{"id":4451582,"date":"2026-08-19T16:25:13","date_gmt":"2026-08-19T14:25:13","guid":{"rendered":"https:\/\/www.deptagency.com\/insight\/three-data-types-that-actually-predict-customer-decisions\/"},"modified":"2026-08-20T16:22:48","modified_gmt":"2026-08-20T14:22:48","slug":"three-data-types-that-actually-predict-customer-decisions","status":"publish","type":"article","link":"https:\/\/www.deptagency.com\/en-uki\/insight\/three-data-types-that-actually-predict-customer-decisions\/","title":{"rendered":"Three data types that actually predict customer decisions"},"content":{"rendered":"
\n\t\n\t\t\n\t\t\t<\/mask><\/g><\/svg>\t\t\t\n\t\t\t\tInsights\t\t\t<\/span>\n\t\t<\/div>\n\t<\/div>\n\tThree data types that actually predict customer decisions<\/h1>
\n\t\t\t\n\t\n\t
\n\t\tYang Song\t\t
\n\t\tAI Solutions Architect\n\t\t\t<\/div>\n<\/div>\n\t
\n\t\t\t\t\t
\n\t\t\t\t\n\t\t\t\t\tDate\t\t\t\t<\/span>\n\t\t\t\t\n\t\t\t\t\tAugust 19, 2026\t\t\t\t<\/span>\n\t\t\t<\/div>\n\t\t\t<\/div>\n\n\t\n<\/div>\n\n\n\tDepending on whose numbers you use, between 70% and 85% of new consumer packaged goods fail within two years of launch.

Nielsen’s Breakthrough Innovation research sits at the top of that range: of roughly 30,000 new products launched in the US each year, only about 15% are still commercially viable after 24 months.<\/h2>\n\tThis happens in an industry that spends an eye-watering amount of money on consumer research. Those funds are invested in much the same way across brands, on tools that can track a mouse jitter, sentiment shifts in 40 languages, and every click your prospects have made since 2017.

Yet, despite all that raw capability and expense, the vast majority of products still<\/em> fail. So, where\u2019s the problem?

It\u2019s in the data\u2014or, rather, in the interpretation of it.

Many brands rely on three distinct types of data: what people say, what people search, and what people do. The problem is that they\u2019ve built their entire strategy on the assumption that each of these three types of data is an interchangeable input to the same output, generically labeled \u201ccustomer insights.\u201d

However, none of these three data types are interchangeable. Each produces its own type of insight that, on its own, is incapable of accurately reporting on consumer behaviour. But, if triangulated together properly, they can help your brand drive growth<\/a>.<\/p><\/div>\n\t\n\t<\/div>\n\n\n\tThe three data pillars and what they\u2019re used for<\/h2>\n\tThere are three distinct data types that (used properly) actually predict customer behavior:<\/p>\n\n\t
  • Attitudinal data:<\/strong> what customers say they believe\u00a0<\/li>
  • Intent data:<\/strong> what customers are researching<\/li>
  • Behavioral data:<\/strong> what customers actually do<\/li><\/ul>\n\nThese measure different psychological states. They predict different moments in the customer journey. They carry wildly different levels of confidence about what anyone will actually do with their money.<\/p>\n<\/div>\n\t\n\t<\/div>\n\n\n\t\n\t\t\t\t\t\n\t\t\t\n\tPillar 1: Attitudinal data, \u201cThe Why\u201d<\/h2>\n\tThe subjective layer. Opinions, feelings, values \u2014 the stuff you ask people about in surveys and focus groups to approximate the motivations behind their decisions.

    If we think of behavioral data as the scoreboard, attitudinal data represents the post-game interview. It\u2019s useful for understanding how the coach frames the loss but, at the end of the day, it still doesn\u2019t change the final score.

    What it\u2019s good for:\u00a0<\/strong>

    Long-term brand positioning. Psychographic segmentation. The emotional scaffolding of your category. Spotting cultural shifts early.

    What it\u2019s misused for:\u00a0<\/strong>
    Predicting whether or not people will actually buy a product.<\/p><\/div><\/div>\n\n\n\t\n\tTake sustainability as an example. About     65% of consumers<\/a> say<\/em> they want to buy from sustainable, purpose-driven brands. In reality, however, only about 26%<\/a> actually do. This 39-point gap of aspirational fiction shows up year after year, in study after study, and yet every year the original 65% statistic re-emerges at the core of a new product strategy.\u00a0

    If you\u2019ve ever sized a supply chain or set a price point based on \u201cpurchase intent\u201d from a survey, you already know how that story ends. The data didn\u2019t lie to you. You lied to yourself about what the data was.<\/p><\/div>\n\t\n\t<\/div>\n\n\n\t\n\t\t\t\t\t\n\t\t\t\n\tPillar 2: Intent data, \u201cThe When\u201d<\/h2>\n\tIntent sits between belief and action. The data representing it shows up in the digital breadcrumbs people leave while actively researching products, things like keyword surges, content downloads, competitor page visits, or their third trip to your pricing page in a week.

    What it\u2019s good for:\u00a0<\/strong>
    Timing. Knowing when a buyer has moved from \u201cmildly curious\u201d to \u201cactively trying to find a product that will solve this problem.\u201d

    What it\u2019s misused for:\u00a0<\/strong>
    Mistaking a topic surge for a buyer.
    <\/p><\/div><\/div>\n\n\n\t\n\tIntent data has moved from a niche signal to a default line item in enterprise B2B. That ubiquity is the problem, because not all intent signals are created equal.

    Not all intent signals are created equal. Someone hitting your pricing page three times is a completely different creature from someone whose employer happens to fall inside a \u201ccloud computing\u201d topic surge. Your tool will score them similarly. Your SDR will treat them identically.\u00a0

    The first gets a reasonable call. The second gets a breathless email that reads like it was written by someone who\u2019s never met a human being, which (in today\u2019s AI-driven world) is a likely possibility.

    Intent signals also decay fast. If your sales SLA is measured in days instead of hours, the signal was cold before you finished your stand-up. Despite this, most companies respond to poor conversion on cold signals by buying more<\/em> intent data. Inevitably, the cycle continues.<\/p><\/div>\n\t\n\t<\/div>\n\n\n\t\n\t\t\t\t\t\n\t\t\t\n\tPillar 3: Behavioral data, \u201cThe What\u201d<\/h2>\n\tWhat people actually do. Logins. Feature usage. Purchase frequency. Data imports. API integrations configured. The empirical, auditable trail of fingerprints.

    What it\u2019s good for:<\/strong>\u00a0
    Almost everything that matters in the short-to-medium term. Things like churn, renewal, expansion, and conversion. The stuff your team needs to report on.

    What it\u2019s misused for:\u00a0<\/strong>
    Explaining why any of it is happening.
    <\/p><\/div><\/div>\n\n\n\t\n\tThe foundational principle of behavioral science is simple: the best predictor of future behavior is past behavior. A SaaS trial user who logs in three times a day converts at a dramatically higher rate than one who logs in once and ghosts\u2014and neither their demographic profile, nor their firmographic fit, nor the blissfully positive intent form they filled out at signup will change that outcome.

    Behavior also exposes what customers haven’t admitted to themselves. In 2010 a Target statistician, Andrew Pole,     showed he could score a shopper’s likelihood of being pregnant<\/a> from about 25 products (like unscented lotion or calcium, magnesium and zinc supplements) and estimate a due date within a narrow window.\u00a0

    There was no survey involved. The tell was in the basket.
    Once the model worked, Target began mixing unrelated offers into the mailers to disguise how much it knew. The behavior was a better predictor than anything a customer would have told them, and it was accurate enough to be a problem.

    The same logic applies far less dramatically every day. The user who just integrated your API and imported their entire team’s data is extremely unlikely to churn next quarter, even if they gave you a 4 on your NPS survey and typed a one-star review. In this example, the integration represents a switching cost they have chosen to pay. The survey score is an opinion they will have forgotten by Thursday.\u00a0

    Action overrides words every time.

    But behavior alone traps you in a reactive loop. Your analytics will tell you that 80% of users drop off at step 3 of onboarding\u2026 but they\u2019ll say absolutely nothing about whether they quit because the copy is confusing, the price is hidden, or the product simply isn\u2019t worth it.\u00a0

    It\u2019s also worth noting that behavioral data has one more hard limit: a product that does not exist yet has no behavior to observe.\u00a0

    This is exactly the trap the launch failure rate lives in. With no behavioral signal available, teams fall back on stated intent (the least reliable data type) and they size supply chains with it. The way out is not to ask better questions. It is to manufacture a small amount of real behavior and measure that instead, which is what the fake door test below is for.<\/p><\/div>\n\t\n\t<\/div>\n\n\n\tMost companies over-rely on just one type<\/h2>\n\tIf triangulating these three data types is so obviously superior, why do most organisations cling to one? Three reasons:<\/p>\n\n\t
  • The silos are load-bearing.<\/strong> Marketing owns attitudinal. Sales owns intent. Product owns behavioural. Each team sits in a different tool, on a different dashboard, with different KPIs, reporting to different VPs who are quietly competing for headcount. Telling a business that they need to \u201cbreak down silos\u201d is the corporate equivalent of telling a divorcing couple to just communicate better: It ignores the fact that the entire incentive structure is designed to pull them apart.<\/li>
  • Every team picks the data that flatters it.<\/strong> Marketing gravitates to attitudinal and top-of-funnel intent because that includes values like impression counts and engagement rates. Product gravitates to behavioral because feature adoption validates the roadmap. Sales clings to pipeline stage because it\u2019s the metric that maps to their comp plan. No team voluntarily chooses the data that says their last twelve months were a waste of money.<\/li>
  • The problem with Customer 360.<\/strong> Enterprises spend mind-boggling sums on these platforms, expecting them to be solutions capable of spitting out definitive insights right out of the box. They do not. Aggregation is not synthesis. Without a governance rule that says \u201cwhen behavior contradicts stated preference, behaviour wins,\u201d you haven\u2019t built a predictive engine. You\u2019ve built a confusion machine.<\/li><\/ul>\n<\/div>\n\t\n\t<\/div>\n\n\n\tHow to actually fix your data (and why it\u2019s hard to do)<\/h2>\n\tThe fix is organizational, which is what makes it so hard to accomplish.

    Map each data type to the right stage of the journey.<\/strong>\u00a0
    Attitudinal at the top, where you’re shaping category associations before anyone is buying. Intent in the middle, identifying who’s in-market and when. Behavioural at the bottom and beyond, the only reliable signal once someone is actually using the product.

    When signals conflict, behaviour wins. Always.<\/strong>\u00a0
    Make it a written rule. If anyone on your team objects, overrule them.

    Stop asking and start testing.<\/strong>\u00a0
    This is the answer to the new-product problem: when no behaviour exists yet, create some. If you want to know whether customers will use a new feature, build a fake door (a non-functional entry point that tracks clicks) and measure it. Click-through data from people who don\u2019t know they\u2019re in a study is worth more than every survey response you\u2019ll ever collect.

    Kill the insight PDF.<\/strong>\u00a0
    Insights trapped in a quarterly deck are already dead. Your systems should automatically flag when behavioral decline coincides with competitor-research intent on the same account and route a contextualized alert to CS this week, not summarise it for the QBR three months from now.<\/p><\/div>\n\t\n\t<\/div>\n\n\n\tThe TL;DR<\/h2>\n\t\u201cCustomer insights\u201d are three things, answering three different questions:<\/p>\n\n\t
  • Attitudinal tells you why<\/em><\/li>
  • Intent tells you when<\/em><\/li>
  • Behavioral tells you what<\/em><\/li><\/ul>\n\nIn isolation, none of these is a reliable reporter of consumer behaviour. But together, they triangulate toward something approaching truth.\u00a0

    A version of this article was <\/em>    originally published <\/em><\/a>on Medium.<\/em><\/p>\n<\/div>\n\t\n\t<\/div>\n\n
    \n\tON OUR MINDS<\/span><\/h2>\n\tVIEW ALL INSIGHTS<\/a>\n\t\n\t\t\n\t\t\t
    \n\t\t\t\n\t\t\t\t\t\t\t\t\"\"\n\t\t\t<\/picture>\n\t\t\t\t\t<\/div>\n\t\n\t
    \n\t\t
    \n\t\t\t
    \n\t\t\t\t<\/svg>\t\t\t<\/div>\n\n\t\t\t\n\t\t\t\tInsight\t\t\t<\/span>\n\t\t<\/div>\n\n\t\t\t\t\t
      \n\t\t\t\t\n\t