In our guest blogs we feature leaders in market research and marketing, sharing their views about best practices, innovation, and emerging trends. The topic of AI is dominating industries across the board, including marketing and marketing research. In this post, MRII Executive Director Ed Keller speaks with Sarah Fay, partner at Glasswing Ventures, where she leads and evaluates investments in early-stage AI-powered companies about AI and the future of marketing. Sarah is also the Chair of Ziff Davis, and prior to joining Glasswing she was CEO of Aegis Media North America (which was acquired by Dentsu in 2012.)
QUESTION 1
Ed K: You have a lot of experience in the world of AI, and you have three predictions about the impact of AI on the world. Let’s look at the first one – AI will change the creative process. Can you explain how?
Sarah F: Yes, but to consider the role creative AI might play, it’s helpful to look at the other side of advertising: media. Media practices have used various forms of machine learning and AI for years. As such, media planners and buyers have come a long way and are adept at using data and programmatic advertising. Today, about 72% of media budgets are programmatic. Compared with manual media placements, these campaigns have typically improved ROI by around 20%, so AI has generated real results in media. But media teams have depended on the creative given to them to deploy these campaigns, so the returns have been based mainly on using AI for targeting. And because of the recent deprecation of favored tracking methods, such as cookies and Apple’s IDFY (Identification for Advertisers), programmatic media has taken one step backward and won’t likely be where to turn for the next mother lode of ROI. Creative AI will likely be the next method for a significant bump in returns.
QUESTION 2
Ed K: In short: data-informed decisions are creating results on the media side. But what about the creative side?
Sarah F: Several studies have shown that more than 70% of an advertising program’s success depends on the creative. It matters which visuals or products you show viewers – what experience you create for them, and how the brand or product is made relevant to the viewer. The advertising industry has always made this a priority with insights teams generating data and information about customers. Creative teams have used those insights to create relevant content that will create a deeper connection with audiences, leading to improved ROI. But the insights process has been time-consuming, using focus groups or contacting people by email or phone. Getting valuable insights needs to happen more quickly, like in real-time.
AI can do three key things to generate higher ROI using creative:
- Provide insights that allow creative professionals to get started faster in architecting their stories and designs.
2. Leverage data to understand what creative is working best, driving desired behaviors from specific audience segments.
3. Automate the creation and versioning of these campaign elements.
Observed behavior in the wild can deliver insights to algorithms on what type of creative should be delivered to viewers in real-time to get the best response. AI can help Creatives produce the most favorable creative selections and combinations of copy, images, audio, and video for audience micro segments. Programs can be set to learn and morph to augment creative in real-time.
Some people wonder if AI will replace advertising creative developers. I submit that AI will make superheroes out of human creators. Tools are being leveraged to augment advertising toward what works, to feed the insatiable beast of fresh content needed, and to reduce the time it takes to turn out great work. Clients are asking for more while wanting to pay less. Only AI will make that possible. But AI’s talent is synthesizing data – using what has worked in the past and refining/regurgitating winning combinations. Talented creative professionals are better at imagining what has never been created before, what isn’t out there to be copied. The human driver of AI tools will be able to imagine, create and test new creative ideas faster than ever in the history of advertising. I think this new way of working will energize many, and the world will see the most fantastic advertising ever.
QUESTION 3
Ed K: Of course, ChatGPT recently released v4 and has been heavily talked about. How do you see its impact on the creative process?
Sarah F: Suddenly, everyone gets it! ChatGPT has shown the general population what generative AI can do. Generative AI content creation platforms can now reduce hours, days, weeks, and even months of manual effort into minutes – it is truly miraculous. These platforms leverage vast amounts of data to create content that is contextually relevant and stylistically consistent with the brand or person they represent. Most generative AI platforms require human oversight or “a human in the loop” to ensure the generated content is accurate and meets the intended purpose. For instance, a platform that generates product descriptions for an e-commerce website may need to be monitored by a human to ensure the descriptions or images do not misrepresent the product. The humans also help to train the algorithms to become more accurate over time. The quality of a generative platform depends mainly on the quality of the data it uses to train its algorithms. Specialized generative technologies are distinguished by their ability to learn and replicate a brand or person’s desired voice and style. This means that a platform trained on a specific set of data, such as a brand’s marketing materials, can produce content consistent with the brand’s tone and messaging.
Other forms of generative content creation (think Midjourney or Dall-E) can be leveraged to create dynamic, eye-popping creative effects with a few clicks. A creative professional would have formerly poured hours into their design, animation, video, and music projects. This translates to a creative team’s ability to generate magnitudes more content. However, these are still the tools to bring a professional’s dreams to life and not the replacement of that person. The quality of the content will rely upon the creative ideas, sensibilities, and awareness of how the creative is driving results.
QUESTION 4
Ed K: Your second prediction has to do with automating decisions – could you tell us about that?
Sarah F: Every manual process that involves a spreadsheet will become automated. If you consider the way marketing research analysts have traditionally worked, a lot of time has gone into gathering information. Analysts are like detectives, seeking data and entering it into spreadsheets. When the data is all neatly assembled, the analyst eyeballs that information and considers the data relationally to make “best judgments.” The most time-consuming part of the process yields little value, and the insights that are gleaned are imperfect – given they are based on the amount of data that can be held in a human brain. In the future, data will be generated from audience actions and signals and automatically ingested into an algorithm, which may put the insights directly to work in a generative advertising program.
Even complex decisions involving many data sources will be made by machines. AI can make complex decisions faster and more accurately than humans. Think about media planning, for instance. A media planner considers many different bodies of data, including audience research, attribution reports, and unique brand considerations such as context, price and KPIs. A massive amount of information is assembled in a spreadsheet where the planner looks for indications and direction on how the budget should be allocated. The recommendation of the media plan usually sounds something like this: “Based on the data we’ve gathered and results achieved in the past, we believe that this plan represents the proper budget allocation.” Of course, I’m simplifying the conversation, but it comes down to a human assessment – a justified opinion – and there is no prediction to make the recommendation accountable. AI will offer an absolute prediction of ROI and tell a planner how that ROI will change with every added element of a plan. AI will learn how results match the predictions over time and recommend additional ideas to test and improve media plans. We invested in Elsy (now part of VideoAMP) which uses AI to create media plans and predict ROI, reducing planning time dramatically.
Media planners will still have plenty to do in a time-starved business. No client has ever valued their late-night hours on spreadsheets! Newly found time might be better used to negotiate creative brand activations with media partners that AI has identified.
QUESTION 5
Ed K: That is the key – it’s not about replacing people; it’s allowing them to focus on what they’re good at and using machines to do what they’re good at. Your third prediction is about how intuitive AI will make life easier. Can you explain how?
Sarah F: Well, consider how much technology has changed our lives over the last few decades. Can you imagine life without GPS? Or not having all your contacts within easy reach? Or having to go back to watching programmed TV? Now think of AI providing as much change to our lives in the coming decade. AI intuitively knows what we need. Already my iPhone tells me I should leave soon for an upcoming appointment that requires travel. Already Google Search guesses the longer phrase of what I am searching for, so I can just hit enter. The various writing technologies I use are correcting my grammar and suggesting words and phrases. I didn’t ask for these things or buy new devices to make them possible. AI crept into my life as it has done everywhere with everyone. And now, I expect the ease of asking a question or giving a voice command wherever I am – in a car, in the kitchen, watching TV… and getting the desired response. My face is now my password, and it suddenly seems inconvenient to type one in. And, I remind you, we are at the very beginning of all this. When my firm, Glasswing Ventures, was founded in 2016, we predicted the AI innovation wave would be bigger than the Internet, and I hold that will be true.
QUESTION 6
Ed K: At MRII our focus is on marketing research. What are the most interesting applications of AI in market research that you’ve seen, and what’s on the horizon that intrigues you?
Sarah F: Glasswing invested in Craft which uses AI to provide a holistic picture of any organization. From growth over time, social presence, and brand trends, to market positioning and the latest company news, Craft gives a view of key statistics and how they compare competitively. This is all updated in real-time.
We also invested in trender.ai, a research tool for social selling. It helps salespeople identify ideal prospects and monitors them for the right time to reach out.
Some other companies we’ve seen and liked include:
- Revuze – gathers insights from thousands of written user reviews and condenses the findings to represent key insights about how customers perceive products.
- Veritonic – tests and measures the ROI of audio assets and campaigns, pre-market, in-market, and post-campaign.
- Magellan – uses speech recognition to find and report on digital audio advertising in the voices of podcasters and radio hosts. This can be used to understand the competitive advertising landscape or the best places to advertise in the highly fragmented podcasting world.
- Vizit – uses computer vision AI to understand which images will appeal most positively to specific audiences.
I can give you a very long list, but these are some of my favorites.
QUESTION 7
Ed K: Going back to the popular topic of ChatGPT, how do you see it impacting the work people are doing in marketing and market research? Is all the attention well founded, and will it be transformative; or should we stay cautious?
Sarah F: ChatGPT is already changing behavior. It has been the most quickly adopted technology in history, gaining over 100 million users in a few months. A college student (who will remain nameless) recently confided, “If I have a three-page paper due the next day, I’ll Chat that. But I will edit what it writes so it sounds like me. But ChatGPT is getting pretty good at sounding like me.” Note the student used “Chat” as a verb, just as we do with Google. This behavior is a little scary because what will happen when students don’t need to work very hard to write papers? Can there be an upside? Maybe the most successful graduates will be the ones who understand how to use AI to their advantage.
ChatGPT can be leveraged by anyone who is tasked with writing marketing content. They can use ChatGPT to get started. After all, it’s easier to begin with something to edit versus just a blank page. Just as it is with any writing task, refining is usually needed to produce the best outcome. Marketers will need to learn the right prompts to massage and improve copy through ChatGPT. And in the end, they may need to provide final edits.
GPT technology is also foundational to many paid generative SaaS tools which can be used to perform specific marketing tasks – such as creating videos, websites, landing pages, presentations, and even writing better blog posts and emails (which also can technically be created by ChatGPT). Similarly, market research professionals can use generative platforms trained to gather data to produce insights – it is possible to quickly create questions and quizzes that reach large audiences quickly, providing valuable information on interests, engagement and buying motivations. And ChatGPT may help to quickly and easily summarize findings.
One thing is clear: while innovations and tools can help to improve our lives, there will always be a role for people. That means it’s crucial to stay on top of the latest trends and insights and ensure you and your people have the solid foundation and access to advanced knowledge to succeed. To learn more about MRII and our online market research courses with University of Georgia, click here.