Hero Conf London was one of the last stops in our world tour of conferences this year and it was certainly one of the best. Acquisio attended as a speaker and gold sponsor, and we were delighted to host the drinks reception (featuring our signature drink the PPC Breeze and our ADlibs napkins!). But the best part was chatting all things PPC and machine learning with attendees!
— Acquisio (@acquisio) October 22, 2018
We wish we could have brought you with us, but since we couldn’t, here is the last installment of the series. Enjoy these session takeaways from one of our favorite events of the year!
Takeaway 1: Understand Where Human vs. AI Strengths Are
Session Title: The PPC Expert’s AI Handbook
Presenter: Fred Vallaeys
Fred advised that we would all do well to begin leveraging the power of AI. In his talk he outlined the ways in which human ingenuity is put to best use and what tasks are highly suited for machine learning. He also covered important questions to ask when analyzing data. His ultimate conclusion? AI and humans have complementary roles to play in the future of PPC.
Even though humans still excel in disciplines like analyzing keyword and ad text data, proper tech and information processes can be as important as hiring A-players on your team. Sometimes, machine learning outperforms the human mind, especially when it comes to things like automated reporting and bidding. In those cases, letting AI do its thing will allow you to skip the steep learning curve. Free yourself to focus on real value-add work!
When examining machine learning data, make sure that:
- You are asking the right questions
- That the available data is the right data for the question
Fred brought up a problematic WWII practice of reinforcing fighter jet fuselage only in places where planes returned with bullet holes. This was a faulty reading of the data since doomed planes that did not return were probably damaged in different, much more crucial spots. As you can see, even with all the data in the world, you are headed down the wrong path if you are not asking the right questions!
Takeaway 2: Automated Bidding Should Be Used Strategically
Session Title: Why You Should (Not) Be Using Automated Bidding Strategies
Presenter: Sven Wilhelm
Sven cited some examples where he was burned while using an automated bidding script (made by one of our competitors) that got out of hand with runaway CPCs. He also showed a number of examples where target ROAS optimization “didn’t work”. In these examples the scripts worked as advertised but had other side effects. Sven says they should have “read the not-so-fine print on the bidding strategy”.
To avoid getting into trouble with bidding scripts, Sven recommends not using optimization for:
- Cases where lead values differ (call vs. form fill)
- Cases with ad-hoc requests (anecdotally many Google clients turn off bidding strategies for special holidays, eg. Black Friday)
He also advises campaign managers to structure their accounts for automated bidding, grouping things by similar CPC or CPA. Similarly, he says that if you’ve been using automated bidding and you want to turn it off, check where bids are set as they could be surprisingly high.
For more information on how to navigate automated bidding and make it work for you, check out our automated bidding ebook.
Takeaway 3: Let AI Take the Wheel for Mundane Stuff, But Understand Its Limitations
Session Title: Machine Learning Tech You Could and Should Use Tomorrow
Presenter: Stephen Kenwright
“Machines are not here to take our jobs, they are here to enable us to take other people’s jobs,” – bold quote from Stephen Kenwright
Stephen suggests letting machine learning take over mundane tasks and focus on things that will differentiate you. According to Gartner, by 2020, “customers will manage 85% of their relationship with the enterprise without interacting with a human.” But PPC machine learning has limitations too, so be aware of them! He also had some suggestions about which machine learning tools to choose.
Before using machine learning, make sure you have a solid analytics practice in place! Then you should use machine learning for things like:
- Mining search queries or social posts
- Analyzing user complaints to predict churn
At the same time, make sure to take note of machine learning’s limitations. Some of the AI out there does not fare so well with:
- Seasonal campaigns
- Brand safety issues
- Campaigns requiring manually customized audiences
Also, when choosing a tool, be wary of Google. They may have the best tech, but they also have many people working towards their own private goals, which do not necessarily align with yours. Their profits come before helping their users. Stephen recommends avoiding Google altogether and trying independent platforms, mentioning Hero Conf exhibitor Acquisio as an option (we agree!).
Takeaway 4: AI is Taking Marketing by Storm, Future-Proof Your Brand
Title: Keynote: AI is Smarter Than You: Adapting your Retail Strategy to Keep Up
Presenter: Cady Condyles
What if we could build computers that one day could see, hear, talk and understand human beings? – Bill Gates, 1991
Based on Microsoft’s internal learnings, Cady provided predictions for the future of AI in marketing and how to leverage it.
Microsoft made the shift to AI for almost everything they do – from being able to detect the possibility of cancer based on user search queries, the creation of the AI for Earth program, Seeing AI to help those with impaired vision navigate, to the creation of an AI Rembrandt painting that fooled art experts, and more.
Bing is Microsoft’s largest AI application. Its current market share in Australia, the UK, and the US are 12%, 25%, and 35% respectively. Bing search was designed to not use autocomplete which can reinforce personal biases. Instead, Bing provides a selection of refinement options, allowing for greater objectivity.
AI is essential for the engaging customer experiences, gaining data insights, and enhancing marketing operations. Cady shared the prediction that by 2020, 85% of businesses will use AI and 30% of web browsing will be screenless (think digital voice assistants). By 2025 95% of interactions will be powered by AI. So what’s a marketer to do?
To future-proof your brand for the AI era, you’ll need to be in the right place at the right time when people are looking for answers. Here’s what you should do, according to Cady:
- Optimize for voice search
- To be discoverable, reach consumers where they are and use conversational agents
- Use cognitive services so your technology can perceive and understand the world around us
- Use intent-based AI to identify and reach your shoppers
Takeaway 5: Take AI Pointers from Top-Performing Marketers
Session Title: What We Can Learn from Award Winning Paid Search Campaigns in the US and Europe in 2018
Presenter: Anders Hjorth
Anders discussed how paid search gets more complex every day. He shared the knowledge gained after reviewing a group of award-winning PPC campaigns using new strategies and techniques that helped them gain an edge over the competition.
Top marketing teams make sure to leverage AI everywhere they can and have strong knowledge acquisition and sharing mechanisms. Here are some of Anders top tips:
- It’s important map your campaigns to user journeys
- If you’re in ecommerce, make sure you use Amazon Ads (it now rakes in about 10% of Google Ads)
- Use a data-driven approach to understand your audience and strategize
- Innovate by combining existing tech in new ways
- Use available technology to drive online users to physical stores using radius targeting, location extensions, and Maps campaigns
- Use negative audience targeting in a similar way as you use negative keyword targeting
- Take advantage of new functionalities in paid search like RLSA, store visits, dynamic search ads, and in-market audiences
That’s a Wrap
That wraps up our top 6 takeaways from our favourite speaker sessions at Hero Conf London 2018! We hope you feel as inspired, and motivated to dive into more automation and machine learning strategies. Some of these speakers said some pretty bold things, but we have to agree with one overarching thing: machine learning is here to make our lives easier, not harder! It’s just a question of trying, learning, and growing.
All screenshots by author via Hero Conf slide decks, taken October 2018.