{"id":3792,"date":"2026-09-30T00:00:00","date_gmt":"2026-09-29T22:00:00","guid":{"rendered":"https:\/\/humes.pl\/consumer-insight-how-to-derive-it-from-data-rather-than-team-intuition\/"},"modified":"2026-09-24T15:36:55","modified_gmt":"2026-09-24T13:36:55","slug":"consumer-insight-how-to-derive-it-from-data-rather-than-team-intuition","status":"publish","type":"post","link":"https:\/\/humes.pl\/en\/consumer-insight-how-to-derive-it-from-data-rather-than-team-intuition\/","title":{"rendered":"Consumer insight: how to derive it from data rather than team intuition"},"content":{"rendered":"<p>At a creative workshop, someone says: &#8220;our customers want to feel appreciated.&#8221; Everyone nods, the statement goes into the brief as a <strong>consumer insight<\/strong>, and three months later no one can say whether the campaign based on it worked, because the assumption was never testable. The difference between such a wish and an insight derived from data is that the latter is grounded in measurement, identifies a specific mechanism or tension, and can be tested in subsequent research.<\/p>\n<h2>What is a consumer insight and how does it differ from an observation?<\/h2>\n<p>A <a href=\"https:\/\/humes.pl\/en\/glossary\/consumer-insight\/\">consumer insight<\/a> is an interpretation that explains why people behave the way they do, particularly when their behavior contradicts what they declare. An observation says what is happening. An insight indicates the non-obvious logic that may underlie it.<\/p>\n<p>A practical example from the FMCG category. Assume that 63% of respondents say they read product ingredients, but in a shelf test, the average time spent engaging with the packaging may not be enough to read the full list of ingredients. These are two facts that require interpretation. An insight hypothesis might be: consumers do not always read ingredients to understand them in detail &#8211; sometimes they check them to confirm a decision made based on a visual shortcut, while the statement &#8220;I read ingredients&#8221; may also be an identity statement (&#8220;I am a conscious consumer&#8221;), rather than a literal description of every purchase behavior.<\/p>\n<p>This difference has operational implications. An observation describes a state at a particular point in time. An insight describes a presumed mechanism, so it may help predict a response to a stimulus that has not yet been tested. This is why a <strong>marketing insight<\/strong> may retain value longer than a single report, provided it remains relevant in a given market context.<\/p>\n<p>Three elements that often distinguish an insight from a well-phrased statement:<\/p>\n<ul>\n<li><strong>Tension<\/strong> &#8211; an insight often describes a conflict: between declared attitudes and behavior, between two needs of the same consumer, or between a social norm and actual motivation. A statement without tension may be a description rather than an insight.<\/li>\n<li><strong>Grounding in data<\/strong> &#8211; it should be possible to identify a specific questionnaire question, transcript excerpt, or observation result from which the insight was derived.<\/li>\n<li><strong>Testability<\/strong> &#8211; it should be possible to design a measurement that could challenge the insight. If no result can be imagined that would contradict it, the proposition may be a truism or an overly broad interpretation.<\/li>\n<\/ul>\n<p><\/br><\/p>\n<p>The statement reversal test can serve as an auxiliary check, but it is not in itself evidence of validity. Reversing &#8220;customers want to feel appreciated&#8221; gives us &#8220;customers want to feel disregarded&#8221; &#8211; an absurd statement, so the original provides little information. &#8220;Consumers read ingredients to confirm a decision, rather than always to make one&#8221; can be contrasted with an alternative proposition: &#8220;consumers read ingredients primarily to make a decision.&#8221; Both propositions are meaningful and testable. This indicates that the original proposition does in fact distinguish between alternatives.<\/p>\n<h2>How can you formulate a consumer insight step by step based on data?<\/h2>\n<p>The process of deriving an insight from data follows a specific sequence. Skipping any stage increases the risk that the team will revert to its own beliefs and selectively support them with a quote from the research.<\/p>\n<ol>\n<li><strong>Collect declarative and behavioral data, where possible.<\/strong> Declarations come from questionnaires, IDIs, or FGIs. Behaviors come from transaction data, analytics, ethnographic observation, eye tracking, shelf tests, return data, and complaints data. Combining both layers is particularly useful because it makes it possible to compare what consumers say with what they do.<\/li>\n<li><strong>Identify discrepancies.<\/strong> Look for instances where the declared importance of an attribute does not translate into its weight in a model explaining choice, where declared usage frequency differs from sales data, or where the declared reason for opting out does not correspond to the point at which the process was actually abandoned.<\/li>\n<li><strong>Code qualitative material for language, not just themes.<\/strong> In <a href=\"https:\/\/humes.pl\/en\/content-analysis-in-qualitative-research-how-to-turn-hundreds-of-responses-into-decision-making-insights\/\">content analysis in qualitative research<\/a>, metaphors, justifications, and moments of hesitation can be valuable. The statement &#8220;well, you know, it just sort of happened&#8221; in an interview about canceling a subscription may signal that the respondent does not have a ready rationalization &#8211; and that is precisely where it is worth probing further.<\/li>\n<li><strong>Formulate an explanatory hypothesis.<\/strong> Use one sentence, in the consumer&#8217;s first person or in the third person, but always with a verb that describes a mechanism: &#8220;avoids,&#8221; &#8220;puts off,&#8221; &#8220;justifies to themselves,&#8221; &#8220;treats X as evidence of Y.&#8221;<\/li>\n<li><strong>Test the hypothesis using data that were not used to generate it, where the study design allows.<\/strong> This is particularly important when the insight is intended to inform a high-stakes decision. If the hypothesis emerged from twelve IDIs, it can then be tested on a quantitative sample, in an experiment, or using an independent set of behavioral data.<\/li>\n<li><strong>Record the conditions that would challenge it.<\/strong> For each insight, it is worth noting what result in a subsequent measurement would weaken or challenge the proposition.<\/li>\n<\/ol>\n<p><\/br><\/p>\n<p>From this perspective, an insight is not a data point, but an explanation of the tension between what customers declare and what they actually do &#8211; or between other observed phenomena. It should also be testable in a subsequent measurement. If the team cannot identify a question or observation that, in the next wave of research, would make it possible to assess whether the insight still holds, it has probably formulated an opinion rather than a research finding.<\/p>\n<p>It is worth paying attention to the structure of how an insight is documented. An insight written as a single sentence is convenient in a presentation, but it loses context. In project documentation, a four-part format works well: observation (what is visible in the data), tension (what does not align), explanation (why this may be the case), and test (what could challenge the insight). This format also makes it easier to retain context when the team changes.<\/p>\n<p>An example from financial services. Observation: users of a banking app declare that the spending analysis feature is highly important, while usage data show that they access it rarely and briefly. Tension: high declared value alongside low usage. Explanatory hypothesis: the mere presence of the feature may serve a symbolic function &#8211; it is evidence that the bank is &#8220;keeping an eye on&#8221; the user&#8217;s finances, so actively using it is not always necessary to gain its psychological benefit. Test: if reducing the feature&#8217;s visibility does not lower app ratings or the sense of control over finances, the hypothesis about its symbolic function will be weakened.<\/p>\n<h2>What mistakes most often undermine a consumer insight?<\/h2>\n<p>Most unsuccessful insights do not result from poor data, but from rushing the interpretation stage. <strong>Research results analysis<\/strong> is often treated as a formality between fieldwork and the presentation, even though this is where the project&#8217;s value is created.<\/p>\n<p>The most common pitfalls when working on an insight include:<\/p>\n<ul>\n<li><strong>Team confirmation bias.<\/strong> The team enters the analysis with a ready-made proposition and selects quotes that confirm it. A safeguard is to document hypotheses before the analysis begins and account for them, including those the data disproved.<\/li>\n<li><strong>Confusing an insight with a need.<\/strong> &#8220;Customers need faster delivery&#8221; is a need, and one declared directly. An insight begins when the question is asked why the same group that declares speed to be a priority chooses the cheapest, slowest option in the shopping cart.<\/li>\n<li><strong>Extending an insight from one segment to the entire population.<\/strong> A mechanism identified among early category users does not necessarily apply to occasional users. An insight without an indication of the segment it concerns is incomplete.<\/li>\n<li><strong>Overinterpreting a small qualitative sample.<\/strong> Qualitative research can generate credible hypotheses, but it usually does not allow their prevalence in the population to be determined. The statement &#8220;most consumers believe&#8221; should not appear in a report based on eight interviews.<\/li>\n<li><strong>An insight without a mechanism verb.<\/strong> If a statement can be written as &#8220;consumers are X&#8221; or &#8220;consumers value Y,&#8221; it is usually an attitude description, rather than an explanation of behavior.<\/li>\n<li><strong>No validity date.<\/strong> Consumer mechanisms may change along with the category and market context. An insight without information about the source material and period in which it was developed may, over time, become an untestable belief.<\/li>\n<\/ul>\n<p><\/br><\/p>\n<p>A separate issue is the difference between an insight and a <strong>research finding<\/strong>. A finding is a structured statement resulting directly from the data &#8211; for example, that satisfaction with post-purchase service differs significantly across contact channels. An insight is one level higher: an interpretation of why this may be the case, in terms of consumer motivation. A report should include both, but these layers should not be confused, because a finding is a description of a fact, whereas an insight is an interpretation subject to the risk of error.<\/p>\n<h2>When is an insight ready to use? A checklist<\/h2>\n<p>Before including an insight in a brief or report, it is worth conducting a short review. A consumer insight is better prepared for use if the answer to most of the questions below is yes.<\/p>\n<ul>\n<li>Can a specific data source &#8211; a question, variable, or transcript excerpt &#8211; from which the insight was derived be identified?<\/li>\n<li>Does the insight describe a tension or mechanism rather than merely a state?<\/li>\n<li>Is an alternative explanation plausible and testable?<\/li>\n<li>Has the insight been tested using data other than those that generated it, or has such validation been planned?<\/li>\n<li>Have the relevant segment and the period in which it was developed been identified?<\/li>\n<li>Has a condition that would challenge it been recorded &#8211; a specific result from a future measurement that would weaken the proposition?<\/li>\n<li>Is the insight understandable to someone outside the research team without additional commentary?<\/li>\n<\/ul>\n<p><\/br><\/p>\n<p>The final point has practical significance. Even an accurately derived consumer insight loses value if it gets lost among charts in a presentation. <a href=\"https:\/\/humes.pl\/en\/how-to-communicate-research-findings-to-executives-so-the-report-does-not-end-up-in-a-drawer\/\">How research findings are communicated to senior management<\/a> affects whether the insight will be used or remembered merely as an interesting point from the report.<\/p>\n<p>Insights are more likely to withstand confrontation with reality when they are designed from the outset as hypotheses to be tested in the next wave, rather than as final truths that close a research project.<\/p>\n<h2>Frequently asked questions<\/h2>\n<h3>How does an insight differ from an observation?<\/h3>\n<p>An observation answers the question &#8220;what is happening&#8221; and describes a state visible in the data. An insight answers the question &#8220;why might this be happening&#8221; and explains the mechanism behind consumer behavior, often by identifying a tension between declaration and action. An observation may become outdated as data change, while an insight retains value as long as the described mechanism and the context in which it was identified remain valid.<\/p>\n<h3>How can you determine whether an insight is valid?<\/h3>\n<p>The basic test is to examine the insight using a data set that was not used to generate it &#8211; for example, a quantitative sample, an experiment, or behavioral data if the hypothesis emerged from in-depth interviews. A second tool is the falsification test: determine what specific result from a future measurement would challenge the proposition, and assess whether such a result could be obtained. If no condition that would challenge it can be identified, the insight may be a truism or an overly broad interpretation.<\/p>\n<h3>What types of research most often generate consumer insights?<\/h3>\n<p>Valuable insights often emerge from mixed-methods projects, where the qualitative layer (IDIs, FGIs, ethnography, consumer diaries) helps generate hypotheses about mechanisms, while the quantitative layer, experiments, or behavioral data make it possible to assess their prevalence, variation across segments, or consistency with actual behavior. Qualitative research alone can provide hypotheses without information about their reach, while quantitative research can describe the scale of a phenomenon without always explaining its causes. Transaction data, digital analytics, and point-of-sale observation add further value because they show behavior independently of declarations.<\/p>\n<h2>Ask about research that will provide insights for your brand<\/h2>\n<p>If your team needs data-based insights rather than beliefs formed in a workshop, it is worth starting with a conversation about which declarative and behavioral data should be combined in a single project. <a href=\"https:\/\/humes.pl\/en\/contact\/\">Contact<\/a> Hume&#8217;s Institute to discuss research scope tailored to your category.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>An observation from research is not yet an insight. We explain how to derive a consumer insight from the gap between what customers say and do, and how to test it.<\/p>\n","protected":false},"author":4,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"inline_featured_image":false,"footnotes":""},"categories":[912],"tags":[],"slowa_kluczowe":[],"class_list":["post-3792","post","type-post","status-publish","format-standard","hentry","category-badania-i-analizy"],"acf":[],"_wp_attached_file":null,"_wp_attachment_metadata":null,"wpml_media_processed":null,"_wpml_media_usage_in_posts":null,"_wp_attachment_context":null,"_oembed_35c905c64c03156f243b94f18c4eb80f":null,"_wp_attachment_image_alt":null,"rank_math_description":"A consumer insight explains the tension between what customers say and what they do. Learn how to derive it from research data and test its accuracy.","rank_math_focus_keyword":"consumer insight","rank_math_contentai_score":null,"_wpml_post_translation_editor_native":null,"_menu_item_type":null,"_menu_item_menu_item_parent":null,"_menu_item_object_id":null,"_menu_item_object":null,"_menu_item_target":null,"_menu_item_classes":null,"_menu_item_xfn":null,"_menu_item_url":null,"_wp_page_template":null,"rank_math_og_content_image":null,"_wp_trash_meta_status":null,"_wp_trash_meta_time":null,"_wp_desired_post_slug":null,"rank_math_primary_category":null,"_acf_changed":null,"wp_pattern_sync_status":null,"_form":null,"_mail":null,"_mail_2":null,"_messages":null,"_additional_settings":null,"_locale":null,"_hash":null,"_config_validation":null,"_wp_old_slug":null,"rank_math_internal_links_processed":"1","_top_nav_excluded":null,"_cms_nav_minihome":null,"_thumbnail_id":null,"_last_translation_edit_mode":"native-editor","_wpml_word_count":"2204","_dp_original":null,"_edit_last":null,"_edit_lock":null,"rank_math_seo_score":null,"_wpml_location_migration_done":null,"_wpml_media_duplicate":null,"_wpml_media_featured":null,"_wp_old_date":"2026-09-24","copied_media_ids":[],"referenced_media_ids":[],"rank_math_title":"How to formulate a consumer insight | Hume's Institute","job_department":null,"_job_department":null,"job_location":null,"_job_location":null,"job_offer_external_link":null,"_job_offer_external_link":null,"footnotes":null,"inline_featured_image":null,"blog_podtytul":null,"_blog_podtytul":null,"blog_czas_czytania":null,"_blog_czas_czytania":null,"blog_dalsza_lektura":null,"_blog_dalsza_lektura":null,"slownik_krotka_definicja":null,"_slownik_krotka_definicja":null,"slownik_cytat":null,"_slownik_cytat":null,"slownik_na_stronie_glownej":"1","_slownik_na_stronie_glownej":null,"slownik_slowa_kluczowe":null,"_slownik_slowa_kluczowe":null,"slownik_w_praktyce":null,"_slownik_w_praktyce":null,"slownik_powiazane":null,"_slownik_powiazane":null,"slownik_kluczowe_punkty":null,"_slownik_kluczowe_punkty":null,"lang":"en","translations":{"en":3792},"pll_sync_post":{},"_links":{"self":[{"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/posts\/3792","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/comments?post=3792"}],"version-history":[{"count":1,"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/posts\/3792\/revisions"}],"predecessor-version":[{"id":3793,"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/posts\/3792\/revisions\/3793"}],"wp:attachment":[{"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/media?parent=3792"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/categories?post=3792"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/tags?post=3792"},{"taxonomy":"slowa_kluczowe","embeddable":true,"href":"https:\/\/humes.pl\/en\/wp-json\/wp\/v2\/slowa_kluczowe?post=3792"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}