We’re all fizzing about the future this week as we officially went live with our very first Artificial Intelligence powered marketing campaign.
Specifically, we have implemented an AI media buying robot, to run the performance media for our friends at Lucky Rentals: a car and campervan rental operator offering great rates in NZ and AU.
We’ve been working with Lucky for a number of years now and have driven solid results along side them, seeing them grow from a fleet of 100 vans to more than 1000 vehicles across NZ & Australia in the last 5 years.
With the tourism market going gangbusters, and competition increasing significantly, we’re constantly looking for tools to help give Lucky a competitive edge, so when we heard about AI’s ability to significantly improve performance media results we knew we had to give it a shot.
The major claim is that he (yep, we refer to him as a person) ‘removes the silo’s between channels’. What this means, is that when Albert buys and optimises performance media, he’s looking at all channels and ad accounts as a whole rather than in isolation. So he’s able to not only analyse and optimise how each channel performs, but also how channels interact with each other. In some ways, it’s like programmatic but across all of your ad channels. This enables a more true drive towards performance optimisation.
The clever stuff with Albert goes far beyond cross channel optimisation. After providing ad account access (we’ve got him working on Adwords, GDN & Facebook), ad creative, a budget and very specific objectives (we’re working to a Return on Ad Spend target), Albert is given free reign to run the accounts as he sees fit. That means he will shift budgets cross platform, set bids, define audiences and targeting, add/remove keywords, turn ad groups on/off and any other activity that delivers the most efficient result based on the objective we set.
In addition to this, Albert will look for opportunities to improve marketing activity as a whole. He’ll analyse competitors sites, their bidding strategies, and their use of other channels to predict market opportunities outside of the accounts he has access to.
And he does all of this, all the time. He’s a 100%, 24/7 committed resource. It’s fair to say we think it’s pretty bloody clever!
The key thing Albert doesn’t do is creative. He’s able to pro-actively mix pre-approved copy and images and will objectively tell us when creative has worn out and will ask for more and he can give us insight on the best performing creative, but he doesn’t create anything himself. Given AI’s predisposition to err towards the worst of human behaviour when given creative freedom, we think this is a pretty good idea.
What did we learn?
Firstly getting a business ‘AI media ready’ takes work. The setup process for Albert was really robust, albeit relatively complex – this isn’t a plug and play solution, which has meant we’ve had to go all-in and have a long term commitment to using the tech. The tagging architecture and backend set up required a fair chunk of development work and back and forth with the team at Albert to ensure everything was set up perfectly, and we needed to load Albert up with a significant amount of creative to start with. On-boarding took longer than we anticipated, however the process in itself was highly valuable in forcing us to get to an agreed best practice – in essence it helped us to develop a marketing tech stack that works with AI media.
Interestingly our *human* media managers are loving it to. For Lucky they are now tasked with strategic and insight driven work – the high value stuff that they get the most enjoyment out of. There’s no risk of job losses either, in fact quite the opposite is true – our media team can now put their time and energy in to the creative things that humans are best at.
We acknowledge that relying on a robot to take responsibility for the key revenue driver for a business is a brave move for all involved and we’re thankful for Simon and the team at Lucky for giving this a crack.
It’s early days for Albert and time will tell if he’ll deliver the success we’re hoping for, but early data is looking promising. We think it holds huge potential for us and our clients and is symbolic of our commitment to really pushing to the edge of digital marketing.