Market Research and Consumer Insights Case Studies
Marketing5 Case StudiesUpdated Sep 2026
These five cases come from working problems in consumer insights: picking one growth bet with limited money, tracing a recurring failure in survey quality, settling a dispute over research priorities, testing whether reported activity reflects real outcomes, and weighing a short-term gain against exposure that arrives later. Each case follows the same four lenses, business objectives, available evidence, operational constraints, and stakeholder impact, because most insight work goes wrong on one of them before method is ever the problem. The organizations are fictional. The judgments inside them will look familiar to anyone who has had to present a finding that nobody in the room wanted to hear.
Case Study 1
Choosing one growth bet from three promising options
Larkspur Outdoor, a regional maker of camping cookware with about 140 staff, two warehouses, and a four-person insights team
The situation
Larkspur Outdoor had three growth options on the table at the annual planning review. The first was a refill subscription for compact fuel canisters, which had drawn about 2,300 sign-ups on a voluntary waitlist from existing customers. The second was a trail-running cookware line aimed at specialty shops, a channel where the company had steady sales but little direct contact with buyers. The third was a mid-priced skillet range for a large online marketplace, where competitors were growing quickly. The discretionary budget covered one full launch and a small pilot of a second. Each option had a senior champion, and none of the champions had seen the others' numbers.
The challenge
The decision had to be defended to finance in the fourth week, and the evidence was uneven. The waitlist was real but self-selected, so it set a floor rather than a forecast. The shop interviews numbered eleven, and none of those buyers had placed a firm order. Marketplace demand could be estimated from search volume, but the company had never sold through that channel and had no cost history to lean on. A new channel would also pull two people from the warehouse team for most of the season.
Analysis
The first task was to restate the objective in terms finance could check. Larkspur wanted contribution margin growth within the coming year, not revenue alone, and that changed the comparison. Teams grounded in Market Research and Consumer Insights usually begin by writing down what the business needs to learn before they open a spreadsheet, and here that meant asking which option could add margin with the cash and people available. The subscription had the strongest unit economics on paper but needed fulfillment capacity the warehouse lacked. The marketplace had the largest apparent demand, but fees and returns thinned its margin fast.
The evidence was weakest where the money was largest. The waitlist was useful for direction but could not predict how many people would pay for a recurring offer, so the team treated it as a floor. The shop interviews pointed to real demand for lighter pans, yet buyers described purchase cycles tied to seasonal orders, which made a first-year revenue estimate fragile. Dana, the head of product, noted that the marketplace estimate rested on competitor listing counts, not on any sales Larkspur had made. Stakeholder impact cut both ways: the warehouse lead would absorb any surge, and the accounts manager had promised two specialty buyers a product within the year.
- Week oneWaitlist, shop interviews, and marketplace search data pulled together.
- Week fourFinance reviews margin assumptions and fulfillment capacity.
- Season startSubscription launches; shop pilot begins with six accounts.
- Mid-seasonRepeat-order rate checked against the margin threshold.
- Season endPilot sell-through and marketplace hold reviewed together.
Actions taken
- Run a paid pilot of the trail cookware line with six specialty shops for one season, with a fixed order minimum and a sell-through target agreed before shipment.
- Fund the subscription as the primary launch, but cap the first cohort at the number the warehouse could pack without overtime, and measure repeat orders at the end of the second month.
- Hold the marketplace option for review, and ask for a two-week test of listing fees and return rates using a small set of existing products before any inventory is committed.
- Brief the warehouse lead and the shop accounts manager together, with the capacity numbers in front of them, so that promised timelines are set against the same constraints the decision used.
Results
Larkspur funded the refill subscription as its primary launch and ran the shop pilot alongside it. By the end of the season, repeat orders on the subscription cleared the margin threshold, though growth was slower than the waitlist had implied. The shop pilot sold through about two-thirds of its first order, enough to justify a wider release next year but not the full range originally planned. The marketplace option stayed on hold. The written record of assumptions and observations shortened the next planning cycle considerably.
Questions for discussion
- Which of the three options would have looked strongest if the team had only used the evidence that was easiest to collect, and what would that have cost the business?
- How much evidence is enough when the budget allows one full launch, and who should decide that threshold?
- What would have to change in the warehouse plan before the marketplace option could move to the front of the queue?
Case Study 2
Tracing a rise in flagged survey completes
Corvane Field Insights, a fifteen-person fieldwork firm that runs online consumer panels and quota surveys for grocery and household brands
The situation
Corvane ran online surveys for grocery clients, and over several recent waves the share of completed interviews removed by quality checks had climbed. The usual rate sat near 5 percent. In the most recent waves it reached about 14 percent, and in three studies the re-fielding pushed client reports back by a week. Quotas filled on schedule, which hid the problem from the operations dashboard the team checked each morning. The panel vendor, the survey programming group, and the field operations team each suspected a different cause, and each had a reason to think so. Clients had started asking whether the toplines could be trusted.
The challenge
The team needed to map the path from invitation to cleaned data, find which stage produced the bad completes, and propose corrective actions that could be measured within two waves. The programming group could not freeze the questionnaire mid-field without losing a client's trend line. The panel contract was up for renewal in the next quarter, and any change to sourcing needed sign-off from the vendor relationship lead. Two analysts were available for the review, and clients were waiting on answers.
Analysis
The business objective was simple to state and hard to protect: keep client reports on schedule and keep the numbers credible. The evidence was the flag log for six waves, with each removed complete tagged by reason, but the tags had been applied by a person reading a summary rather than by the system at the moment of capture. The team mapped the process from panel invitation to screener, through the main survey, to the cleaning rules. Teams working through a problem like this often find the cause sits in a handoff rather than the instrument. The log showed mostly speeding and straightlining, but not where those respondents had come from.
The stakeholder picture mattered as much as the data. Account leads were fielding questions about revised timelines, and the programmer had been blamed in two review meetings for a problem that started upstream. The team also had to avoid a fix that protected quality by quietly biasing the sample, since a tighter screen could remove the very shoppers a grocery client most wanted to hear from. Each proposed change therefore had to be tested against both the flag rate and the demographic mix of the completes it kept. That meant any change needed a before-and-after comparison, not just a lower number.
Actions taken
- Split the flag rate by panel source, device type, and questionnaire length for the last six waves, and share the table with the field and client teams in the same week.
- Pause invitations from the source with the highest flag rate until a fresh validation sample passes, and track that source's share of completes each week against a limit of 20 percent.
- Add a minimum time threshold and a straightlining check at the screener so poor completes are caught before quota counts them, with a target flag rate under 6 percent within two waves.
- Hold one shared weekly review with the vendor, the programmer, and the accounts lead, working from one dashboard with a named owner for each measure and a dated fix log.
Results
The flag rate fell from about 14 percent to 6 percent across three waves. Most of the improvement traced to one panel source, which was paused, re-validated, and returned to use only after its respondents passed a fresh check. The screener changes removed a further slice of speeders without measurably shifting the age and household mix of completes. Re-fielding delays ended, and two client reports went out on schedule the following month. Mobile layout still produces more flags than desktop, so the next task is a redesign of the grid format.
Questions for discussion
- If the panel source had not been the main driver, which of the proposed actions would have carried the most risk to the client trend lines?
- How should a fieldwork firm weigh a tighter screen against the risk of removing the respondents a client most needs to hear from?
- What evidence would convince the vendor that a quality problem is its responsibility rather than a design flaw the firm should fix?
Case Study 3
Negotiating one research priority across two teams
Pell Ridge Foods, a mid-sized maker of shelf-stable sauces with two plants, national grocery distribution, and a nine-person insights group
The situation
Pell Ridge had a single research budget for the year and two teams who wanted it spent differently. The brand team, led by Marcus, wanted a quarterly tracker that could inform packaging and promotion decisions within a few weeks of fieldwork. The insights group, led by Ines, wanted a larger segmentation study that would take most of the year to field and analyze but would reshape how the company understood its shoppers. Regional sales managers asked for store-level data that neither plan included. The commercial director had approved the overall budget but had not said how it should be divided.
The challenge
The disagreement had reached the weekly operations meeting, and each side had started building its case in writing. Marcus saw the segmentation study as slow and abstract, while Ines saw the tracker as a set of numbers that would be read without context. Finance wanted a decision before the next budget cycle, and packaging changes had to be signed off in the third week of the planning period. Neither team had yet said what decision its research actually needed to support.
Analysis
The first move was to separate positions from interests. Marcus's position was a tracker, but his interest was a packaging decision he could defend without waiting a year. Ines's position was a segmentation study, and her interest was a shared picture of who buys the sauces and why, which the tracker could not supply. The evidence was thinner than either side admitted. The last tracker had covered only two regions, and the segmentation method had never been tested on the company's own shopper panel. Teams with a shared objective often find both sides need the same decision supported.
Constraints narrowed the field quickly. The budget covered either a full segmentation study or four tracker waves with a light segmentation module, not both at full scale. Sales had no budget line and would need a cost-sharing arrangement for store-level cuts. Decision rights were the real gap: the commercial director owned the budget, but the steering group had no written authority over scope. Stakeholder impact fell hardest on the regional managers, who would act on whichever numbers arrived first, and on the plant schedule, which could not wait for a late study.
- Clarify interests Each team states the decision it needs to support, not the method.
- Map constraints Budget, panel timing, and plant calendar set hard limits.
- Confirm decision rights Commercial lead decides scope; steering group reviews results.
- Test the agreement Pilot one quarter, then review whether the trade still holds.
Actions taken
- Ask each team to write the single decision its research must support, in one sentence, and bring those sentences to the steering group before any method is discussed.
- Agree a split of roughly three parts tracker to one part segmentation, with the light module built on the tracker sample so that the two outputs can be read together.
- Put the commercial director's approval in writing, naming the steering group as owner of scope changes and the regional managers as the only recipients of store-level cuts.
- Set a review point for the tracker after two waves, where the packaging team reports whether the numbers changed a decision, and reallocate the segmentation budget only if they did not.
Results
The steering group agreed a split in the third week, with the tracker as the main spend and a light segmentation module built on its sample. Sales received store-level cuts for two regions under a shared cost line. Marcus had packaging numbers in time for the spring review, and Ines kept a segmentation question the tracker alone could not answer. The review point after two waves turned up one packaging decision the tracker had shaped, which gave both teams a reason to trust the arrangement.
Questions for discussion
- Which of the two original positions was really a proxy for a different interest, and how would you test that before the meeting?
- When decision rights are unclear on paper, who should be asked to settle scope, and what should happen if that person prefers one team's position?
- How would you design a review point that gives both teams a fair chance to show the value of their approach without rewarding the most recent result?
Case Study 4
When survey volume hits target but follow-up stalls
Calder Home Supply, a regional home improvement chain with forty-two stores and a six-person central customer insights team
The situation
Calder's insights team measured store performance through a monthly customer survey sent after each purchase. The team had met its completion target in every month of the past season, with about 6,400 responses arriving in a typical month. Yet the share of surveyed customers who had a follow-up action logged by a store manager had fallen steadily, and two regional managers had started asking why the survey was being run at all. Store managers saw the survey as a reporting task rather than a source of work, and the central team saw completion counts as proof the program was working.
The challenge
The team needed to decide which indicators deserved a place in store targets, which were noise, and how to design a review cycle that would catch a drift like this sooner. Targets were set at the chain level, but the data came from stores that coded follow-up actions in different ways. Store staff had limited time for a second review, and the survey vendor could change question wording only at the start of a quarter. The regional managers wanted something they could explain to store leads in one meeting.
Analysis
The objective was clear on paper: improve customer outcomes at the store level. The indicators were not. Completion counts were accurate, since the vendor's system captured every submitted form, but they measured effort rather than response. The follow-up action rate looked closer to the outcome, yet store coding varied, and some managers logged a call as an action while others recorded nothing. The team asked whether the drop was real or a recording artifact, and the answer depended on which store codes had changed recently. Without a fixed definition of follow-up, the indicator could not separate a store that was improving from one that was simply recording less.
Constraints shaped what was feasible. A rewrite of the store coding list needed a two-week window for training, and managers had already been asked to take on the monthly store briefing. The vendor could change survey wording only at quarter start, so any fix to the outcome question would wait. Stakeholder impact was uneven: store managers felt measured on activity they could not control, regional managers were losing confidence in the program, and the central team had reported a success that the store floor did not recognize. The review cycle had to be light enough that managers would keep using it, and specific enough that the next drift would show up within a month.
Actions taken
- Freeze one written definition of a follow-up action, with three permitted codes and no free text, and retrain store managers in a single session before the next survey wave goes out.
- Retire completion count as a target and replace it with the share of responses that received a logged action within five days, reported by store and by region.
- Add a monthly audit of twenty randomly chosen store records, checking whether each logged action matches what the customer was told, so the indicator is tested against a second source.
- Run a quarterly review in which the central team, two regional managers, and a rotating store lead compare the indicator with the audit results and agree one change to the process each time.
Results
Within two survey waves, the logged-action rate had settled into a steady line instead of falling, and the audit showed that roughly a fifth of logged actions did not match what customers had been told. The team tightened the coding guidance, and the reported rate moved up. Part of that rise was real improvement and part was better recording, and the audit was what separated the two. Regional managers kept the survey in the program and asked for the quarterly review to replace the monthly targets meeting. Store managers said the three-code list was quicker than the old form.
Questions for discussion
- Which of the indicators in this program would you keep if the team could report only one, and what would you lose by dropping the others?
- How much weight should an audit sample carry when it is small compared with the full set of store records?
- What would a store manager need to see in the review meeting to believe the indicator reflected their work rather than their paperwork?
Case Study 5
Weighing a quick revenue gain against consent exposure
Quill Harbor Analytics, a boutique consumer research firm of about sixty staff that runs app-based panels and rapid concept tests
The situation
Quill Harbor ran a consumer panel recruited through a mobile app, and its participants had agreed to answer product surveys for research purposes. A national beverage client asked for a rapid concept test using panelists' past purchase behavior, matched to a loyalty dataset the client already held. The contract would have been the largest the firm had signed that year, and it would have paid for the panel refresh the team had been postponing. The consent language, however, covered product research only. Resale or sharing of individual-level data with the client's advertising partners was not described.
The challenge
Rosa, the managing director, needed a recommendation before the client's review in the second week, and the team disagreed about the risk. Felix, who ran panel operations, argued that the panel's trust was the real asset and that one misuse would drain recruitment for a long time. Adaeze, the compliance lead, said the current wording would not hold up if a participant asked what was done with their data. Sales wanted the account, and the client had threatened to move the work to a larger vendor.
Analysis
The business objective was to grow revenue and protect the panel, and on a short horizon those goals pointed in opposite directions. The evidence favored the client's plan on speed and fit. A concept test run on past purchase data would give a sharper read than a fresh survey, and the matched data would have made the result unusually specific. But the evidence on consent was clear, and it counted against the plan. The original wording did not describe sharing, so using participants' data that way would be a use they had not agreed to, whatever the commercial logic. Options had to be judged on that basis first.
Operational constraints were tight. Re-consent would require a new notice in the app, a review by the privacy adviser, and a fresh invitation to panelists, which the operations team estimated would take most of the quarter. Stakeholder impact reached beyond the firm. Participants who had shared purchase habits in good faith would be affected if their data moved, and the client's own shoppers would be implicated if the match was later found to be unauthorized. The team also had to think about the next client. A panel known to have been used for resale would lose recruits, and the reputational cost would fall on every study that followed.
Actions taken
- Decline the matched-data use as written, and tell the client in plain terms what the consent language allows, so the conversation starts from the same facts on both sides.
- Offer a hybrid pilot: a small re-consented group, with a clear notice describing any sharing, running a concept test on fresh responses while the full matched study waits for review.
- Rewrite the consent notice before any further client work, with separate choices for product research and data sharing, and have the privacy adviser sign off on the wording.
- Set a written rule that matched or shared data requires panel-level approval and a participant opt-in, and record each request in a log the managing director reviews each month.
Results
Quill Harbor declined the matched study as proposed and ran a smaller pilot on fresh responses from re-consented participants. The client moved part of the work to a larger vendor, and the firm lost the largest line item in the proposal. The re-consent process took most of the quarter, and recruitment slowed for a time. Two later clients asked for the same consent structure, and the new notice became something the firm could show prospects rather than a cost it had absorbed. The panel grew more slowly than planned, without the complaints a resale would likely have produced.
Questions for discussion
- If the client had agreed to pay a premium for the matched data, which parts of the decision would have changed, and which would have stayed fixed?
- How should a research firm weigh the trust of its panel against the revenue of a single large account when the two conflict?
- What wording would have made the original consent notice adequate for matched use, and what would participants have needed to see first?
Learning Resources
Marketing CASE STUDIES
China, India, United States, Indonesia, Pakistan, Nigeria, Brazil, Bangladesh, Russia, Japan, Mexico, Philippines, Vietnam, Egypt, Turkey, Germany, United Kingdom, France, Italy, Canada, South Korea, Australia, Spain.