The AI Q&A Series: Matt Garbutt, Creative Director at Brave Bison Unveils the AI Secrets of Digital Marketing
Get ready for our very first AI Q&A series, where we dive deep into the world of artificial intelligence and its fascinating impact on the ever-evolving realm of digital marketing.
We are excited to kick things off with Matt Garbutt, Creative Director at Brave Bison, one of our valued member agencies. Join us as we uncover their insights and expertise on how AI is shaking up the digital marketing landscape!
1. How has the role of a digital marketer evolved with the integration of AI technology? What skills or knowledge areas are becoming essential for digital marketers to excel in this changing landscape?
Where many tasks were manual and needed several people to complete them (e.g.: marketer, writer and designer), we are now in a place where individual marketers have the tools to do their jobs largely independently. Content creation and sharing has become automated, data analysis has become real-time, hyper-personalisation is possible and campaign assets are optimised in real time.
Marketers who have recognised the sea change have moved fast to adopt generative AI and become skilled at prompt design, overseeing data analysis and the use of tools like Jasper at the mass market level. Other tools like Hubspot and Salesforce are releasing new tools in their ecosystem ongoing. So, a marketer not only has to get familiar with existing tools, they also need to spend time researching and learning new tools. That’s a lot, but once up to speed, they’re finding that they can be the executive and strategist where the machine becomes the means of production, delivery, distribution and optimisation.
2. “Hyper-personalization” is a term used to describe the impact of AI on delivering customized content and recommendations to individual customers. Could you elaborate on the techniques and benefits of AI-driven hyper-personalization?
One of the things that Gen AI and the stronger prediction models we’re using, we get faster, real time inference and the ability to know people in specific clusters in detail and create ads just for them. We can test the ads with techniques like silicon sampling before our campaign goes live and then use the machine to optimise and iterate the campaign as it progresses.
We have been talking about the right context, right person, right time for a long time but now we actually have the tools to do this.
3. Automation is a key feature of AI. How is marketing automation evolving with AI, and what advantages does it offer in terms of campaign management and customer interaction?
One of the biggest examples is in data and data analysis. Now that we can use zero-party data and silicon sampling for predictive analytics, we have a much better-defined starting point to make a successful campaign. Once the campaign is live, I have mentioned some of the new practices and techniques we have to target, deliver and optimise campaigns further.
4. How does AI aid in understanding consumer preferences and targeting effectively? Could you provide real-world examples of how AI has helped you refine your targeting strategies?
As mentioned in previous answers, with prediction modelling and silicon sampling, we can use AI generated audiences to test and refine our targeting before the campaign goes live. With us, it has been a process of training and upskilling our team to use the machine to do this and build trust with it, to the point that we have begun to use these techniques in campaigns.
AI allows us to open up targeting giving more opportunity for smart bidding strategies to find potential customers. By analysing signals of intent, bidding algorithms will serve ads across the most relevant placements to the most relevant consumers. We refer to this as value-based bidding, bidding to the value of that sale as opposed to as many sales as possible.
We can serve highly relevant ads delivered by AI. Instead of providing set creative and ad copy; we provide headlines, images and video and AI will mix and match to serve the most relevant ad to that individual consumer at that moment in time. This also includes the placement the ad is served on which is a decision made by the algorithm.
It all hinges on measurement and fuelling algorithms with enough data to make those decisions – our role is training algorithms and setting campaigns up for success to allow algorithms to thrive.
Predictive audiences in GA4
It is like Minority Report – a tool which now allows us to group consumers based on how they are likely to behave in the future, rather than solely relying on their past behaviour.
We use a tool called Lunio to remove fraudulent click bots from impacting our client campaigns. Lunio is a bot detection and prevention software designed to help businesses identify fraudulent, invalid and unintended clicks. The platform enables us to assess traffic behaviour and define the threat level of traffic across networks on a unified interface.
5. What are some specific AI-driven techniques that you employ to analyze your clients’ campaign performance, behavior, and market trends for better decision-making?
I have mentioned this for Q4 but applicable here too.
Predictive audiences in GA4 are worth a scan here but essentially it is like Minority Report – a tool which now allows us to group consumers based on how they are likely to behave in the future, rather than solely relying on their past behaviour.
Data-driven attribution involves a huge amount of data and computing power to model which customer touchpoints are adding the most value to the digital marketing channel mix.
6. Given the potential for AI to shape the future of digital marketing, what are your key considerations while incorporating AI into any marketing strategies? How did you ensure a successful implementation?
There is a balance to be struck between quick and long wins. Over the last year, we have seen a multitude of tools released to market and while many of them are helpful, there is a lot of noise out there so it is important to retain your focus on what is really important. LLMs are that, along with what is happening in image and video generation and production at a broad stroke.
We also have to keep our eyes on legal. For example, we had a campaign parked at the last minute recently due to late-breaking concerns about IP ownership of the visual output of a Gen-AI-made campaign. So, we need to keep abreast of legal developments.
One important thing to note is that we never have or will work with clients or sensitive data in publicly available versions of any tooling. Neither should anyone else.
When it comes to implementing tools and tech into our workflow and agency stack, we are in the mode of exploration and testing, which will often lead to the ‘explorers’ upskilling their colleagues before we implement tooling off the shelf.
In terms of building our own tooling, we now have a dedicated AI team who are shaping the future of the company’s AI stack with both off-the-shelf and custom-built tooling which is being assessed and delivered on a priority use case basis. Watch this space.
7. Can you provide examples of AI-powered tools that you are using to streamline processes, and how do these tools improve efficiency and effectiveness?
We’re in the process of building out our own stack with an LLM underpinning it and then working on apps and agents ourselves and with partners to allow us to chat with our data, carry out better prediction modeling and creation of ads at speed, scale and deeper personalisation, to give an idea of a few. All these tools give us powers we previously didn’t have and then save us huge amounts of time for tasks such as report creation, campaign analysis and many more.
We built a proprietary copybot based on GPT two years ago that we have been iterating. Of course, we use OpenAI’s tools for non-sensitive day-to-day work and others in the public domain such as Midjourney, Runway, D-ID plus other lesser known tools.
We find that we can research, plan, strategise, execute and produce faster, up to 40% as a given average. The machine gives us happy accidents and nice surprises to work with too, from a creative standpoint in particular.
8. Among the services you offer, in which areas do you harness AI technology’s potential the most? Could you elaborate on instances where AI has been a game-changer in enhancing the delivery and effectiveness of these services?
Creative and Performance Media are our biggest specialisms, ML has been helping us for years in Paid Media, with the inbuilt features of the big tech services we use, for campaign planning, audience building, testing and optimising. Albeit that is now shifting to a wider and more proprietary view with us building our own tech, which gives us better clarity on our own and clients’ data to deliver ever more optimal and tighter-targeted campaigns and delivery.
When it comes to creative, we ran a successful community-led Gen AI campaign earlier in the year with WWF that gained a lot of exposure and was a big success.
9. According to the poll you have run on LinkedIn, even though the striking majority of %67 said they think that AI will make us faster and more productive, the remaining 33% were either skeptical or pessimistic about emerging AI technologies. Why do you think people are concerned or hesitant about AI?
I imagine people are concerned for their jobs immediately and then possibly concerned about doomsday scenarios like the Terminator etc. when it comes to AGI. Also, a lot of professionals in the art, design and creative fields are concerned about plagiarism.
10. What do you think about emerging “AI Automation Agencies”, are they a goldmine or another bubble that will pop soon enough?
I think they are an obvious output of what has happened over the last year. We are going to see a lot more solopreneurs and small teams rise up and challenge bigger companies because there is so much more power and potential at peoples’ fingertips now. As with any other sector, the cream will rise to the top.
11. As Matt Garbut, Brave Bison’s Creative director highlights (hAIlights?) it is a time of “adapting or getting left behind.” What would you suggest to other digital marketing experts to adapt to this AI wind?
Follow your curiosity. Explore and experiment. Get comfortable with the tech. It will be a revelation and you’ll understand that the world has changed.
This is like the early days of the internet all over again.
We are ALL late here and we are learning as we go along. With Gen AI, we have created more of an organism than a piece of tech.
Explore and understand: Audit your own stack, systems, channels, practices and propositions, does any of it look like it’s ready for better efficiency or economy with Gen AI assistance? Take time to grasp the various AI applications and their evolution. This involves distinguishing between simple task automation and advanced machine learning, as well as standalone applications versus those integrated into larger platforms.
Stagger your Implementation: Start with rule-based, standalone applications that aid decision-making. Over time, transition to more sophisticated, integrated AI systems for customer interactions. Or, start bigger by building a platform for your ongoing AI stack with your own, secure Large Language Model implementation in your own domain. Then add apps and tools on top of it.
Ethical and Legal Implications: Ensure that AI tools respect data privacy regulations and avoid biases that could lead to discriminatory practices.
12. What potential challenges and obstacles might arise when integrating AI into marketing campaigns, and how can agencies address these challenges effectively while leveraging AI to maintain a competitive edge in the evolving role of digital marketers?
I have touched on some of the legal concerns. Otherwise, we have seen ethical concerns aired about human replacement. Then of course, there are challenges around expertise, with this tech being so new, how do you separate the good from the bad and make choices about what to use and adopt? Who can you rely on to give you good advice as there are not many bona fide AI experts out there?
We are all finding our way and learning this as we go to a certain extent. We are at a point where I think the brave will win and those who decide to invest in the tech and create teams around it to explore and build new tech will win out.
13. How will companies address data security and privacy concerns in the near future? What strategies and best practices can be adopted to ensure ethical AI usage?
Data security is one of the biggest issues but clarity on that is fairly simple. Do not expose any data to any public AI tooling. If companies want to use LLMs and apps like Advanced Data Analysis for their or their clients’ data, they are going to have to build their own LLM stack and keep it in their own domain or risk legal and IP meltdowns.
Regarding privacy, we are entering a world where, as marketers, we can use a relatively small dataset to input into the machine and quickly get augmentation, deep analysis, inferences and insights from it – much more than we would ourselves in the hours we have in any working day.
Clearly, data is going to be an even greater and more precious commodity going forward but we are moving to a Web3 world where individuals have much greater ownership over their own data. Therefore, companies are going to have to be much more transparent about what data they harvest, why, and how. We will likely soon be in a place where people are rewarded for sharing their data, not spammed with ads.
When it comes to AI training data and privacy around that, the big players have (by their admission and PR) anonymised the training data used.
So, the rules will likely be some of the below:
- Opt in data collection
- Transparent data use
- Data anonymization
- Bias Detection (mitigating bias in AI models)
- Encrypted Data Processing (AI models that process data while it is still encrypted, keeping user details inaccessible)