Natalie Kelly [01:38]
Today, Steve and I are digging into the third annual Connected Insights Imperative report. Based on research with more than 250 marketing and insights professionals, we're going to explore why the expectations around insights are rising, what separates AI adoption from true AI acceleration, where synthetic data fits, and what organizations need to do next to turn connected consumer understanding into better decisions that ultimately power better growth. Steve, it's always a pleasure to get the two of us back together. And I have to say this year's findings give us plenty to talk about. So let's start with the one that immediately jumped out to me, and I'm betting it jumped out to you. Connectedness has increased significantly from 38% to 48%, but overall satisfaction with insights has fallen from 60% to 48%. How do you explain that apparent contradiction? Is satisfaction with insights falling because insights are getting worse or because expectations for what insights should deliver have actually changed?
Steve Phillips [02:46]
Great to be back with you, Natalie. I should say that to begin with, these are always fun. I think it's a fascinating finding. I think great kudos to the insight industry. We are working as, you know, with a lot of clients, as are many other people, to try and connect the dots in the data that we provide. And it's only in that connectedness that our clients can really get the maximum value from the data that they've already bought, paid for, and connect that data to create insights on top. But connectedness was always a platform to me. It was never the end state. It was the platform on which you can build the greater understanding, the more holistic understanding of the consumer and use that to empower better decision making. So whilst I think what the survey is showing, we have made progress on the core of it, we've made progress on the platform, but maybe the progress we've made on top of that platform on providing those insights, not just getting to the insights, but getting the insights themselves into the hands of the people who need them when they are making a decision. And I think that's the next step we've got to take is great. We've done the core work, now let's build from that. And as you know, AI is changing so quickly, everything now is changing so quickly that we don't have time to rest on our laurels and say, yeah, hey, we've managed to connect some insights. That's. That's job number one. That's job number one. Now we need to prove the value of that in a very rapidly changing world.
Natalie Kelly [04:23]
Yes. And, you know, I think you point out something really great there, Steve, that it seems like the pressure to get more value from AI has increased because people believe in it now, but they've had some experience and it's like everybody can kind of sniff this opportunity out, but they don't know how to derive the value. So it's like the frustration with insights to you teams, you know, and I think a lot of that results in like, reorgs and things that are happening in customer, you know, accounts that we speak to all the time, but in the market at large, because there's this frustration like, oh, we've got these great tools, we want to unlock the power of them. We know the insights team has something to do with this to give us really great insight so we can make the best decisions to power business growth. But the secret is it's complex, you know, there's org changes, there's, you know, data connectivity, there's all a lot of pieces in there. So I think that is my read on it, just knowing what the market is saying.
Steve Phillips [05:22]
Yeah, I think you're right. And also there's an alternative which sounds really good. So if a marketing team wants to understand, you know, what their consumers are thinking, what the Trends are we both know they can go to Claude or ChatGPT and get an answer to that question. They can get an answer to that question in two minutes. Now, my guess is that that answer to that question is nowhere near as robust as if you went through the data that we've connected, right? So actually the result, you would get, the insight you would get, the ability to make a smart decision would be significantly better if you went through the insight process. But because it's available instantly, sometimes you're like, oh, well, it's over here. It's instant. Now what we've got to do is make the great data available instantly rather than a complex process of, oh, I'll talk to my insights Aidan, who'll talk to the person who might have read the report, who talks to someone else. So we've got to get that brilliant insight that we've got and embed it into our client systems embedded into, you know, the software platforms that they're using. Whether it's Salesforce for the sales team or Adobe for the creative teams. We've got to make sure that the brilliant insight we've got, which is significantly better than just going to an LLM. We know that there are various studies where I've been reading several this this week, there was a meta study talking about how much more important it is to get genuine human insight into the system. And besides that, that's, that that's the competitive advantage your organization has. Every organization can go to ChatGPT. Not everyone. Every organization can go to our client's data. So we know where, we know where the value resides. And what we've got to do is get it into the systems where the decisions are being made as quickly as possible and close to pretty much instantaneously.
Natalie Kelly [07:22]
Yes. And it's so important, you know, it's interesting because I think before people were thinking in the vein of like, oh, maybe with AI and synthetic, like, I can replace humans or I can get this thing to done faster. Now it's about where do humans matter the most? And what you said really resonates with me. It's the speed and the quality, but the depth of the quality. Like as a marketer, I can get a signal, like you said, from asking ChatGPT or Claude, and that's going to get me so far. But then I'm like, well, but my growth segment is actually millennials in this age group. And this, this is, you know, this profile. What about them? That's really because marketers want to target with precision that I'm Never going to get, unless I import my data or can connect my data and get the, that data embedded into these other places where these decisions are made. That is the heart of what I think the frustration is that's reflected in our survey data on this topic. Because people like, I know there's a there there, I just can't connect all the dots. I just, you know, and so we're helping them do that on the zappy side. But I know every single customer is facing this exact challenge and I feel it as a CMO as well. I want, I want all my customer data everywhere, all the time.
Steve Phillips [08:35]
Yes, and that's absolutely right. And, and, and you know that, that customer data is coming from different sources, right? Some is coming from, in our case, you know, salesforce, some is coming from Gong, some is coming from Notion, some is coming from Slack, blah, blah, blah. So they, it's about aligning, it's about going back to what we were talking about, connecting that data. But then that data has to live and breathe where you live and breathe, breathe. And those are the changes, the architectural changes we're making with our clients. And I think their end clients, the people who are making decisions in these corporations, are frustrated because it's not there. And they think everything should be instantaneous. And to be honest, they're right. The setup process takes a bit of time, but then it has to be instantaneous. It has to be with them in the same way that other tools are with them. But better. Better. Our data is better. We know, we know from study after study after study that good quality human insight is much better for making smart strategic decisions. Is there a world for AI, Just AI? Is there a world for AI trained on other data sets? Yes, absolutely. Yes, absolutely. There is that world. But if you want, if you genuinely want a competitive advantage, you have to bet it on real human insight.
Natalie Kelly [09:56]
Yes, absolutely. Your data is your moat and your competitive advantage. And leveraging it with AI and putting it into more places where you're making those decisions is how you will win with consumers.
Steve Phillips [10:10]
Exactly.
Natalie Kelly [10:10]
Yeah, exactly. Yeah. Well, we've solved all those problems.
Steve Phillips [10:15]
Okay, Natalie, I've got a question for you. So there's another tension in the data. 46% of brand manager respondents say consumers are brought in before creative development. But only 15% of insight and consumer research professionals say the same. Now, why do you think marketing and Insights have such different views of when consumer understanding is actually entering the decisions? Could marketers be drawing on signals outside the traditional insight function?
Natalie Kelly [10:45]
Oh, absolutely. I mean, the consumer insights Teams generally are limited in terms of where and how they are used. So the marketers are looking across many, many more signals, a much broader set of tools and signals across everything. And they aren't often connected, which is, in my opinion, unfortunate because insights people have a superpower. You know, market researchers look at data in a very careful, specific way and marketers are looking, you know, they want that, but they also need to look at the broad trends and what's happening in the business. And if they even could talk to each other more about that, the power that would be unlocked would be huge. Because often the marketers are looking at social, you know, social data and, you know, maybe click data and things like that that they don't feed to their insights teams and share with them. They don't often have deep conversations about that insights people might hear snippets of it, but they usually go to them and say, I want you to do this type of study, I need to know about this. And they give them the research question. But they, the insights people are suffering from the same thing that we just talked about that AI is suffering from. Lack of context, lack of nuance, lack of access to the data. The more you can give that access to the insights team to leverage that data, the better. And guess what? They could use it to inform a lot of their studies and map it. It's just that the human bridge is not there. And why isn't that human bridge there? Lack of time, I think. In every company it's the same. The people around the company have all the right information. They just can't share it fast enough with each other to make a darn difference, you know, and that's the hard and frustrating part. And that's where AI could help us so much.
Steve Phillips [12:34]
Yeah, I get frustrate, get frustrated by this. This has been a bugbear of mine throughout my career. And insights, and you said it, you encapsulated it. So the marketing person is listening to different forms of data and then asks the insight person for some data. My view is that we're too reactive. So what I want insights teams to be is proactive. You don't need to wait for the ask. You can deliver, you can embed, you can stream insights into those decision makers so that sure, they might be looking at social, but that social is tied to the human survey data that we've got and we've proactively made those two align and tell a similar story and enable people to get that signal really easily. So we have to stop reacting and waiting to be asked for an answer, we have to proactively thinking about how do we deliver that answer into the hands of the decision maker at the moment that they need it in order to make decisions. So I think, again, I think the connected insights piece, the platform, we talked about that the platform is great, the platform is great, but that is not the answer. The answer is what you can do once the platform's in place. And that requires us to have much more strategic conversations with people. Not at the moment they need an answer to a question, but when they're thinking about systems, we have to think of ourselves much more like an architect than the project manager and architect systems to enable data to stream into their hands when they need it.
Natalie Kelly [14:14]
Yes. I mean, when you think about it, if we had no platform and we had no tech, how would we get those insights to each other to connect the insights? It would happen in a room where people are talking about what they learned from watching people live on the street or whatever. But we're in a different world now. We have online digital multiple channel, you know, channel proliferation. We need the digital signals as well as the offline signals. And guess what? The online signals and the digital signals are often more indicative in many ways because people will tell you quickly and in real time. You get that. I'm not discounting the value of qualitative research, but I do believe that all qualitative research is also data that can be fed into a system where it can be action with AI and technology to enable better delivery of those insights to the teams who need them in the decision place that matter.
Steve Phillips [15:08]
Absolutely. And I'm all for, and I'm all for. And so if you think about the data that is coming in, you know, there's, there's social media data, there's clickstream data, there's survey data, there's qualitative data, there's a whole range of data. All that is wonderful. Now I increasingly, and we often go onto this topic, what I see the role of the insight person doing is it doesn't matter what type of data it is. What matters is what the data tells us, how robust that data is, how good quality is, does it cover all of our audience. So, for instance, social media data I think is really powerful, but I think it's powerful for trends. So I wouldn't use social media data to understand what my brand position was. Maybe I'd think about it for where my brand is going. I wouldn't use it to design an advertising campaign. I might use it for innovation of a New flavor variant, you know, where I want to capture a market. So we as inside people are actually pretty good at understanding the power, the potential use case of each of these streams of data when that's the point of connecting them. So we have to connect them so that we can use the right type of data in the right type of situation and to make the right decision. You know, that's what the system is there for. So I'm not worried about these different types of data. I'm excited by them and excited about the opportunity of putting them together.
Natalie Kelly [16:31]
Me too. And bridging the offline and the online is the thing I really, really care about and talk about every day to my own team. I think having the sources of data from various different touch points are part of the journey and understanding that as a marketer is essential, which is why embedding that consumer data everywhere you can is going to be so important. You know, when I think about old, older models that existed, my limitation as a marketer would be, oh, I want to give my Consumer Insights team access to my social tools so they can log in and see the same reports that I see. But reality is you can't afford enough licenses so you're not going to be able to pay for that out of your budget.
Steve Phillips [17:10]
Yeah, yeah.
Natalie Kelly [17:11]
So there's blockers that are literally financial blockers. But if we can get the data into places where people already live, into the systems they already use, that is how the true power of Consumer Insights will be unlocked, in my opinion.
Steve Phillips [17:25]
And I hate to get boring and technical, but MCP and agents is the way forward we have. That's, that's, that's the way it works. That's how we do it. That's how we're going to solve this problem.
Nataly Kelly [17:35]
Absolutely. I am seeing as a marketer the value of being able to query all these different systems where all the data lives that I don't normally have access to, but I need because it's where my customers live and I need to know what those signals are in order to make good decisions for how I use my budget. So yeah, every CMO needs that. And it's not boring and technical. It's easy. MCPs are easy.
Steve Phillips [17:59]
It is absolutely true, absolutely true.
Nataly Kelly [18:01]
It's actually the most human-friendly type of thing. I love it. One other question I have for you related to MCPs and AI and all the great things that we love about all the new tools we have available, but AI could actually widen that gap further. So my question for you is the following: 93% of organizations are already using AI, which is great, but just 8% are doing what we describe as AI accelerated in our report. So most current use cases are still focused on things like survey design or just reporting or summarization or data cleaning, which is work that is not the most impactful but is important work to do along the way. So Steve, what really separates using AI to make the existing insights process faster from actually transforming how consumer understanding gets used inside a business?
Steve Phillips [18:54]
Yeah, I think this talks to a split, which is not about insights, it's about the application of AI. And there are two ways that people are applying AI within their business. The first is efficiency, efficiency and effectiveness. And that's doing the old job better, right? So, and it's a natural place that people go. So I don't think it's a problem, but if you stop there, it's indicative of a lack of imagination, because actually AI also enables you to do different things, not just do what you're doing smarter and cheaper, but to do something else on top of that. And I suspect that, you know, for the first — like the survey says, you know, 80, 90% of applications of AI will actually be about, "Oh, I've got this frustration in this part of the process, let me use AI to make that quicker." Not a problem at all. It'll save some money, make things quicker. Then you've got to get onto creating using AI to create a competitive advantage. So then you've got to be creating new ideas, doing new things. And we are seeing — I think it's really interesting — we are seeing some of our clients, some of the senior people in insight, being on the teams that are doing new things at organizations. And I go back to what are our core skills. Our core skills are understanding and knowing and combining data. We think about this stuff naturally. You know, ever since I started in consumer insight, we've combined qualitative and quantitative data into one report. And I think that skill set is helping us be a part of the change that is not just about efficiency and effectiveness, but doing new things in new ways. And that bit is slower. It is slower because it needs more time, it needs more thought, it needs more imagination. But the great thing is that consumer insights departments are playing a role — at least we're seeing that in some of the pioneering companies — and that's something that really excites me. And there's a whole range of things that we'll be doing, like streaming insights directly into, you know, media trading applications, so that it's not just click data that means media is optimized around, but also brand lift, you know, or brand impact. It might be around ensuring consumer data is automatically included in, you know, in Adobe to ensure that every piece of creative is always on strategy. So it's just using things in different ways in our world. But those things will just take a bit more time than the easy ones, which are, "Ha ha, I can write a first draft of a report really quickly, isn't that great?" And yes, that is great. So both bits are exciting. Obviously, you know, the new thing is probably more exciting to think about than just the simple efficiency-effectiveness thing.
Nataly Kelly [21:54]
Yeah, you know, a lot of the devil's in the details on where in the process you decide to use AI and where you decide the power is for your creative team, or in this case the insights team. You know, and I'm thinking about both, because as you mentioned that, I was thinking, yes, you know, the best humans you want pointed at the hardest problems and the most impactful problems, and you want those high-value humans to do that work more and less of the low-value work that AI could do without you. And so that is kind of where people are naturally focusing more on leveraging AI for those things that are mundane tasks and things that nobody really loves to do but has to do — but the knowledge of how to do those right lives with those humans who train the AI and who give the prompting and who develop the skills. I'm thinking about that because, like, on the creative side as well, it's the same thing. Like, I want my creative team not doing low-value assets that are for every single employee that Zappy has on social media. Like, they can go in and do this themselves with a template that my team created and use AI for the actual creation of those. But where I do increasingly put my creative team now that we're leveraging AI more is in strategic things. Okay, how do we really think through this customer delight moment, for example, at Zappy, to give a Zappy example. And I think it's true of the insights teams as well — now they're getting access to new horizons where they can influence more and be part of the next wave and the next challenges, which is super exciting, because their brainpower and the way they think about data and the knowledge they have of how to read the consumer is so important that every CEO needs them next to them, right, as the right-hand person. Not just the CMO — the CEO needs to understand, and the CMO is often a proxy for that. But the power of the insights voice there is so key.
Steve Phillips [23:54]
Amen to that, Natalie. So, Natalie, there's another finding I think marketers should pay attention to. Connected organizations are much more likely to combine the what consumers do with the why behind that behavior. 47% balanced behavioral and motivational data, compared with only 14% of disconnected organizations. Now, why does that combination matter so much when you're actually making these marketing decisions?
Nataly Kelly [24:22]
Oh, this is such an important topic that every marketer cares about, because if you only look at the what data and not the why data, the what data can mislead you and lead you down a very bad path. Similarly, the why data can too.
Steve Phillips [24:40]
Yeah.
Nataly Kelly [24:40]
Either in isolation can lead to the wrong conclusion, but especially when you're trying to figure out what to do next. You know, we recently did a test of this Callaway "good, good" creator ad that was out there, and I think that's a perfect example of why these things matter so much. Because when we got the AI, we tested it in Amp AI, and we got the results back, and it flagged all the things that were out in the media of why it would be a bad idea and why it would go wrong. But when we tested with humans and we looked at the specific cohorts that matter the most for them, there was a difference between what women felt and what men felt in their target audience. A very big difference. So it actually made men like the brand more and women hate the brand more. Now, if you're a marketer and you're like, "Okay, I have to grow a market share and I need to partner with some creators and I need to influence this segment," which segment am I trying to shift? I have to think surgically about where I want to make that shift. And that creator activation, you could argue on paper, would have shifted that segment in the right direction if you only looked at that data point. But if you didn't think about the why, if you didn't marry the why data and the what data, that is the type of thing that can lead to that outcome. I'm not saying that's happening in that example, because I think there's many things that went wrong in that example. But that's an example of how, if you only look at one of the two, you can be misled. And this is why it's so important for marketers to be involved in thinking about brand risk and connecting that to every single ad that you put out on behalf of your brand, every single activation that you have with a creator, with an influencer, because each of those rolls up to impacting your brand health scores, your brand health metrics. We can do a Zappy brand, you know, health snapshot and see this very quickly, point in time, in addition to continuous trackers that you have to make sure that you're looking at those big things. You can also do a snapshot to figure out, "Oh, did that actually damage our brand in some way? Did this impact this segment in some way?" That's the type of data that every marketer needs. But you can't just look at it in isolation. You need the why and the what. And then you can figure out, "Oh, is it good that people clicked on this ad? Is it good that more men went and bought these golf clubs because they like this?" You know, maybe it drives a short-term sales activation spike that helps you hit your quarterly numbers, but it might in the process hurt your growth segment, which is women — that is your future. And so you're again making a short-term trade-off, you know, a long-term trade-off for the short-term sales spike. So that is really at the heart of a lot of marrying these things together.
Steve Phillips [27:36]
Yeah, that makes perfect sense. And I think, again, going back to my point from before, is the insight people have to be proactive. So the ask may be, "Can I have some what data, or can I have separately some why data?" But we should make sure that we deliver them together. We have to connect these insights, and it's the combination of the two that really creates the insight. So, you know, I've been a qualitative researcher before, and you come up with this great anecdote, but if that anecdote is literally for one person out of a thousand, that's not interesting. But if that anecdote resonates with 250 people out of a thousand, then it becomes really powerful. So that's one of the core jobs of the insight team — pulling that data together.
Nataly Kelly [28:22]
Yes. And validating, right? Because I remember many research studies when I was in a research company, and we always started with the qual, but we had to validate with the quant, and then we go back to the qual and do deeper qual, and then it was like a back and forth, because you have to use one to inform the other. So I think the best of the best is when you've got all of those connected. So that's really exciting. And now we have a question for you, Steve, about another both-and scenario: human and synthetic, qual and quant, why and what. So let's talk about synthetic research for a second, because this is obviously going to be a much bigger part of the conversation at large in the industry. It's a big part of conversations with all of our customers. One really interesting finding from this report is that we found only 7% of organizations are currently using synthetic respondents, and people are much more comfortable using them for early-stage, reversible decisions than for major strategic calls. There's also a pretty big divide between functions — 47% of insights professionals are unwilling to use synthetic for major strategic decisions, compared with only 18% of marketers. So, Steve, where do you think organizations should draw the line on this topic? And is some of that caution from the insights team actually a good thing?
Steve Phillips [29:46]
I go back to the data science behind this to start with. So let's think about the data — how do you train and manage a synthetic respondent, or a synthetic answer, a piece of synthetic data? So imagine you have a very strong data set of people in Paris and what their favorite color is, and you say, "Okay, I've now got that data set, I've trained my synthetic model on that data set about what colour they like," and you say, "It's a Tuesday, what colour are people most likely to like today?" Then you will get a pretty damn good answer from that synthetic model, because it's trained on very good synthetic data. However, if you ask them where they're going on holiday next month, you're very unlikely to get a very good answer. In fact, you'll probably get exactly the same answer you get from ChatGPT. So ChatGPT may be accurate, it may not be accurate, but you haven't done anything in terms of training a model by telling it all about people's favourite colour. If you ask a question which is unrelated to the training data — I think the problem with a lot of synthetic data is that the training data is unrelated to the questions people are asking of it. Now, for that reason, we've decided not to do generic synthetic. What we're doing is synthetic data that is related to the training asset that we have. We can predict advertising results, we can predict innovation results, but there is something else around it as well, which is you've got to think about what that training data is. So that training data is what people said about previous ads they've seen and previous innovations they've experienced. As long as you're creating something that isn't genuinely new, then synthetic is going to work really well. So if it's a minor line extension, if it's a version of an ad that you've already tested and you know is good, not a problem at all. If you're trying to create something new — a new idea, a new piece of creative, a new way of engaging with respondents, a new flavor variant that you haven't seen before — then synthetic, pretty much by definition, won't work, because it's based on past data. So to me, that understanding of how the models work aligns with how and when you use synthetic. So we recommend you use synthetic for low-risk decisions — so a situation where your choice isn't really between using humans or using synthetic, it's between using synthetic and not using anything at all. So if you think about an advertising campaign, you might have the big TV campaign or the big video campaign, you might have a few key pieces of digital, but you've probably got another hundred or two hundred pieces of campaign material. Synthetic can give you a really good understanding of, are they going to land well, are they going to be good for the brand, are they aligned with the rest of your campaign, so that you don't make a really poor decision. But you wouldn't want to test your main pieces of communication with synthetic, because hopefully they're saying something new and creative and exciting, and you've created something that's worthwhile putting out there. In the same way, I would say with innovation — it's great if you've got 50 ideas and you want to cut it down to 20 that you're going to put into proper testing. So once you understand how the models are built, then I think you know when to use them, and also when to use which type of model. So there are different models that are good and accurate at predicting different answers, different questions, different situations. Ours are great for innovation and advertising — they're not great for, you know, general knowledge or preferences for holiday items or whatever it might be. So you need to understand the variety of synthetic models around, and people need to adopt the right synthetic model in the right situation, which is low risk. And I think it's very early. So there's been a lot of chatter about synthetic, and a lot of companies and offers that are sort of generic digital twins, and I can see the attraction of them. It's going back to the conversation we had before, which is, yeah, there's no difference really between a generic digital twin and asking Claude or ChatGPT, and you go, "Well, that's actually quite useful because I like talking to ChatGPT and Claude." You have no competitive advantage — if you're in genuine insight, every other competitor has exactly the same answer. But it may mean that you can move really quickly for something that's very low risk. I'm not against that — you just wouldn't use it for an important decision.
[34:56] Nataly Kelly: Yeah. And you know, it's interesting because it's the importance of the decision, but it's also where it is in the process. I feel like consumer insights professionals are going to become the almost like a medical professional who will diagnose the exact dose of when and how to use synthetic that is appropriate for the, the problem that we're trying to solve or the pain that needs to be. Because I'm thinking, you know, there are times when I currently use AI for forecasting and I think this is a great use. Like in that Callaway Good, good example, you know, we could have, they could have forecasted what would happen, you know, based on just general prompting, general questions with AI and synthetic respondents. But if I were developing the ad and I wanted to understand specifically which part of the ad is the piece that is now AI even called that out. It even called out the specific moment in the ad. But if I were the creative team in charge of updating it, then I would run or want to run by the final finished ad by human. So it's really interesting what you're saying, the stage and the impact of the decision. Decision. But there are various points in the process where the decision really, really matters. But it's not always flagged as like, okay, this is a high risk decision. Sometimes that's just a designer making a call really quickly because it has to get out the door because the super bowl is launching. You know, this ad has to be finalized and handed off. Sometimes it's a timing thing and people don't have the time to consult. And so in that instance, because AI did such a good job flagging the exact scene in that Callaway Good, good, you know, piece of creative. They at least could have said, based on what you know of, of how humans behave in this target audience in this segment, is this likely to be better or are there other problems now with this execution that need to be flagged? And I probably would get a quick response back even if I had limited time, that would be directionally helpful. And that's where the power of the human judgment and the AI informing it would come into play. I think in that example there was clearly a lack of human judgment, multiple phases, but also there wasn't a human working with AI to even get that sense check on the final finish. So, you know, there's multiple places where you could have used AI and human judgment. But I think insights teams will play a more important role there in diagnosing strategically which tool to use when, at where, you know, along the way.
[37:33] Steve Phillips: Yeah, no, I completely agree. And, and I think, I mean, I mentioned the architect thing before. I think we have to play a role architecting the data in the systems that our users are using. And yes, there's a role for ChatGPT and yes, there's a role for a digital twin and yes, there's a role for more sophisticated synthetic. The, the types of stuff we do. And yes, there's a role for human and throughout all of them understanding and delivering the right solution built off. And remember, all the synthetic is built off the human data. So we have to make sure that all of the input to human data is really top quality. So it used to be, you know, we do a survey and then someone would get a report, make a decision and move on and no one would look at that again and that data was thrown away. Now we're using that data over and over again and we're using it to train models. So it's become, the human bit has become way, way more important and way, way more powerful. So the importance that, I mean, you know, every insight leader I talk to is placing more and more importance on the quality of that data because that data is being used over and over and over again. So it is creating effectively, it's creating the algorithm that the rest of the organization works from. So this is the role in many ways I see for the insight teams is managing that architecture, ensuring that data quality and using the architecture to integrate the right models in the right situations, making sure they have the right training data that's up to date, that's well thought through, all of that sort of stuff. And I think it's a really exciting role. I think there's a, there's, there is also, as we know from the satisfaction, because we're not there yet, there's a lot of work to do.
[39:25] Nataly Kelly: Yeah. And you know, as you said, you brought us back to the architect layer, which is AI accelerated level four of the Connected Insights framework. And we've written about this now for years. And what I didn't predict when we were starting to see more AI adoption last year is that AI would be something that would really push people into level four and force them to become architects, I kind of thought, well, if a company is motivated to do that, that will happen naturally and they will lead the charge. What's actually, I think happening is it's forcing it to happen in different areas and including the role of the insights team, which is the most important piece to all of this, in my opinion.
[40:06] Steve Phillips: That takes me on that to the next point. There's a structural challenge behind all of this. Half of organizations still rely on ad hoc research and separate vendors, while only 38% have centralized research in a common system and just 12% have continuous streams of insight. Now what does that operating model mean for marketing's ability to actually use consumer understanding when and where decisions are being made?
[40:32] Nataly Kelly: It means a lack of agility. Marketers need to be agile in order to respond to the change in consumer behavior and consumer trends and where the consumer is headed. And if the data is decentralized and the research is in different systems and it's all over the place and it's not connected through technology, it can't be continuous. Connectedness is what leads to continuous. You can't have continuous consumer insights and respond to consumer change at the level you need as a marketer if you don't have connected insights. That is so important. You know, ad hoc studies really do make it difficult. That throwaway data is just a minefield, that it bothers me a lot to think that we pay money, marketing budget, company budget, to do consumer studies as marketers that then don't get leveraged over time because those could be heating and training your AI and your synthetic respondents so that you can leverage them at more decision points. It's so vital to make sure that that change happens. But the only way it will happen is like we just talked about, through making sure that the consumer insights team is architecting smart systems and enabling the data to connect and the data sources to connect. You know, as a marketer, I want my team, every single member of my team, to be able to action that consumer data and consumer insights at every decision point so that I can make sure and feel confident that I am choosing the right creative, you know, making the right decisions on how much money to put against that ad. How many dollars do I put behind this asset versus that? What are my strategic choices from a product development and innovation perspective? If I don't have the consumer connected and the data informing my decisions, I cannot leverage the continuous change that's happening for competitive advantage. So moving toward continuous is so important. But I don't think we will see companies out there be able to take advantage of continuous insights until they connect. And it's going to be a both and in a gradual process. But I do think connectedness is a leading indicator of ability to leverage continuous insights.
[43:04] Steve Phillips: I completely agree. The only thing I would say is you say it's a gradual process. It's just the pressure seems to be so, so strong at the moment and the way you look at the rapidity of change in so many parts of marketing, but also of course the underlying technology. And I think, you know, the pace we worked five years ago is not the pace we work now. And we have to jump on these trains. And the amazing thing is that AI is incredibly good at helping you become agile if you structure your data in the right way and are aligned with the model. So it's both a scary and exciting time.
[43:47] Nataly Kelly: I agree it's scary, but it's exciting. And I hope people will move past the fear toward action against the most strategic areas. Like you know, when I say gradual, it's like gonna be gradual to connect all of those things. But not all of those things matter equally. So what are the systems that actually are where the most important decisions are made? Where did those decisions happen? By whom are they made? Once we map that consumer insights teams can position and orient how you know, what part of the architecture do they action first? What areas are the most important to really go in and impact? And I will tell you, just having worked in big companies before, sometimes you can't action the things that are the most important. You have to find the places where you have willing collaborators and willing partners. Because humans will always be the gatekeepers to change. So as much as goes back to behavior change, doesn't it like humans are going to be the blockers? Always it's the humans who are the blockers. So where are the things that we can find the best partnerships and alignment and collaborate to prove that this is possible and find our way in and start making traction there. But I would go in order of priority for the business. Okay, so Steve, I have another question for you. We've talked for years about insights teams moving from order takers to strategic partners. And we're seeing progress there, which is great. 42% are now viewed as strategic partners. This is wonderful. But only 11% are viewed as architects of an organization wide insights operating system. So Steve, is strategic partner still the destination? Or does the rise of AI mean the role of insights now needs to evolve again?
[45:31] Steve Phillips: I think it needs to evolve again. The strategic business partner was moving away from just what we talked about before. Just being an order takeout. Oh, marketing comes along and says, go and test this thing. Then you go off and test it. And what to me, being a strategic business partner is about saying to the marketing team before you've even created something, I'm here, I can help. I've got some data on the market. I can ensure, I can be at the kickoff meeting, I can help brief the ad agency, I can be involved in the process. I can advise you when to, when to test, when not to test, what data you need to make a smart decision, et cetera. So that's a great strategic business partner. But the changes that we're talking about, the rapidity, the volume of decisions, the volume of creative, the volume of innovation, means that it's not, you know, a strategic business partner sitting in a meeting room giving you quality advice is really important and really useful, but no longer enough. And actually, increasingly we need our insight not delivered by a human, but embedded into a system. And that to me goes back to the architect idea. It's much more. Yes, we need, and I suspect the insights department will fragment a bit with some people being much more strategic insight professionals sitting in strategic meetings and some people managing data flows and ensuring that data flow is hitting the right parts of the organization at the right time, which may enable an agent to make a decision or a human to make a decision. It almost doesn't matter. What matters is that there's great quality insight to help either the agent or the human. And then of course, there's the macro role of saying, okay, where do I need that strategy? Where do I need the embedded insight? How do I ensure the whole system is based off really great quality insight? And that insight is up to date and in the right systems in the right way. So I think it's a necessary but not sufficient step for organizations to make. Now, it's a bit like the move to Connected Insights. That's a great platform, but that's no longer the destination. That's part of the answer, not the full answer. Okay. And that brings us probably to the biggest new idea in this year's report, the Connected Insights Flywheel. We've talked a lot about connecting insights, but the flywheel is about what happens to the learning next. Now, how does connecting learning across innovation, advertising and brand actually change the way an organization makes decisions?
[48:13] Nataly Kelly: Well, this is a very big topic and I think a lot of very smart people have worked on this exact topic for, for many, many years. But what I love about the Connected Insights Flywheel is that it helps us visualize some of the things that we know are happening and how technology can play a role in solving these challenges. So innovation learning will inform what you should bring to market. Advertising learning will inform how to connect it with consumers. And brand learning shows what happens in market. So having those three buckets of insights and weaving them together and connecting these learning loops really enables business leaders to leverage what the organization already knows. We talked about throwaway data and ad hoc research and all that money that you pay that doesn't go anywhere and not feeding into a system that is powering the way that you can make decisions that with quality data and quantity of data, the real opportunity is to shift from just connecting the data to connecting learning and enabling a business to really leverage a lot of the power of all of that consumer insight to make better decisions that compound rather than individual studies starting from scratch over and over, which is wasteful and doesn't actually give you the leverage that you need. You know, and when I say leverage, I literally mean budget leverage. Budget leverage. You know, if, if somebody in that organization has already paid for a study with that exact group of people and there is information in there that could be leveraged for a different team, why would we not use it? Marketing budgets are precious and finite and we have to prove the, the ROI and dollar of every dollar that we spend. And yet all of that data we're paying for just goes into an abyss and we never leverage it again. It's a crime.
[50:26] Steve Phillips: I'm with you. I'm with you. I'll get geeky for a minute. So we have to remember as insight people that that data is made up of multiple things. And I always talk about this internally, right? So there's three things that we're within that data. There's data about the person. So the demographic data, who they are, where they go, what hobbies they have, interests, all that sort of stuff. So great data about the person, data about what the person thinks about the thing. So whether they like a brand and a piece of advertising, a piece of innovation. So data about that and then data about the thing. And tag data is so important for creating genuine insight loops. And so having a tagging infrastructure around advertising and innovation and brand that can talk to each other, so everyone thinks of the second piece of data, which is what I've said about the thing. But without those other two pieces of data, you really don't have the sort of power of insight. And I think being able to architect a system that makes it possible to look across those three data streams and really get a combined view of the market and what people are thinking about products and campaigns and brands and bring that together is incredibly powerful when done right.
[51:45] Nataly Kelly: Yes, well, and the last piece of information that marketers use that relates to that is what did they do in response to what they thought about the thing?
[51:54] Steve Phillips: Yes, yes, yes, yes, yes. You know, and like that's absolutely fair.
[51:59] Nataly Kelly: Are always trying to drive and that's the click behavior and like going through to purchase data and all of those pieces that really are what the marketers are charged with is like you got to make an impact on the top line revenue growth, at the bottom line profit. You know, you have to make sure that you are thinking about those things and connecting the dots there. And that is I think the heart of the challenge. And that's what I love about the connected about the new flywheel in our model. It really shows how these things connect because if you only stop at well what do they think and why do they think that and you know, but it doesn't turn into dollars or pounds and action and reaching people's households and pockets, then we failed.
[52:45] Steve Phillips: Absolutely.
[52:47] Nataly Kelly: Because we have to grow the brand and grow the business in order to stay alive and afloat and keep our jobs, everybody's livelihood going. Like this is the heart of business and why business people should care about a Connected Insights Flywheel because it's Connected Insights has the potential and power to actually grow the business. And that's where I see consumer insights professionals making the biggest impact. But they won't get there unless they embrace AI technology, synthetic, all the things that we've been talking about and architecture in the future.
[53:20] Steve Phillips: Exactly, exactly.
[53:22] Nataly Kelly: Great. So Steve, we are going to move into our lightning round where we're going to ask a few quick hitting consumer insights related questions. So ready to go?
[53:34] Steve Phillips: Ready.
[53:34] Nataly Kelly: Okay. What's one insights activity you think AI will fundamentally reinvent over the next 12 months? Not just make faster.
[53:43] Steve Phillips: I think it's the consumption of that insight. I think it's moving from PowerPoint meetings to embedded into systems. And the more we embed into systems, it just means the insight is used more often by the people who are working at the coal face building whatever it is we're building. So definitely for me it's embedding. What's one marketing decision where consumer input still enters the process far too late?
[54:08] Nataly Kelly: I can't give you one. It's all of them. There's lots of little decisions that are made all the time we're considering consumer insights matter and we typically only reserve consumer insights for specific decisions and not as many where they could really, truly be impactful. Steve, what's one type of decision you'd be comfortable making with synthetic data today? And one where you'd still insist on human validation?
[54:39] Steve Phillips: Yeah, I go back to the risk thing I said earlier. So if you're doing a campaign and you've got some hero piece of communication that you think are genuinely creative, if you think they're really dull, test them with synthetic. If you think they're genuinely creative, then absolutely you need humans. But when you see lots of digital instances of those communications, lots of different versions, et cetera, then I would say use synthetic and use synthetic because you want some insight about them. You want to know, have you made a mistake? Is there something in there that is problematic? That's the first thing you want to know. But you also want to know, does it connect with the rest of the things I'm doing? So just at the tagging level, we can give insights that say there's similar distinctive assets to your main pieces of creative. So it's part of the same campaign. So there's a lot of things you can get from synthetic, but for the major decisions, definitely stay with humans. So, Natalie, what's one thing marketing leaders should stop doing if they want to get more value from their insight teams?
[55:40] Nataly Kelly: Don't put baby in the corner. I don't know if you've seen Jody Danson.
[55:46] Steve Phillips: Yeah, yeah, yeah.
[55:47] Nataly Kelly: Insights teams are baby in the corner. You know, triggered, siloed out there. You know, I really believe if there's one corner the insights team should be in, it's the corner office. They should be in that corner fighting with the CEO, the CMO, to inform the strategy. That is the biggest thing that I think any CMO can do to bring the consumer insights team in there and stop putting them in the corner.
[56:15] Steve Phillips: I should say, as the parent of a daughter, that when my daughter was 17, she definitely should be put in the corner.
[56:25] Nataly Kelly: Well, maybe in time out when they're dirty analogy. I think it works. Steve, finish my sentence. The organizations that win the next year of Connected Insights will be the ones that are curious.
[56:44] Steve Phillips: So I think now to expand on it slightly, we're talking about a changing world with incredible new technologies coming along, and we've always been curious. As an industry, we have to maintain that curiosity and learn, and then we can apply.
[57:02] Nataly Kelly: Okay, before we wrap, Steve, we have one final question we ask our guests. So it only seems fair that we answered ourselves. What's the one thing that marketing and insights teams need to do together in the next 12 months to help their company win with consumers and grow their business.
[57:18] Steve Phillips: So I would like to see them working together on that architecture piece — piece of mapping the roles, the functions, the workflows that would be improved with insight — and then the insight teams work on how to deliver that insight into that map. And Nat, I'm throwing the same question back to you.
[57:39] Nataly Kelly: Focus on the relationships. Build the relationships, because once you have the relationships, you can talk about building the architecture. And then, Stephen, I will build half of come true.
[57:52] Steve Phillips: Brilliant.
[57:54] Nataly Kelly: Great. Okay, so that wraps up this episode of the Inside Insights podcast. Steve, we have had a lot to think about in this episode. I think the big takeaway is pretty clear. Getting connected is no longer the finish line. The opportunity now is to make consumer understanding continuous so every decision can build on what came before. If you'd like to dive deeper into the findings, we've linked the full Connected Insights Imperative 2026 report directly of the show notes wherever you're listening. If you'd like to contact either Steve or myself, you can find links to our LinkedIn profiles in our show notes or at insideinsightspod.com and if you haven't subscribed yet and you want a regular stream of conversations with some of the world's leading marketing and insights professionals, hit that subscribe button in your podcast app or follow us on YouTube. Okay, that's all for today. See you on the next episode of Inside Insights.
[58:46] Steve Phillips: Thanks. Thanks. It.