The three-minute version
(We get it, you’re busy and just want to cut to the chase.)
In today’s fast-paced marketing world, we are inundated with data but often struggle to extract meaningful insights. The document highlights a critical issue: while AI can process vast amounts of data quickly, it often misses the emotional and cultural nuances that only human interpretation can provide.
At Spark, we believe that the brands making a real impact are those that use smart technology to deepen their understanding, not replace it. This is where our What3™ approach comes in, combining three essential elements:
What’s Happening
This involves understanding the objective data and patterns of customer behaviour. It’s about identifying who is buying what, how attitudes are shifting, and where customers are dropping off. We use cutting-edge quant tools and AI-powered solutions to bring clarity to chaos.
So What?
This is the emotional and cultural decoding behind the data. It’s not just about what people are doing, but why they are doing it. We blend video-based qualitative research, mobile-first ethnography, AI-powered sentiment analysis, and human interpretation to bring depth and meaning to your data.
Now What?
This is the commercial action plan. Research isn’t valuable until it helps make better decisions. We translate insights into strategy recommendations, messaging frameworks, innovation opportunities, and experience fixes. Every Spark project wraps with clear, practical next steps your team can act on immediately.
While AI has revolutionised research with faster survey analysis, real-time video coding, predictive modelling, and automated insight dashboards, it often falls short in interpreting intentions, social context, and what actually moves people to act. We use AI to power speed and breadth but layer it with human interpretation to unlock insights that stick.
The article below goes into some case studies, e.g. a major UK bank’s conversational AI debacle illustrates the pitfalls of relying solely on AI. The bank’s AI platform failed to resolve over a third of customer issues due to a lack of human interpretation and context, leading to confusion, frustration, and lost trust.
In conclusion, modern marketers need more than just data; they need confidence in their decisions. By combining the raw speed and processing power of AI with the interpretive intelligence of human researchers, leading brands can make decisions with more confidence, more quickly. At Spark, we believe in using AI smartly, not blindly, blending cutting-edge tech with sharp, experienced human minds to turn complexity into clarity. #intelligencemeetsinstinct.
The full version (Including steps to ensure you’re getting the best from AI)
The New Research Imperative
Today’s marketers face a strange kind of problem. We’ve never had more data at our fingertips and yet, somehow, we’re losing touch with the people behind the numbers.
- There’s plenty of data, but not enough real meaning.
- We’re moving faster than ever, but sometimes at the cost of depth and emotional clarity.
- AI is changing the game, but it can’t always tell the full human story.
At Spark, we think the brands making a real impact aren’t the ones blindly chasing “AI, first.” They’re the ones using smart tech to deepen their understanding, not replace it. That’s where true insight lives, in the space between the numbers and the nuance.
At Spark, we call this the What3™ approach, a proven model that combines:
What’s Happening, The Truth in the data
This is where we start. The objective signals, behaviours, and patterns that tell you what your customers are doing, not what they claim they do.
- Who’s buying what?
- How are attitudes shifting?
- What content’s being ignored or engaged with?
- Where are customers dropping off?
We use cutting, edge quant tools and smart, AI, powered solutions to bring clarity to the chaos, but we don’t stop there because at Spark, we know that data without interpretation is just noise (you’ll hear us say that a lot).
It’s not just about the numbers. It’s about making sense of them in a way that’s useful, human, and actually helps you move forward with confidence.
So What? The Emotional and Cultural Decoding Behind the Data
This is the layer most research misses. It’s not just about what people are doing; it’s about why they’re doing it.
This is where our human expertise comes in:
- What are the emotional drivers behind behaviour?
- How do cultural cues shape interpretation?
- What’s the gap between what people say and what they feel?
We blend video, based qual, mobile, first ethnography, AI, powered sentiment analysis, and good old, fashioned human interpretation to bring depth and meaning to your data. We believe that insight shouldn’t just inform, it should connect, inspire, and stick.
Now What? The Commercial Action Plan
This is where we earn our keep. Research isn’t valuable until it helps make a better decision, faster, clearer, and with more confidence.
We translate insights into:
- Strategy recommendations
- Messaging frameworks
- Innovation opportunities
- Experience fixes.
- Market entry or growth roadmaps
Every Spark project wraps with more than just findings, it ends with clear, practical next steps your team can act on straight away. Whether you’re shaping a campaign, refreshing your brand, or rethinking the customer experience, our What3 framework makes sure you walk away with direction, not just data.
It’s not about machines replacing researchers. It’s about machines freeing us to be better ones, faster, sharper, and more human.
AI Is Here, But It’s Not the Whole Answer
There’s no doubt AI has changed the game in research. It’s delivered:
- Faster open, end survey analysis
- Real, time video coding and clustering
- Predictive modelling and sentiment scoring
- Automated insight dashboards
And it’s tempting to believe that AI alone can deliver ‘truth.’
But here’s the reality: most AI is great at the “What’s Happening,” and pretty poor at the “So What?”
It can detect words, patterns, and reactions.
But it can’t always interpret intentions, social context, or what actually moves people to act.
That’s why Spark’s What3™ framework is so critical. We use AI to power speed and breadth, but we layer it with human interpretation to unlock the insight that sticks:
- The right questions (not just faster ones)
- Contextual judgement
- Cultural understanding
- Sense checked intuition.
Without this, you’re just automating noise, not insight.
Case Study: UK Bank’s Conversational AI Debacle
A major UK bank invested heavily in conversational AI and chatbots aimed at reducing operational costs. However, the result was not deeper insight or better customer experience, it was confusion, frustration, and lost trust.
What Went Wrong:
- The AI platform was built around 180 labels and 120 chatbot intents, yet there were over 480 distinct customer journeys online, many of which weren’t supported. This mismatch meant the AI couldn’t resolve over a third of NPS, detractor issues, it simply didn’t recognise or could not handle them. (Source cxtoday.com)
- The chatbot often deflected queries that required deeper context or nuance, pushing customers into dead ends, or looping back to live agents.
Why It’s a Lesson in “Automating Noise, Not Insight”:
- The AI captured top, level intent, what customers were saying.
- But it completely missed why they were saying it, context, emotion, complexity.
- No mechanism existed for human interpretation, decision, making, or corrective escalation.
- As a result, instead of uncovering meaningful insights, the bank accumulated false positives, repeated issues, and customer friction.
Why This Matters to Marketers
Modern marketers don’t just need more data, they need confidence.
Not just in the numbers, but in the decisions that come next.
If you’ve ever sat through one of those research debriefs, you know the kind: “this figure went up 5%, that one stayed the same” you’ll know what we mean. Technically accurate, sure. But how did it leave you feeling? Probably not inspired. And definitely not brimming with confidence about your next move.
At Spark, we think insight should do more. It should energise, clarify, and give you the confidence to act, not just tick a box.
Confidence is a preference of the habitual voyeur of what is known as…Sparklife (sorry Damon, couldn’t help myself, if you know, you know)
Marketers need confidence to:
- Launch faster, without second, guessing whether the insight is robust enough.
- Kill bad ideas early, before budgets, teams, or reputations are on the line.
- Justify spend in the boardroom, not with vague signals, but with evidence grounded in real customer understanding.
In a world where marketing budgets are under pressure and attention is scarce, there’s no time for guesswork, gut feel, or generic dashboards.
What you need is a research model that balances speed with substance.
The AI + Empathy Model Delivers That Confidence
By combining the raw speed and processing power of AI with the interpretive intelligence of human researchers, leading brands are getting the best of both worlds, and making decisions with more confidence, more quickly.
Speed
AI takes care of the heavy lifting, coding open ends, tracking sentiment shifts, and flagging emerging trends in real time. What used to take weeks now takes hours, so you can act while your competitors are still reading the charts.
One retail client spotted a sudden shift in sentiment around a key product and adjusted messaging within 48 hours, before it became a problem.
Depth
Human researchers bring the “feel,” they pick up on tension, read between the lines, and ask the questions algorithms don’t know to ask. That turns surface, level feedback into something richer, more emotionally resonant, and genuinely useful.
Cultural Relevance
AI can spot patterns, but it takes a human to know if something is ironic, aspirational, or just plain cringe. And that matters, especially if you want your campaign to land well in real, world markets, not just look good in a deck.
Commercial Actionability
Insight only counts when it leads to action. That’s why Spark’s What3™ model always answers the question: so what now? You don’t just hear what people said, you know what your brand should do next.
It’s Not About Being Cutting, Edge for the Sake of It
This old chestnut. Spark has been banging this drum for almost 18 years. There was a time when we advocated Moderator, less research to a room of researchers. I think we may have been perceived as being witches, but time has proven we were right. There is a time for letting the machines do their thing and a time for human input – research expertise helps us dictate when those times are and how to manage each project accordingly.
This isn’t innovation theatre. It’s about using technology to get closer to your customer, in a way that’s faster, truer, and more commercially useful. When AI is fused with empathy, and insight is filtered through the What3™ lens, you’re not chasing trends. You’re making smarter decisions, with clarity, confidence, and impact.
So how do you ensure you get the most from AI and your agency, well you can shortcut it by just hiring Spark…let’s suppose you want to know the best steps to having it all, AI + Human. Here are what we see as the 8 steps.
8 Steps to Ensure You Get the Best of Both AI and Human Insight
1. Start with a Sharp Brief, Not Just a Broad Topic
- Define what you need to know (not just what you think you want to measure).
- Clarify the business context: Is this informing a product launch? Repositioning? Campaign validation?
- Be clear on what “good” (or “success”) looks like in terms of output, strategic clarity? Innovation direction? Stakeholder, ready storytelling?
A good agency should positively challenge your brief, not just take it at face value.
2. Ask How AI Is Being Used, And Where Humans Still Add Value
- Don’t just ask if AI is being used, ask how.
- Examples:
- Is AI used to code open, ended responses?
- Is it analysing sentiment?
- Is it being used to surface themes before humans interpret?
- Ask where human judgement is applied, and why it’s essential to balance the tech.
The best agencies will explain not just what their AI does, but what it doesn’t do well, and where people step in.
3. Look for Frameworks, Not Just Platforms
- Ask how the agency integrates both AI and human insight into a structured model (e.g., Spark’s What3™).
- Avoid tech, for, tech’s, sake approaches. You want joined, up thinking, not disconnected tools.
- Check if the agency can move smoothly from quant to qual, or digital qual to action, without friction.
Frameworks ensure you’re not just receiving data, you’re receiving thinking.
4. Insist on Human Interpretation of AI Output
- AI can cluster and summarise, but it doesn’t know your brand, market context, or customer nuance.
- Make sure a senior researcher will review the AI, led themes, apply cultural understanding, and connect dots to your business problem.
Robots + Human = Speed + Substance = Happy clients and happy Sparkies!
5. Demand Real, Time Transparency
- Ask if you can observe the process, not just receive a post, project report.
- Can you view real, time dashboards?
- Can you preview emerging themes during fieldwork?
This builds trust and helps you course, correct early if needed. At Spark we can facilitate this, but I’ll be honest many clients just leave us to get on with it. As I said, resources are tight, so clients have enough of their own work to be getting on with.
6. Validate with Human Voices (Even in Quant)
Even the biggest quant studies get better when you add a human touch, open, ended prompts, verbatim quotes, even a few short video clips.
These little moments bring the data to life. They help stakeholders feel the insight, not just read it, adding empathy, believability, and real, world relevance to the boardroom conversation.
At Spark, we combine the power of scale with the power of story, turning numbers into narratives that move people to action.
7. Make Time for a Strategic Debrief, Not Just a Download
- Don’t settle for a 100, slide deck or a PDF dump (Both are actually illegal in Spark)
- Ask for a facilitated session where the agency walks through:
- The story in the data
- The implications
- The “Now What”, clear, prioritised recommendations
This is where AI insights leads to real business impact.
8. Choose Partners Who Push You
- Work with agencies who challenge lazy assumptions, debunk overreliance on AI, and know when to pause the tech and talk to people.
- Ask for examples of when AI alone wasn’t enough, and what human thinking added.
The best partners combine rigour with imagination. That’s where confidence comes from. AI is the engine. Human interpretation is the steering wheel. (you can steal that one!)
You need both to get where you’re going, fast, safely, and meaningfully.
Case Study: If a Robot Ran Your Research
What happens when you let AI take the wheel – and forget to bring a human along for the ride?
Client Context
A well, known UK retail brand wanted to test a new campaign concept aimed at 18–30, year, olds. Keen to move fast, they opted for a “fully automated” DIY research solution promising:
- AI generated surveys
- AI analysed results.
- Instant dashboards with predictive scores
No researchers were involved. The promise? Insight at speed, with zero fuss.
What Went Wrong
1. Surface, Level Questions
- The AI wrote generic, templated questions: “How appealing is this idea on a scale of 1–10?”
- No tailoring for the cultural context, tone, or emotional drivers of the target group.
Result: Respondents clicked through with minimal engagement. The AI got clean data, but no depth.
2. Misleading Analysis
- The dashboard flagged the concept as “positive” based on sentiment, scored open, ends.
- But human review later showed sarcasm, irony, and even frustration that the AI failed to detect.
Example: “Yeah, great idea… if we were still in 2005.”
The AI read this as positive due to “great idea.” Note: A lot of AI Robots were born in the USA; British and Irish consumers are masters in sarcasm in a way US consumers are not; we need to be mindful of this when deploying AI.
“Will you really?” “I will, yeah” Interpret that, robot.
3. Blind Spots in Segmentation
- The tool couldn’t segment responses meaningfully by life stage, lifestyle, or context, only demographics.
- It missed a crucial insight: students loved the idea; early, career professionals found it patronising.
No one thought to look for emotional divides, just age brackets.
4. ‘Actionability’ Fell Flat
- The client presented the findings to the board with confidence: “The concept is a hit.”
- They ran the campaign. It flopped.
- A later Spark project revealed the core issue: the idea lacked authenticity and emotional resonance, something the AI simply couldn’t assess.
What Spark Would’ve Done Differently
At Spark, we start by getting a clear picture of what’s happening, using robust, large, sample quant combined with real behavioural data. It’s not just about what people say, but what they do. A robot, only approach might stop at an auto, run survey, giving you top, level results without the context or depth that real decisions require.
Next, we ask so what? That’s where the magic happens. Our AI assisted tools help surface themes quickly, but it’s our experienced researchers who decode the cultural signals, spot the tensions, and bring meaning to the data. A robot, on the other hand, will likely give you a few sentiment scores, useful, but often missing the why behind the numbers.
Finally, the most important bit: now what? Spark delivers clear, commercially relevant recommendations. You walk away with actions you can take, not just insights to admire. The robot, only version might offer a predictive score or output, but without guidance, it leaves you with more questions than answers.
Takeaway: AI ≠ Insight Without Humans
AI is undeniably powerful. It can scan thousands of responses in seconds, spot patterns we’d never see on our own, and churn out neat, looking outputs in record time.
But speed isn’t the same as substance.
Without human input, the curiosity, the challenge, the context, AI is just a blunt tool. Fast, yes. But effective? Not always.
Letting AI run the show might feel like a time, saver, especially when you’re under pressure to “do more with less.” But here’s the risk:
If it skips the real questions, misreads sarcasm, or misses the emotional and cultural context, you’re not getting insight, you’re getting noise. Nicely packaged, but dangerously misleading.
And that kind of noise? It leads to:
• Misguided strategies
• Tone, deaf messaging
• Wasted budget
• And lost trust with the people who matter most.
At Spark, we believe in using AI smartly, not blindly. We blend cutting, edge tech with sharp, experienced human minds who know when to dig deeper, when to ask the awkward questions, and how to turn complexity into clarity.
In today’s world, it’s not just about being fast. It’s about being right, and real.
Get in touch via our Sparkbot at…
Nope! Let’s discuss how we can apply the What3 framework to your brand challenges to help build brand growth and power better, more confident decisions. Drop us a line at hello@sparkmr.com and let’s chat!
