Example · Market Research Analyst
Production work is leaving your job. Deciding what gets studied is not yet your title.
20 of your 40 hours still need you
- 14 stay yours
- 6 faster with AI
- 20 tools could take
A typical week at the middle pace. Your plan uses your own tasks.
Is market research a dying career?
By our task-by-task count, 20 of a typical 40-hour week as a market research analyst still need a person. Tools could already take about 20 hours. Production work is about half of your week, and a model does it today. It is also the half that justified hiring you.
Read more
The industry's own vendors now sell survey programming, tabulation, the tracker refresh, the deck and the competitive scan. Your profession also faces a rare threat, because the respondent can be synthesized too. Qualtrics started selling prebuilt synthetic panels this spring, and a consumer brand was among the first to test them.
What neither the model nor the synthetic panel can do is the part people already come to you for: knowing whether a sample can answer the question, turning a brand manager's email into a study that will change a decision, and sitting in a backroom while a brand team hears a consumer call the concept confusing. That work is often not yet your title, and it is exactly what an Insights Manager posting describes. The plan is about making it your title before the brand teams learn to run the synthetic panel themselves.
The hard truth
Online opt-in samples can mislead you.
Their answers can swing results, especially for young adults. Checking the data before you share it protects your work and your name.
Sources (2)
- In a Pew test, 12% of opt-in respondents under 30 claimed a license to operate a nuclear submarine. Pew Research Center, 2024-03
- An opt-in poll found 20% of under-30s called the Holocaust a myth; Pew's probability survey found 3%. Pew Research Center, 2024-03
What stays yours
Telling a brand manager why the survey in their email will not answer their question
This rests on how hard mistakes are to catch. A bad study returns a confident number and nobody finds out until the launch underperforms. Catching it before fielding needs someone who has watched a leading question produce a false winner.
Turning "should we launch the spicy line?" into objectives, method, sample and a questionnaire architecture
This depends on context. The model can write a concept test, but it does not know that sales already promised the retailer a Q2 slot, that R&D has two viable formulas, or that the last test failed because the sample skewed to current buyers.
The backroom and the same-day debrief
This needs presence and context. AI-moderated one-to-one interviews are live products (Outset, Conveo). A live group with the brand team behind the glass, and you reading the room and re-ordering the guide on the fly, is a different job.
Saying "that move is inside the margin of error" to a VP
The accountability here is borrowed. Someone must stand behind the number when a brand team wants to act on noise. Today you likely do it as the defendant, after the fact. The plan changes when and who initiates.
Knowing which past study already answered the question
What protects this is context that won't last long. It is worth a lot because nobody wrote it down. In Week 7 you write it down on purpose and trade the task for the credit.
What's next for a market research analyst?
Plan A is your best move inside the work you do now. Most sessions in the 13-week plan build toward it.
- A
Research lead for one brand
Why it fits
Why you. If you already tell brand managers why their surveys fail, want to be deciding what gets studied rather than building what someone else decided, and the seat above you is open or thin, this is the move.
The gap. You have no charter and no ledger of studies you designed and decisions they changed. Your method breadth (conjoint and MaxDiff, two ways to measure trade-offs, and pricing) is thinner than an Insights Manager posting asks. You also need to handle the politics of telling a brand director "that isn't a research question."
The trade-off. You depend on one brand director, and you will be arguing, in writing, for tools to do the work you were hired for. That is the right argument, and it still costs you. You get fewer studies, bigger bets and more visible mistakes.
Other moves that fit
These are also good bets in this field. Read them if Plan A doesn't suit you.
Own the admissibility gate for synthetic and AI-moderated methods
Why it fits
Why you. If synthetic scores worry you because they look right, that is the job. Many companies have no rule for when a synthetic result may reach a retailer or a claim.
The gap. You need the parallel-test evidence from Weeks 4 and 8, a careful read of the Bisbee and Park papers, and a ledger of where synthetic agreed and disagreed with humans on your own categories.
The trade-off. At a mid-size company this is a half-role, so it only works paired with the research lead move. You will be the person who says "not yet" to something the CFO likes. And if the parallel tests come out well, you will have documented, under your name, why the company needs fewer human studies.
Qualitative moderator (the hedge)
Why it fits
Why you. If you already moderate when a senior colleague can't, and the backroom is a part of the job you enjoy, the craft is within reach.
The gap. You need the craft. Burke's four-afternoon workshop or RIVA's four-day course, both moderator training programs, gives you the frame; fifty groups give you the skill. At a handful of rounds a year in-house, that is years.
The trade-off. The middle of the qualitative research market is going to AI-moderated one-to-one interviews now (Outset, Conveo and Remesh all sell it), so what remains is the high end, which is supplier-side and relationship-driven. It pulls you toward a smaller employer and away from the decision room. This is the plan if the research lead and method gate moves stall by Week 8.
What's a plan B for market researchers?
Plan B is a job AI is creating that these skills already fit. One session a week in the plan trains you for it, so it is ready if your job shrinks or you want a change.
BPlan B, picked for this role
AI evaluation specialist
Catching a leading question before fielding is the same instinct that decides whether an AI output, or a synthetic panel, can be trusted.
Other options
AI value and ROI analyst
You already tell a VP when a move sits inside the interval, and AI investments need someone who will say that about their claimed returns.
AI product manager
Turning a vague launch question into objectives, method and sample is product discovery, and AI features need people who can frame the right question.
In your own plan you can switch to any of the 47 roles.
The first two weeks
The first 2 of 13 weeks. Most weeks have three sessions of 30, 60 and 60 to 90 minutes, plus one 40-minute Plan B session. Weeks 12 and 13 have two sessions plus Plan B. Each session ends with something you can show. By week 13 you'll have a tracker memo with thresholds the brand teams agreed to, written rules for when synthetic results can be used, and a published AI test write-up.
- Week 1: Measure your week
What you do
- ALog how you spend your time for one week. Sort each hour under one of the 14 tasks on your Pathcast task list.
- AList every study from the last two years. For each one, write the question, the decision it was for and whether that decision changed. This is your study ledger.
- AWrite down every time this year you explained why a survey would not work. Use one line each.
- BTake the first half of AI Fluency: Framework & Foundations, Anthropic's free course. Ask AI to draft a short made-up survey, then mark every flawed question.
- Week 2: Test AI on survey building
What you do
- AAsk your Director whether the AI copilot or the survey platform's AI may touch respondent-level data and exports from syndicated data, the reports your company buys.
- AHave the approved tool build this month's concept-test survey from the brand brief. Count leading, double-barreled and logic errors using the checklist in Questionnaire Design by Marsh and Bolton.
- AIf the rules allow it, have the tool build last quarter's tracker deck from the data tables. Time it, and list what it could not know.
- BFinish AI Fluency. Download a free Pew Research Center dataset and pick one question to explore.
Three courses from the plan
Checked 2026-09-22.
- Bisbee, Clinton, Dorff, Kenkel & Larson, "Synthetic Replacements for Human Survey Data? The Perils of Large Language Models"Political Analysis, 2024 (doi 10.1017/pan.2024.5); free preprint at osf.io/5ecfa · Free (open access)
Yes, Week 4. It is the clearest evidence that synthetic panels get the mean and miss the spread.
- Park et al., "LLM Agents Grounded in Self-Reports Enable General-Purpose Simulation of Individuals"arXiv 2411.10109, v3 June 2026 · Free
Yes, Week 4. It is the strongest case for the other side, so know it before someone quotes it at you.
- FTC Policy Statement Regarding Advertising SubstantiationFTC · Free
Yes. Week 9.
Risks beyond AI: moderateBudget risk
Research budgets are discretionary and get cut with marketing budgets in a volume downturn. Agency retainers are often the first to go. Consumer-goods office headcount has been contracting on its own, before synthetic panels shipped. Brand-side research budgets are rising overall, but the money is flowing to platforms rather than analyst seats.
Where this could be wrong
- The shares are guesses until Week 1. If your log shows far more deck production than we assumed, more of your week is automatable, the pressure on entry-level jobs is higher than 40, and the research lead move becomes urgent.
- Treating survey programming as automatable rests on vendor claims, not a benchmark. Your Week 2 count is the number that matters. If you find four faults per twenty questions, AI only helps with the task rather than doing it, and your seat is safer.
- The synthetic evidence is moving. Bisbee tested ChatGPT in 2023; Park's v3 is June 2026. If a vendor shows segment-level spread that matches your human waves on your categories, the method gate narrows fast.
- We assumed a mid-size company with no MBA brand program. At a P&G or General Mills the brand-management door is real and our decline of it is too strong.
Also matched: market research analyst, consumer insights analyst, insights analyst. Research checked 2026-09-22. Page updated .