Tim Bock

Segmentation analysis is an extremely powerful tool for providing stakeholders with strategic guidance - but it’s also a potential minefield if you’re not careful with how you do it… Here’s a quick reality check on the kinds of pitfalls you should avoid: 1. Too Many Variables, Not Enough Focus Throwing every variable into the mix sounds thorough, but it’s a recipe for confusion. The “curse of dimensionality” means that with too many variables, meaningful segments disappear, and everyone starts looking the same. Imagine trying to find patterns in a haystack made of haystacks. 2. Sample Size Struggles Small samples and lots of variables? That’s a fast track to unreliable segments. If you don’t have enough data, your results won’t hold up - and your segments may just be random groupings. 3. Masking & Redundant Variables Including variables that don’t actually differentiate customers (masking variables), or those that are just duplicates (highly correlated), can hide real structure and create “rubbish bin” segments. It’s like trying to tune a radio with too much static. 4. Algorithm Instability Many clustering methods are sensitive to starting points and can produce wildly different results each time you run them - especially if your data isn’t well-prepped. 5. Response Biases Rating scales can often serve the creation of artificial segments (like “high raters” vs. “low raters”) as different geographies (and even individuals) can use scales entirely differently, responding with their own bias. The output of your analysis then becomes just a descriptor of these biases. Ultimately, segmentation analysis comes with a myriad of issues and simply running a clustering algorithm on your collected data is not likely to be a good enough approach if you care about a quality result. At our Nuremberg event, my colleague Keith Chrzan will be taking the Intro track through his fantastic training on how you too can undertake Segmentation analysis badly – and well (we recommend you do it the latter way).

58

15 Comments

Great collaboration from Greg Allenby and Peter Kurz exploring the buy / no-buy decision in conjoint - particularly in the new product space. They questioned, how much survey takers really know about the market when completing a conjoint and if we should inform respondents to get better willingness to pay data. They found that in a new product space respondents learn as they go and that learning affects their choices. So what does that mean? Well, one idea is to test different information conditions across respondents. Perhaps you’d have 3 cells: Cell 1 - No context Cell 2 - Basic info on brands and prices Cell 3 - Full competitive landscape overview And you could even consider a short video to help set expectations, reduce confusion, and lead to more meaningful data. My key takeaway: Conjoint results are only as good as the mental model the respondent brings into the task. If they don’t understand the features, price ranges, or brand landscape, their choices may not reflect reality. If we want better insights, we have to design with the respondent’s understanding in mind—not just their choices. Great work both! 👏🏼 #mrx #conjoint Sawtooth #sawtoothconf

32

1 Comment

I'm unashamed about Significant Insights - I bloody love it! Reading this piece it’s clear that the rising stars in our industry have a lot to say about what matters most. Their enthusiasm for meaningful, impactful work is a powerful reminder of why market research and insight is so rewarding. The 30 under 30 cohort are drawn to the blend of creativity and analysis, the chance to influence decisions, and the opportunity to make a tangible difference. It’s about solving real-world problems and helping businesses see the bigger picture. For those of us who've been in the game longer, there’s a lesson here: how often do we stop and truly listen to the perspectives of those starting their journey in this space? Their ideas and energy can spark fresh approaches and keep our industry moving forward. If you are a leader in an insight agency, how are you helping to create opportunities for new talent shaping the future of insights? #MRX #Insights #MarketResearch #MarketingResearch

37

10 Comments

Op-Ed: Market Research Is at Risk — But Behavioral Data Is the Savior Traditional market research is on the verge of obsolescence. Survey fraud, synthetic responses, and outdated methods have eroded trust—and generative AI will finish the job. Why pay for human-written reports when LLMs can synthesize thousands of responses in seconds? But not all is lost. The real future lies in behavioral data—what people actually do, not just what they say. App usage, website visits, and location patterns offer a far more reliable picture of the consumer. Companies that own first-party behavioral data will become the power brokers of the AI age. As AI models hunger for real-world inputs to stay relevant, proprietary behavioral datasets will become the new oil—fueling training, personalization, and prediction across every industry. The value won’t be in asking better survey questions, but in knowing the customer before the question is asked. The future of market research isn’t in sample sizes or Likert scales—it’s in human behavior at scale. Firms that capture, unify, and ethically apply this data will lead. The rest will be replaced. Behavior doesn’t lie. And in the world of AI, it’s the only truth that counts.

25

THE NEXT FRONTIER IN UNDERSTANDING CONSUMER SENTIMENT – “QUINTUPLES”, BY LEVERAGING AI AND MACHINE LEARNING. Sentiment analysis is a key tool in understanding consumer opinions, but traditional methods only classify sentiment as positive, negative, or neutral—lacking the depth needed to fully capture complex sentiments. Quintuples solves this by breaking these sentiments into five key components: 1. Subject – The entity being discussed 2. Sentiment (Polarity) – Positive, negative, or neutral 3. Emotion – The feeling driving the sentiment 4. Object (Aspect) – The feature being evaluated 5. Context – Conditions influencing sentiment (e.g., sarcasm, mixed emotions) How does it work? It extracts text data from various sources such as social media, online reviews, and surveys. Then Natural Language Processing (NLP) and Machine Learning Models are applied. Why is this better? 1. Detailed & Precise: Captures specific emotions and aspects. 2. Handles Mixed Sentiment: Accurately reflects complex opinions. 3. Context Awareness: Detects sarcasm and context for better accuracy. 4. Aspect-Based Insights: Identifies key product features driving satisfaction or dissatisfaction. What are the applications in Market Research? 1. Customer Feedback: Dive deeper into survey responses and reviews. 2. Brand Monitoring: Track real-time sentiment on social media. 3. Competitor Analysis: Identify gaps and differentiation opportunities. 4. Market Segmentation: Segment customers based on emotional responses. Incorporating Quintuples in market research can provide a wealth of actionable insights and help make data-driven decisions, thereby helping businesses optimize, improve their #CX, and stay ahead. #SentimentAnalysis #MarketResearch #CustomerInsights #NLP #AI #MachineLearning #CustomerExperience #Innovation

80

15 Comments

The market research industry faces a fascinating contradiction: despite well-documented data quality challenges, the global insights market continues to thrive, now exceeding $142 billion annually. As someone who's spent 30 years in this industry, I've observed first-hand how survey design flaws, programmatic sampling issues, and sophisticated fraud (and most recently, AI-powered bots) can compromise raw data quality. Industry reports suggest 40-70% of responses may need cleaning or removal. It's worse than it's ever been. Yet client spending increases year after year. Why? Because what clients truly purchase isn't data points, it's confidence. The real value of market research has never been in decimal-point precision. It's in the strategic reassurance and de-risking that comes from having external validation for business decisions. When a CMO needs to justify a brand repositioning or a Product Lead requires evidence for feature prioritisation, research provides the necessary armour against organizational scepticism. Research firms understand this dynamic intuitively. The sophisticated data cleaning processes they've developed aren't just technical necessities. They're part of delivering what clients actually need: anxiety reduction. This explains why, despite industry hand-wringing about data quality, client satisfaction remains high. The research delivers on its core promise: making decision-makers feel more confident navigating uncertainty. Perhaps we should stop viewing data quality challenges as existential threats and instead recognize them as technical hurdles in an industry whose fundamental value proposition remains intact. The market keeps growing not despite these challenges, but because the underlying need - decision confidence - remains as strong as ever. What's your experience? Has research helped you make more confident decisions, regardless of methodological imperfections? #MarketResearch #BusinessStrategy #DataQuality #ConsumerInsights #DecisionMaking

28

3 Comments

Curiosity and analytics don’t have to be intimidating. On Data Gurus, I’m joined by Megan Peitz, Founder and Owner of Numerious Inc. to talk about making advanced analytics approachable and practical. Megan shares her journey from statistician to entrepreneur, her work helping brands like Google make smarter product decisions, and how she’s turning complex models into accessible tools through her training program. We also discuss how curiosity, clear communication and human context remain essential even as AI reshapes the research landscape. Click the link in the comments for the episode. #Analytics #M&A #Data #Strategy #Innovation #Acquisitions #MRX #Restech

46

5 Comments

We are running some qualitative research online and we are having to deal with respondents using AI to formulate answers...bit of investigation and it turns out quantitative research using online community sample is rife with AI fake respondents and cannot cope...there are some threads on Reddit being run by quant researchers who talk of how sophisticated AI has got, when it comes to slipping past screeners. The AI bot respondent thing is out of control. We reckon we can spot AI assisted answers in the context of a small qual project, like the one we are running currently but I really would be extra cautious about any research purporting to use panel sample online, or in quant research. Going forward, I am only going to trust f2f, or live f2f online, using likes of Zoom, otherwise i will be very wary of a lot of research. Not that i wasn't before but at least i knew real people were generally being engaged with at the coalface and if the research findings were questionable, that was down to client and researcher, not respondents, usually Ironic i am invited to re - write this with AI. No AI was used in the formulation of this rant. Might have to start deploying that in my research reports

329

126 Comments

Don't undervalue spending even 30 minutes with a consultant in market research to discuss your survey questions, approach, distribution, analysis plan, and best practices. 🎯 With so much AI out there and pre-built templates in survey platforms, it might seem as though getting feedback from an actual life researcher isn't valuable. 👎 With continued growth in DIY platforms and employees being pushed to provide more value from their positions, jumping into a survey project and doing a research study without any experience is tempting. ⏩ Although some survey drafts I've reviewed over the years from other organizations are much further along than others (several might be missing out on a career in this field, including this most recent example 🙂), the constant theme is there are always tips, pointers, ideas, and changes to suggest coming from experience. We have gone through these studies so often that researchers can see things coming miles away - even at the design/draft phase. Some advice from a short 30-minute virtual meeting I had: 1️⃣ Benefits of mutually exclusive/collectively exhaustive categories 2️⃣ Valuable ways to integrate your CRM to data on the front-and-backend 3️⃣ Benefits of a pre-notice ahead of the invites (and a thank you note after) 4️⃣ Discussing survey flow, how to increase responses and engagement 5️⃣ Providing perspective into timelines for fieldwork for strategic planning 6️⃣ Benefits of unique links and separating fieldwork by sample sources 7️⃣ Best-practices for open-and-closed ended questions 8️⃣ What it means to design for mobile-first and above-the-fold 9️⃣ How to handle PII and intrusive questions 🔟 Considerations for weighting data and segment breakdowns All were accomplished in a half-hour with the client, which will set them up for more immediate success, better results, and clear action items even though Drive Research is not commissioned for this project for a non-profit. All pieces can be extracted from a conversation if you work with a researcher who knows what to look for. Need guidance on your research study? Feel free to reach out, and we can arrange something hourly for you. The advice and direction you receive from anyone on our team is worth its weight in gold. We have an A+ research crew here who has seen it all. -------------------------------------- Need MR advice? Message me. 📩 Visit DriveResearch.com 💻  1600+ articles to help you. ✏️ --------------------------------------

32

1 Comment

🚨 New Report: 2025 Australian Consumer Trends: "Lonely, Stretched and Scrolling" From AI acceleration to mental health, affordability pressures to the rise of the sandwich generation — Australians are feeling the weight of change. We’ve just released our 2025 Consumer Trends Report, packed with insights that matter to brands navigating uncertainty and designing for a more conscious, digitally-fatigued, and values-driven consumer. 📊 Inside: 👉 What’s keeping Australians up at night 👉 Generational shifts in digital behaviour 👉 The sustainability vs. affordability tension 👉 Evolving expectations around work, AI, and wellbeing Download the full report here: https://bit.ly/4ksZXtI 📎 Quick summary attached for those short on time If any of these themes are on your radar for the year ahead, let’s chat. We’re already working with several teams to shape custom research around these trends — and would love to support your strategy too. #ConsumerTrends #MarketResearch #DesignThinking #BrandStrategy #ProductDesign #Sustainability #AI #Australia2025

23

2 Comments

I have a new career – I’m officially an industry thought leader. Over coffee, a client recently asked for my perspective on the shifts I’ve been seeing in the Insights and Analytics space. I drew on the conversations I’ve had with Insights Directors, Senior Customer Insights Analysts, Chief Data Officers and agency owners… Fast forward, and I found myself presenting to the entire Insights team at a Big 4 consultancy during a one-hour session. Here are three key highlights from that discussion: - In 2010, insights sat firmly within the marketing function, while data was largely seen as the domain of IT. Today, those silos have blurred and so have the skillsets needed to thrive. - AI is brilliant at summarising data, but it’s Insights and Analytics professionals who bring value by sense-checking, interpreting nuance, and knowing when an outlier deserves a deeper dive. - These professionals are experts in balancing speed with rigour and more importantly, knowing how to adapt that mix to suit unique social or commercial contexts. As we edge further into the era of generative AI, it’s worth remembering: technology offers a one-size-fits-all solution, but true insight is bespoke. Insights and data professionals are the difference between overalls and haute couture.

65

4 Comments