Like any other retail item, motorcycle shopping is seasonal. And when Harley-Davidson decided to try something new to combat the slow fall period, the New York division of the iconic motorcycle brand turned to artificial intelligence to boost sales.
In October 2016, Harley-Davidson’s first AI campaign, “48 Bikes in 48 Hours,” used AI platform company Albert to ramp up audience knowledge to meet the brand’s conversion goals.
Before using Albert, Harley-Davidson NYC’s all-time sales record was eight motorcycles in a single weekend. In its first two-day campaign, Albert almost doubled this record, selling 15 bikes.
“For there to be a technology that literally knows nothing about us and can come and outdo us in ‘his’ [i.e., Albert’s] first shot was incredible,” said Asaf Jacobi, President of Harley-Davidson NYC, through a rep.
Over six months has passed since Harley-Davidson NYC credited Albert with 40 percent of its motorcycle sales and is hiring 6 new employees to keep up with Albert-generated leads. Among the results during that time:
- After three months, Harley had seen an increase of 2,930 percent in leads/month — 50 percent of which were lookalikes
- Harley initially thought of their target audience in broad, buyer-persona based segments (gender, age, location) and projected only 2 percent of New York’s population to be potentially buyers. Albert identified new audiences by correlating online behaviors with their likelihood of conversion, and autonomously targeting them.
- Albert learned that Facebook ads with instantaneous product views generated a conversion rate 8.5 times higher than other Harley ads, so he discontinued low-performing ads and applied a similar user experience across new ones.
- Albert determined that ads with the term “Call” outperformed ads with the term “Buy” by 447 percent, so he began optimizing ads across all channels with this insight.
Now, as nearly a dozen more Harley-Davidson outlets are looking at using Albert’s AI programs to drive sales, we checked in with Or Shani, CEO of Albert (formerly known as “Adgorithms”) on what what it took to move so many bikes.
GeoMarketing: How did Harley decide to use Albert and AI? What inspired the decision?
Or Shani: Harley-Davidson of NYC is one of the top markets in the world, but their sales weren’t competitive with the other top markets. New York City is also an incredibly difficult place to break through using traditional advertising. Harley knew, for instance, that they only needed to target 2 percent of the population (their prospective buyers), but in order to do so through traditional advertising, they had to target the other 98 percent also. That was very expensive and not very effective.
After doing a number of radio ads and other traditional advertising campaigns, it was clear that digital advertising and marketing were the only methods that were going to let Harley really target the niche audience they were after, but they were unsure of how to go about it. Harley NYC’s CEO Asaf Jacobi, oversees all business operations, but wasn’t accustomed to making these kinds of marketing decisions. Suddenly, he found himself in this new position where he was trying to solve what were clearly marketing problems.
Asaf began researching various options and stumbled upon a news round-up of artificial intelligence technologies that were hitting the ad and marketing space, including Albert. Asaf had been reading about other applications of AI, and thought that it had to be the answer for marketing.
So when he saw this list, he knew he was onto something.
Then, he had a synchronistic moment. Asaf hadn’t gotten in touch with Albert yet, and was out in his neighborhood taking a walk when he ran into a guy on the street. Asaf started chatting with him and asked what the man did. He learned that he was speaking to Or, the founder of a company called Albert. Asaf laughed and told Or he had been planning to be in touch after reading about what Albert does. The rest is history.
What were the components of the campaign?
Albert performed a holistic multichannel digital campaign for Harley-Davidson NYC across search, social and display.
In general, Albert gets marketers’ campaigns out across devices (desktop, mobile), paid channels (display, social media, and search) and nonpaid channels (email).
What was the value of lookalike targeting in this campaign? (Basically, trying to get some context around the 50 percent lookalike figure).
Harley wanted to target “lookalikes” – or prospective customers that had similar demographic and behavioral profiles as their highest-value customers. They knew certain customers were likely to spend above a certain amount, so they wanted to target similar customers likely to spend in the same manner.
Albert accommodates lookalikes by integrating with existing CRM data and building out profiles of the types of customers he wants to target. He identifies these audiences and their specific behavioral patterns, and learns what will best serve them based on what they do and don’t respond too. Albert takes audience identification one step further than the traditional marketing practice of identifying buyer personas based on broad behaviors, gross sales and guesswork, to identify real customers based on their actual online behavior.
Additionally, Albert doesn’t take months or years to observe and act on these observations – he does it in hours and days. This quick learning is what resulted in such rapid and drastic results for Harley-Davidson.
What were some of the new audience segments that Harley hadn’t previously discovered?
Albert is able to identify customers and micro-segments on his own. He might create a segment of “users who spend 1-2 minutes on the site” and begin targeting them, knowing that they’re already engaged. He will also dismiss segments, such as “visitors who spend less than 15 seconds on the site,” knowing they’re a waste of his time. He’ll determine what calls-to-action and creative combinations work for individual customers, and will give overarching suggestions, such as “this group of customers is responding to headline A.”
For example, Albert found that consumers likely to purchase diamonds were also likely to purchase luxury motorcycles. He applied this insight to the campaign, helping to build the Harley brand among an audience typically not targeted by motorcycle advertisements.
Could this campaign, which was by its nature, short-term, be expanded? How did this the tight timing of the campaign determine the targets and focus?
Harley initially used Albert for their 48 bikes in 48 hours campaign – a campaign designed to eliminate seasonal overstock and determine how quickly Albert could meet the brand’s conversion goals. The tight timing of the weekend campaign, dictated the goals and KPIs that Albert needed to reach – primarily selling as many bikes as possible in the short window.
After seeing how successful Albert was – “he” nearly doubled Harley’s all time weekend sales record in NYC from eight bikes to fifteen – the motorcycle brand expanded their use of their AI partner. Harley has used Albert in other “burst” campaigns also primarily for moving bike inventory, and has incorporated the platform into ongoing longer-term initiatives such as campaigns for leases, calls, and e-commerce site/inventory.
Some of these longer-term goals include identifying customers with laser precision, understanding who customers are, and getting serious New York-area motorcycle buyers to complete a lead generation from on the Harley website. Harley then aims to convert these online leads into in-store motorcycle sales. Five other Harley franchises have engaged Albert’s assistance, and approximately ten more are in talks with the AI platform.