below data benchmarks for post-click content engagement is based off of millions of dollars in ad spend. It’s important to note that content performs differently across verticals. The top line of numbers is what a human would likely achieve without the assistance of AI trying to optimize for engagement. The bottom numbers represent the performance using AI while optimizing towards content engagement.
As the numbers clearly show, AI outperforms humans every time. It’s a significant performance boost, too. Using AI in a CPE model means that budget is only spent on real content engagement. That’s how AI can eliminate the 66% of waste in a click-based economy.
There’s another benefit of using AI to amplify content – it get’s smarter over time. That means that the targeting will get better and the CPE price will fall. The savings is then reinvested driving even more engagement over time. Below is an example of how it works:
There’s no question that nearly all auto consumers use online content when navigating the buyer’s journey in order to make a purchase decision. This is a good thing, but challenging nonetheless, because it’s becoming increasingly difficult to get content visibility online. With native advertising in a click-based economy up to 66% of every attempted engagement bounces. The only other paid media options are interruptive in nature and consumers don’t like it.
The auto industry is now using AI to optimize in an engagement-based economy. It’s content engagement that advertisers want. With no correlation between clicks and engagement why do we still have click and impression-based pricing schemes? AI can eliminate CPMs and CPCs. It’s the use of AI, as described above, that can help eliminate waste in paid media and deliver the greatest efficiency.