ChatGPT takes the agency world by storm.

On November 30, 2022 OpenAI captivated the world by putting ChatGPT in the hands of the public. Since then, if you work in an agency environment, conversations around ChatGPT and the explosion of AI-powered tools have dominated your water cooler chats and Slack channels, as well as your more serious, forward-looking planning and innovation meetings. Have we used it yet? What are the best ways to use it? What can we do? What shouldn’t we do? Same.

Seeing this as a huge opportunity, my Kettle colleagues and I took a good look at those questions, each of us looking through the lens of our specific disciplines. Honestly, we found too much. So many tools, so many use cases. But we persevered to cut out much of the noise, and narrowed it down–for us and for you–into a series of posts where we’ll be sharing much of what we found.

Since every Kettle endeavor typically starts with strategy, it felt like the best place to start. And I suppose this is a good time to introduce myself. I’m Eric. I’ve been a Director of Strategy at Kettle for a few years now. If you feel so inclined, shoot me a hello on my LinkedIn profile. Now, before we get into the ways we think AI can assist us in our research and strategy work flows, let’s start off with a little background for perspective.

Now an entirely new set of creative and strategic minds can use AI for a much wider set of tasks across a wider set of disciplines.

Why the AI pandemonium now?

I feel an obligation to point out that AI has been in use for years, powering many useful tools and applications. Siri is a perfect example. It’s powered by AI, but we didn’t freak out about it. In the agency context, AI also makes Google Analytics and Adobe Analytics the powerful tools we know them as, and we just accept them as the standard. So why is today different?

Well, frankly, no one wants to be left in the dust. And everyone wants to capture the value that new AI tools can bring to their organization. At a minimum it’s just to keep up. Ideally, it’s to gain an edge, speed things up, make them more efficient, less expensive, or maybe even more intelligent.

But today is also different because now everyone has access to the power of AI. For free in many cases. And it’s easy to use, because generative AI tools like ChatGPT and Midjourney are based on natural human language models. We can use our normal way of speaking to ask them to perform a variety of useful tasks.

There’s also a very creative aspect to generative AI that did not exist in many of the older applications. No longer are AI-powered tools only in the hands of the technically advanced analytics teams. Now an entirely new set of creative and strategic minds can use AI for a much wider set of tasks across a wider set of disciplines.

And to that point, here are a few ways we think AI can help with research and strategy tasks–in the agency environment and beyond. Give them a look, let us know what you think.

Ignite your desk research process

Like one of those quick start fire logs we once used, before we knew how bad they were for the environment, ChatGPT can help you ignite your desk research process. Ask the right questions and you can narrow down your initial search more quickly than using Google alone, and shorten the time it takes to get to a baseline for where you think the research should go next.

Use it to pull together a picture of the brand landscape that you need to understand right at the onset of a pitch. Ask it to give you an overview of consumer needs and behaviors in a certain category before you have the official data, just to get some ideas flowing. Or ask for an overview of the competitive set, or recent press. The list goes on.

Get really good at writing the right prompts and you may find yourself having a good giggle about how fast it comes up with the snapshot you’re looking for. From there you can decide what deeper research needs to be done. Because remember, that quick start log always got you started quickly, but you always had to add in a stack of real logs to have a real fire. And so goes our recommendation here as well.

IMAGE: AI-powered trend signals in the Synthesio platform

See around corners

If there ever were a superhero who could see around corners, their origin story would definitely be based on some research incident that happened during a Strategy Team all nighter. If only, right? Well, with AI, we sort of can now. The more straightforward name for this super power is predictive analytics, and it’s powered by AI.

Consumer intelligence tools like Brandwatch and Netbase have been around for years, giving us a look into what consumers are saying, liking, complaining about, and doing online. But once AI got baked in, the game changed. With predictive analytics we can now, well, see around corners.

AI-powered consumer intelligence platforms like Synthesio can help us see trends forming before they are fully formed. They take today’s data, compare it to past trends data, and decide if it has the signs of becoming a trend in the future. That edge can be incredibly powerful for your team if you want to spot trends as they’re emerging, and get ahead of your client’s competitors with ideas that just somehow magically resonate with your audience. And if you want to sound very official at the next meeting, you can call this new magic AICI. It’s short for AI-enabled Consumer Intelligence.

Cut through the noise

Today we live in a world of data. Data can help us arrive at powerful insights, and it drives many of the decisions we make in the agency environment. These days we sometimes even use the term BIG data because it’s, well, a lot of data.

When that data is untamed and unstructured it can be difficult to find the signal through the noise. Now AI-powered text analysis tools like Deep Talk can help you cut through that noise, and help you uncover valuable insights.

Take customer reviews, feedback, or open ended consumer survey responses as a few examples. Manually sifting through hundreds of lines of unstructured text and trying to extract meaningful patterns can be difficult, and time intensive. But AI-powered tools can now help you identify key themes and trends faster, so you can identify innovation opportunities faster.

IMAGE: AI-powered creative testing dashboard in Kantar’s Link AI

Test creative, quickly

When thousands, or sometimes millions, of dollars of ad spend are on the line, testing campaign creative with a sample audience before it goes live has become standard practice. After all, we want to ensure the success of our work just as much as our client partners do.

The process goes like this. You make the sample creative, write the brief for a research partner, brief them to get them up to speed, give them a handsome sum to run the test, and the test takes a couple of weeks. Then another week or so before the results and findings are ready to go.

This process is fine if you have the time and budget. But today our world moves much faster, and sometimes this process can end up taking up more time and budget than you can afford to lose.

Today, AI-powered creative testing platforms like Link AI for TV and Digital enable you to test your campaign or digital creative quickly, and at a fraction of the cost. Within a day you can get a read on creative performance, behavioral metrics, and brand lift metrics in an easy to read dashboard. And you can run several different tests at the same time. If you don’t have a big budget, or an extended timeline, this could be something to look into for your team.

Bring briefs to life

Strategists are not typically known for their visual design skills, and we all know there are times when an internal-facing Strategy deck could use a bit of color. Mostly I’m talking about briefings or kick offs where creating some excitement always helps. Well, the next time you’re writing a brief and want to show a bit more than tell, check out Midjourney Or DALL·E 2 will work as well. If you haven’t played with either yet, these generative AI tools can create shockingly good custom images that come straight from your imagination, if given the right text prompts. Give it a whirl. Maybe it’ll get some creative juices flowing.

A few watch outs

All in all, the net net of AI-assisted tasks has been very positive so far. We’ve found so many ways to make our work lives easier. But, there are a few things to be aware of when deciding how to use these tools.

• Avoid intellectual property infringement. AI tools like ChatGPT are trained, in part, on information that may be copyrighted, trademarked, or patented. Keep this in mind when considering how to use the information you get from these tools.
• Vet what you get from AI. By and large, what you get from each AI tool is going to be useful, or helpful. But, certain AI models can occasionally give you different answers to the same prompts, and sometimes they can provide responses that contain inaccuracies. So it’s a good idea to vet the sources and information. As we mentioned above, use this information as a strong starting point, not the end.
• Copy is not original, really. While some might see the value in the ability of generative AI to write prose, we have to remember that it is only generating output from things that have been written before, and not originating thought or expression. So if you need original writing, you’ll need a human involved.

Final thoughts

• This is only the beginning. Applications are evolving quickly, with new versions being released weekly or monthly, if not daily. So this post will probably be slightly out of date by the time it gets edited and published. But that’s OK. The point is to just get started.
• Just get started. If you haven’t already begun playing around with some of these AI-powered tools, don’t be afraid, just try it out. But try not to get lost in the sea of tools that are out there. Focus on one, or a few, reputable tools and see where that gets you.
• Garbage in, garbage out. Get good at asking the right questions in the right format for each tool you’re using. Bad prompts usually mean less than amazing outputs. So, do a bit of digging about good prompts. One tip here is to be clear, concise, and specific in what you need. And don’t be afraid to ask ChatGPT for a revision. Ask it to cut down the length. Or speak in a more friendly tone. Or structure the text the way you want.

More to come…

You may be thinking to yourself, “he skipped a lot there.” Well, that’s by design! Remember, I have some very talented colleagues in other departments who will be sharing their thoughts on AI for Design, Tech, Production, UX, and more. Watch this space, we’ve got more coming very soon.