Insights Everywhere #007: Piers Fawkes

AI-Supercharged Trends & Insights

How to best use AI to identify unique insights and trends for your business

‍

Summary

In this insightful conversation, Piers Fawkes, founder of PSFK, shares his journey from a writer to a trend analyst and discusses the evolution of gathering marketing insights and trends over the past 20 years. He emphasizes the role of AI in supporting trend analysis, but the importance of human input in generating actionable insights. The discussion covers how best to leverage these emerging automation tools (with some specifics mentioned) to enhance and speed up the process of trend analysis and insight generation. The discussion also touches on the changing landscape of marketing agencies and the impact of AI on consulting, culminating in a look ahead to the hopes and fears for 2025.

‍

Takeaways

• Previously, trend reporting was quite manual (gathering of ideas, identifying patterns, sourcing data to back it up).
• The issue with future trend reporting is, there isn’t a lot of data to back it up as it’s an emerging concept to take hold at a future point. Therefore it requires intuition and leaps of faith which is very different to traditional market research.
• These days, AI can speed up the trend analysis processes significantly.
• However, AI is probably giving everyone the same, generic answers, yet with trend reporting, a lot of the value comes with providing thought-provoking, unique ideas.
• Therefore, Piers still believes in humans guiding a lot of the inputs and finessing a lot of the outputs of AI.
• You can train AI by feeding back your human-filtered results so it knows what to look for over time.
• AI helps to write good first drafts, but the output currently needs to be edited.
• In the future, humans could be seen as the orchestrators, and the AI the agents to execute.
• Insights must be actionable and unique to be valuable.
• The role of the CEO is increasingly intertwined with marketing, therefore it would benefit agencies to be in touch with the CEO (versus just the CMO).
• The future of work will see a shift in middle management due to AI.

‍

View more episodes and subscribe to our series here.

‍

Hi, welcome to the latest edition of Insights Everywhere. I'm David, the founder and CEO of Studio Everywhere. And with me today is a person whose work I've been a big fan of for about 20 years or so. He's Piers Fawkes, founder of long running trends and insights agency, PSFK. He's based in New York, and we're gonna be talking about the future of marketing insights. Piers, it's great to you with us. Please introduce yourself. Maybe we can start with you telling us your story. Who are you, Piers? How did you get here?

Hey, David, It's great to be with you. I am originally from the UK and about 20 years ago I moved to New York and I moved to New York to be a writer. And when I came to New York, like anybody who goes to a new city, you're always  fascinated with all the new things you've never seen before.

And I kind of turned up just a little bit before social media. And so there wasn't a lot of sharing of what's happening in different cities and different places. And I turned up here and I started documenting all the things I was seeing, the stores, the fashion, the technology that was happening 20 years ago. And started sharing that on a humble blog.

Back then that was social media. That's what we had back then. And my friends around the world, whether they were in Sydney, Australia, whether they were in Paris, they would share what they were seeing as well. And we put this all into one sort of website, one blog called PSFK. And we were just going to share in all these new ideas that we saw going around the world. And it was a time when this wasn't happening. There was no sharing of ideas. We basically had mainstream media. We had some cool magazines that were talking about some things. You either had the fashion magazines or you had things like Wired magazine or Fast Company. But you didn't really have the stuff that the rest of us were looking at.

There was an amazing time between 2003, 2005, 2006 where we had this democratization of media and all of us were just kind of coming in and sharing. So some of the big websites of today, TechCrunch and things like that all kind of came from that moment when new voices came. And so I came with sharing ideas about new ideas and it was just a lot of fun at first. I was still trying to be a writer, still trying to have a project, still trying to find work in New York city when somebody showed me a trend report.

I got sent a trend report by a company from Anheuser-Busch. And Anheuser-Busch said, you know, the Budweiser people they said, "Can you do one of these?" And I'd never seen a trend report before. I didn't even know. I guess I knew what a trend was, but I never saw a trend report. I didn't know people made trend reports back then. And so I looked at it and I thought, it's just a paper version of my blog. I could do this. And that's how it started. I realized that I can make money, and the folks around me can make money by telling folks what what we saw out there, what the patterns were, what the themes were. It's really interesting.

Who benefited from this in terms of using it in their work?

I think for 20 years, we were half a media company and half a consulting company. And so we had this newsletter, which grew - tens of thousands of people each day were opening the newsletters. We had 2 million visitors a month visiting the websites. And so we had all these creative professionals - people who worked in advertising, often the strategy community, the bigger components such as marketers, brand executives, and then retail executives, reading this content.

We were writing reports, sharing reports, and we were doing conferences and talks and all sorts of things. So those are people who are reading the site and the same people who would hire us to write reports. They are in the same types of communities.

Eventually what happened is on a consulting side, we worked with automotive companies, a lot of like Korean companies, German companies. We work with technology companies. So I worked with all the Silicon Valley companies. Apple was one of our very first clients after Anheuser-Busch. We worked with Apple for five years from post iPhone 1 to iPad.

We worked with retailers, so we worked with Walmart. Target was one of our biggest clients, we worked with them for about 20 years. And then lifestyle brands, we worked with Nike, and a whole bunch of other companies as well. So those are the types of companies that either worked with us or read our content and consumed our content.

Presumably creating trend reports was very manual and crowdsourced curation of what is going on. What do you have to say to this?

Well, in some ways I'm not doing anything very different, but now I use these new tools to help us get there faster. So in the old days, to create a trend report, we'd have to gather lots and lots of ideas that we see around the world, do pattern recognition, which is really the manual matching up of ideas into themes. Then find data to back it up, go talk to experts to back up these themes, which eventually baked into trends. And then we'd have to write the report out. So there's a whole bunch of stages there that could be improved or sped up by automation and AI. And so the last couple of years, I've been exploring how to use these tools. We call them AI, but it includes data retrieval, automation and other things. How can we speed the process so we can get to insight faster?

Can quantify it as well, these trends these days based on the velocity of information?

Because you can use the machine to do the boring stuff, you can spend more time validating the trends. So that is for us, our methodology or grounded theory analysis, we identify weak signals, and then we identify patterns. So we can find more data to either validate what we're seeing, or we can question what we're seeing. We can go and find consumer data and consumer insights and stuff like that. The challenge with trends is, especially future forward trends is there isn't a lot of data.

If you're really trying to think about the future, there's not a lot of quantifiable data that proves a trend, it's a future point. So you got to leap often. You got to take some insights and you to leap with it, which is unlike classic market research.

AI is very good at doing all the grunt work, which I agree. As a bunch of strategists at Studio Everywhere, we use it every day. We're always experimenting new ways to try and use it as a sparring partner or a co-pilot or even a copywriter with various degrees of success. Sometimes it's amazing, sometimes it's very generic, but always useful. Makes you think differently to if you're talking to yourself or even talking to another human. But the question is, at what point does it transcend the mundane and the grunt work and start to think on higher order levels that might compete with human intelligence? It seems to be trending that way. And how would you define what an insight is?

To that point I still believe in human in human out, in terms of this process with AI. So today, a lot of the insight gathering is still human led in terms of finding new ideas. And at the end, reviewed by a human to really reflect on this and actually steer the final output. So we might have a tool that can create a trend report in 20 minutes now, rather than 20 weeks. But we've all used these AI tools and it comes up with some smart stuff. But it comes up with the same ideas for everybody. And so at some point, if you're a company out there, how are you trying to create differentiation in the marketplace when the machine is giving you the same insights? I think sometimes that's why you need to put some human powered data into the system to help get to better and more differentiated insights. And so, for me, like the insight is that, I'm sure there's somebody with a better definition for insight, but it's really just an idea that you can action. It's kind of the way I think about it. It's something that's unique, that's actionable.

But to give you an example, I've been working on this future of automotive reports. And even though I have a system that generates automated reports right now called Trend God, I decided what I wanted to do is go back a step and do a little bit more of a manual process because I noticed this issue, which is when the automated system churns out content, it comes up with some generic points of view. And I wanted to break down the process a little bit and put in human-found data and then stop the process halfway through and look at what it was finding.

And then what I did is actually I did the pattern recognition myself rather than ask the machine to do it. And then I told the machine, these are the patterns I saw. Do you agree with me? The problem with AI is that typically when you ask AI if you agree, they're like, yeah, that's great. But I did the pattern recognition and then I can use the machine to fill it out.

So I told AI - I see this angle, I see this theme, help me write it out. Here are the examples to back it up, write it out, and then towards the end, then I was like, okay, now I want to see some greater insights. What's happening on a macro level in the marketplace. And I think again, the machine is okay at that. But again, what's going to make something different. What's going to be stand out to make you think about, how do we be something very different?

So within the auto market, if you ask AI about what the trends are, it's going to be electrification, it's going to be like urban renewal, it's going to be autonomous cars. And it misses the point that actually, in many markets, autonomous cars struggle.

It's probably a longer term vision for most markets. And in fact, maybe that's not something we spend a lot of time talking about. Unless we're talking about certain markets - I know various parts of Hong Kong, China, mainland China are taking advantage of it.

And then, you know, one of the other things that went through, which I had to swap manually was data usage. So one of the big things which is a worrying concern for consumers particularly is the use of personal data by connected cars and connected experiences. We have this issue with our phones today. What's happening with our data? We have companies like Apple protecting our data. We have governments worried about what certain apps and who's using the data. And it's the same, it's gonna happen to cars. Like, where's this data going? How long are you going to hold this data? Who else is having this data? Are you resetting this data? And I think actually there's a big opportunity, I think, for a car company to actually play the data safe card actually.

Cars these days are riddled with cameras on the inside and out. So with Teslas you can see on screen what's coming around the corner. It's amazing. Privacy could be an issue.

It is a big issue. It's a big issue, it needs to be talked about and we will resolve it. I'm a big believer that people, humans with technology will resolve these things. But if you can remember the Mark Zuckerberg with the sticker over his camera on his laptop, that's not too long ago. Wealthy people I know in New York have wired speakers instead of Bluetooth speakers because of hacking issues. There's real background concern about data issues. I don't think the mainstream worry about it. I think mainstream hope that car companies and technology companies are gonna deal with it and look after their data.

I know you're thinking about cars a lot at the moment because you kindly sent me the draft of your new report into the future of automotive. Tell us about how you put that together and what's in it.

Now with these tools, it's much easier just to gather data. In the old days, we would have to bookmark a million different sources. When we have a new project, what happens now is every day when we see something of interest, we are adding those pieces of data into Airtables or Notion tables and things like that. We also have humans bookmarking. We also have AI trained machines that are bookmarking content as well. Particularly when it comes more like to stats and quotes and interviews with executives.

So we have all this data. So what makes things different is when we create this report is we already have the data pool and we can just get started and do that pattern recognition straight away. In the old days, it would take us four weeks to gather data and then we would do pattern recognition. We don't really have that issue so much anymore. And then, as I said, we went through and we analyzed the data and we then came up with these ideas and then towards the end, we brainstorm ideas about where you could take these ideas. And again, we use humans to come up with these brainstorm ideas rather than the machine.

We have used ChatGPT and the other AI tools to brainstorm for us and come up with ideas. What's great about these machines is that they can build, they iterate, they look at ideas and add things, but they're not very good at being contradictory and sometimes coming up with stupid ideas, you know? Like, why don't I blend these two crazy ideas together to make a new idea? The machine is not brilliant at that in some ways - it's rather conservative.

And I think humans working in the workplace might do the same thing, might just go, “I'm going to play it safe.” But what we wanted to do is begin to go, “Well, let's just think a bit more crazy a bit more wilder here,” because we want to have ideas. The whole point of my work is to get a reaction and get people thinking and I don't want people to be nodding along by the end of my report.

You covered the main themes - electrification, autonomy, privacy - but then at the end there are some great ideas and observations about the more consumer things going on.

I've been thinking for years that these electric vehicles are almost smartphones on wheels and it was amusing to see you flagging up some of the tactics that appeared in the phone industry years ago like Nokia with their snap-on covers, you now have electric cars with snap-on interchangeable panels for fashion, changing the look of your car. You have cross-industry collaborations like Porsche and Soho House and you had several more in there as well. It's super cool. Things that you don't tend to see in the old car so much.

Well, I think the old car world has some needs. And so one of the ideas we have is around customization. The traditional car world wants to make money from your ownership after purchase. And that's typically through servicing of the car. And so they're always thinking, "How do we make more money? How do get people to come in for the service? Then how do we make money during that service?”

So one of the ideas in the report is around customization as a service. So in addition to checking the engine, how do you update the car with the latest features, the latest paint job or whatever it is? It's like a bit of a net-a-porter when you come into the service station, you can get a bit of a refresh.

Yeah, it's a fantastic idea. But, I think you have to think about little ideas like that rather than what's mainstream. ChatGPT is just going to continue to tell you that autonomous cars is going to be the future. And in many markets, that's a long way off. So we need to think about what's actionable. If I talked about actionable ideas, that's kind of one of the big drivers that  is behind my work.

Great, that makes sense. Shifting back to AI for a second, the new edition of The Exponential View by Azeem Azhar dropped into my inbox this morning. I was reading it, he's covering five themes going on in the world at the moment. Heading towards an AI society, energy as technology, geofragmentation, social vibe shift.

But he said a really interesting, phrase which I found very thoughtful, it's like "AI is the electricity of thought." which was a thought-provoking idea because it flows from app to app.

I've been playing with this tool called Hunch this week where you work in a no code environment, like a canvas such as Figma, and you can put widgets and link them all up together. Each widget uses a different LLM so you can do multi-stage thought flows and watch how things flow from one stage to the next. I was using it to tweak and model a naming process. You can run through all the different thinking processes that a creative person would do and combine them and see what compound names and ideas come out at the end. Then you can rank them and weight them accordingly based on feasibility, viability, it's a real time saver. So there are some quite powerful things.And I reckon the future of what we do in consulting is master these but with a human output because it still needs judgement and the answers can be quite generic.

Well, what's interesting there is so you've basically created an agent to do that process. And the latest term is agentic AI. But basically you created an agent. And though you can have multiple agents that do slightly different things within a task, one thing as you talk about that is, you said at the end, then you rank the ideas.

Once you've ranked the ideas you can feed that back into the system, basically fine tune the system and add that at the end. So it starts ranking itself. What I find with my work is, as a human, I'm judging the work and then I judge it enough times where I basically have a spreadsheet, which I can then put back into the system to fine tune the results.

And so that eventually the AI grows, the agent, the chain grows because I keep adding tasks. I keep teaching new things for the machine to do.

I've learnt with ChatGPT especially is it gets to know you, what you do, who you are, and how you write. It becomes ever closer to you as a human being or a personality, which is pretty interesting. But it makes the answers more in tune with what one is looking for.

Yeah, I love that phrase electricity of thought, it's great. You know, about the technology and how it becomes more ephemeral and, just how we've moved from hardware to software to this idea of energy flowing around.

Nice, isn't it? I guess that makes us the electricians or something. What's our role? Orchestration of concepts. I think taste and judgments as well. How to filter stuff it spits out? What do you, of that, polish up and take to the client?

Yeah, that is the challenge. What do we take to clients, especially when the client is doing some of their own work as well. So you know, if you're doing a naming exercise. Maybe the clients have done a lazy version of that as well. So what now? What do you do now?

Yes, AI definitely does a good first draft - a vomit draft, as the guy who taught writing called it.

Yeah. Then you think, okay, I can get to this part quicker, but now what do we do? So if you get to the insight, you get this idea. It still has to manifest in the marketplace, whether it's a brand name, brand image, advertising, website or something like that.

And it's going to get you to that place quicker. So then you can work on that and actually make those things richer and differentiated.

I appreciate it for filling in blind spots as well, like things you might not have noticed. And tailoring content to different audiences. If you need an executive summary, it'll tell you what needs to go on that slide in terms of functional content areas, not the answers. So you can really use it to tune your delivery.

I'm benefiting it from other ways as well. There's this account on Twitter called Mind Branches, which uses AI to decode whole books into a mind map so that you can read a book in five minutes based on the logic tree. A super good way to read business books. Do you have any favorite tools? What's in your toolbox?

Well, over the weekend, I wrote down all the tools that charge me. And I realized I spend a lot of money each month on AI tools. And some of them are cheap, you know, like Canva, which I even think is an AI tool. And then there are some things which are super expensive.

But I have about 30 tools - I guess I wouldn't call it tools. It's a stack that I use, from Softr, which is more like a publishing platform to Airtable to Fireflies for transcription, in terms of my automation, use Zapier. But I definitely would look at your tool as well.

There's a gazillion services that I'm using. But I think what's going to happen is some of these LLMs, whichever your favorite LLM is, it's going to do a lot of this work. And I think actually what's going to happen is you're going to have the big LLM. And then I think Zapier or Hunch is going to be the in-between place. Where we need help, is how you take some insight from the machine and then pass it to another place as well.

So I see great opportunity. That's a great place to invest again.

Wiring the thought electricity grid, I guess, across different apps and environments. Definitely.

Yeah, because we hear these great examples of the AI calling the service center then the AI slacked somebody. But what's happening within that system is there's an automation happening, not just the AI. And so I recommend everybody to just learn about automation as much as learn about AI. Because as you said, it the cables together in terms of the electrical grid.

Definitely. I think one thing that is also changing is your classic customer base, agencies. Everything's being transformed at the moment. There are big mergers going on, Interpublic and Omnicom.

There's a lot of talk of the AI-powered agencies. I've seen some interesting examples. What do you see as the future?

I think maybe I'll just talk a little bit about what the challenges are with the clients. And then maybe we can work backwards towards agencies. I think this seems to be a post pandemic, a lack of interest, especially in the Western sector, with North America, Europe, executives have taken their foot off the brakes. A lot of people are doing their nine-to-five.

And they are not bothered or not as interested about their business as they used to be. I have a number of products like competitor tracker and things like that. And with broad minds, I realized that people don't care what their competition is doing, which sounds illogical, but for most people, what the competition is doing, that does not impact their day-to-day work directly.

I used to feel that everybody was obsessed with what the competition is doing, what innovation is doing. And I feel on the client side, there is a very small amount of people who are really interested in the future of their business. They're probably compensated in terms of the future of their business as well. Yeah. But for a lot of other executives, I don't know a lot of people care.

So if you basically have an audience that don't care, what does that mean for agencies and how they support them? So you have agencies that just need to churn work and just keep doing work and just do the work that the clients need done. And then you have a small group of agencies that can service that small clientele who want to do brilliant work who really want to drive the business forward. And the problem with that one is you basically also have the consultancies having the conversation with the CEO and the CEO today is really the CMO.

It often it feels like the CMO is a marketing director, an activator of campaigns and a coordinator of stuff. But it feels more and more these days by tracking earnings calls and things like that, that who's really in charge with customer communication, customer experience is the CEO. And I don't think agencies have that conversation, or are connected with the CEO, so there's a danger there. So in some ways I can understand why you have this consolidation. So that you can go up against the consultancies. So Ogilvy can go up against the consultancies. Maybe that's one of the reasons. So it's a roundabout way to kind of get to a point.

Last question. You're a futurist of sorts, let's think about the year ahead, it's January 2025. What about one hope, one fear, one reality for the year? Where do you see things going?

I mean, one hope is that we all just get back to work. That's how it feels from where I am. There seems to be a lot of, slowness here, and it would be good for us to all get back and get energized about the future and start building the future. There's been a lot of short-termism and I think we need to start thinking about where our business is going to be in five years time. And so that's my hope.

The fear is because we have this lethargy and this slowness within corporations, the machine, the AI will replace a lot of middle management. And so, as you know, a lot of the work can be done by machines. So the fear is a lot of radical change within middle management. And so we see these organizations that are traditionally pyramids that gets flattened. So basically, as I said, the CEO is the CMO now, and has a far reduced middle management, and then you basically have doers who are doing the work in some ways. And so that, just from a societal point of view, that would be pretty impactful. Yeah, so that's a bit dark.

The reality is I think people don't take AI seriously right now. And I think that will continue for a while until we see what's going to happen. It's a bit like the DTC movements or the Airbnb. We won't really be taking AI that seriously until like the Airbnb turns up of the travel industry for this new generation.

I think when we see these new generation of companies that are going to replace the old companies, the legacy companies. We're going to see some of that, I think, this year. They might not replace them this year, but we're going to start to see the Airbnbs turn up and we're going to see history replay itself.

That should be interesting. Do you see any signs of early breakouts on that front?

We see a lot of the little startups, but I don't know if I see a scale up right now come through. But it's definitely something to watch. I think all these startups we're watching, somebody's going to come along and just solve a problem that could have been solved before. And what happened before with the Airbnb scenario is everybody saw Airbnb come along, but they thought it was gonna go away. And now it owns 20% of the market.

And that's going to happen again. So it's going to be interesting. And even some of those old digital companies are going to be replaced by these new companies. AI first companies are gonna be interesting to watch.

We're two weeks into the year already it's a wild ride so I look forward to see what's next.

‍