Agencies Aren’t Dying, They’re Getting an Upgrade: AI for Market Research Agencies and Seven Market Changes Happening at Once

A lot of people I speak to, inside and outside our industry, believe that the effect of AI for market research agencies will be fewer of them. It might turn out to be true. So far there is no reliable evidence for it, and it is the least interesting thing happening in market research right now. Staring at the headcount question has made a lot of agency leaders unnecessarily gloomy about changes that mostly run in their favour.

Seven significant things are shifting at once. Six are good news if you are well positioned. One is a genuine headwind, and I will not soften it. Let me take them in turn.

The headcount question is the least interesting thing happening

We have had tools like ChatGPT since late 2022, and the catastrophe predicted for us has not arrived. Entry roles have reduced, and that is a real issue I come back to below. But there has been no mass unemployment in our industry, and agencies have not gone to the wall in their thousands. AI is not an extinction event for agencies. It is a transformation event.

It helps to remember how little the core of our work had changed. Since the late 1980s, when Word, Excel and PowerPoint arrived on researchers’ desks, there has not been a major productivity tool for the centre of what we do. Fieldwork and data processing were transformed. Everything else stayed exactly as it was: somebody wrote the proposal, designed the questionnaire, built the guide, read the tables and assembled the deck. For nearly forty years, production has dominated our time and most of our mental energy.

That is what AI for market research agencies changes, and it changes it in a way no previous tool could, because for the first time the technology reaches the whole process rather than one stage of it. The end-to-end workflow, brief to debrief shows where. What follows is what it does to the market.

Seven market research industry trends happening at once

One word needs pinning down first. Production is the making of things: the methodology, the proposal, the survey or discussion guide, the screener, fieldwork management, processing and coding, first-pass analysis and the deck. It is skilled work, and it can be specified, which is exactly why a machine can now take a great deal of it. Thinking is everything else: framing the real question, choosing between defensible interpretations, connecting a finding to a commercial decision, and putting a name to a recommendation someone will be held to.

1. Production costs fall hard, and the cost base resets

The agencies I work with typically find that between 50 and 70% of billable hours are production tasks. That is our own working estimate; Forsta’s Torbjörn Andersson, writing in GreenBook and ResearchWorld in 2025, cites a benchmark of 80% of a researcher’s time spent preparing, cleaning and organising data. We hire clever researchers and arrange for most of their week to be spent on formatting.

Agencies using our tools report time savings of up to 50% across proposals, fieldwork materials and analysis. Box Clever took survey design from two days to about one hour. Your numbers will differ, but the direction is the same for everyone: a specialist AI platform permanently resets what a project costs to deliver, and therefore what one can be sold for, whether or not you move first. This is the shift everything else follows from.

2. The agency changes shape: leaner, more senior, with new choices

Falling production costs do not automatically mean fewer agencies. They mean agencies with a different shape: fewer junior production roles and a more senior mix. They also hand you capacity, and a choice.

You can do more with the team you have: more projects, more proposals, more time in front of clients. Or you can be far more selective about the work you take and specialise harder than your economics allowed. Both are legitimate. What is not a strategy is letting the freed hours refill with email, which is the default and the commonest way agencies get nothing back.

3. The value of true insight goes up

This one comes from outside our industry and runs firmly in our favour. IDC estimates around 230 zettabytes of data will be created in 2026, roughly tripling between 2025 and 2029. Gartner puts unstructured data at 80 to 90% of new enterprise data, growing three times faster than the structured kind. Most of what organisations hold is material their systems have never made sense of.

Abundance turns data into a commodity. As the data becomes commonplace, the ability to find the true insight inside it becomes rarer and worth more. Knowledge is not power any more. Only insight is, and insight is what we are in business to produce.

4. Falling costs plus rising value equals more demand

This is the part people miss, and it is the most commercially important of the seven. When the price of good research falls, more organisations can buy it. The company that could never sign off a strategic study now can. The client who ran four strategic projects a year runs eight. And because everyone is drowning in data, the value of genuine thinking rises.

The evidence supports it. ESOMAR’s Global Market Research 2025 puts the insights industry at $153bn in 2024, expected to pass $160bn by the end of 2025: growing in cash terms even as unit costs fall. The assumption that cheaper research means a smaller industry has not been true so far, and I see no reason to expect it to become true.

5. Tactical and strategic pull apart

Synthetic respondents, AI moderation and DIY platforms are turning the tactical end into a commodity: quick-turn testing, trackers, low-stakes studies. That work is getting radically faster and cheaper, and for what it is, it is good. Within the ESOMAR total, research software is now $62bn and grew 11.5%, while the established market research sector, $56bn, grew 4.8%.

The same forces push the value of strategic work up. Once anyone can produce something that looks like professional research, a validated method, real respondents and good thinking stop looking expensive and become the reason clients buy. Both ends are doing well. The squeeze lands on the undifferentiated middle, which does not mean small. It means an agency that cannot say clearly why its answer is better than a cheap one. I set out how to make that case in how agencies compete with DIY platforms.

6. Consolidation: fewer agencies, but stronger ones

Here is the headwind, stated straight. The global agency population will probably shrink. But the agencies that go will mostly be the ones that resist AI altogether, or compete with manual labour at the commodity end, where prices are collapsing fastest. AI-enabled agencies doing strategic work pick up what the departing agencies leave behind.

Consolidation has a second edge. The big network groups will chase this wave with scale economics: advanced AI, offshored production, aggressive cost-cutting. Independents should not try to win that race on size. Level the cost base with AI for market research agencies, then compete on the senior attention and strategic focus a network finds hard to industrialise. Scale is good at making things cheaper. It is not good at making a specific person available to a specific client on a Friday evening.

7. Clients take the simple work in-house

Clients have used DIY tools and in-house teams for simple research for over twenty years, and AI accelerates that sharply. Concept screens, basic trackers and quick polls will keep moving client-side. What stays with agencies, and gets more important, is the complicated, high-stakes work with multiple stakeholders, where judgement, method and independence are what the client pays for.

If your revenue is concentrated in work clients can now do themselves, this is next year’s budget conversation. Do not defend that work. Know what proportion of revenue it represents, and have a plan that does not depend on it.

The two patterns underneath all seven

The market is splitting. Tactical work becomes a commodity while strategic work becomes more important and therefore more valuable. The middle, neither cheap enough to win on price nor differentiated enough to win on judgement, is where the pressure lands.

Total demand is growing. Cheaper research plus a greater need for insight means more research gets bought, not less. The industry expands in cash terms while unit costs fall, which is what happens when the price of a thing drops and its value rises at the same time.

If you run a strategic agency, both patterns work in your favour. Your demand outlook is arguably the best in the industry. That is the future of market research agencies as I see it, and it is a long way from the gloom. Market research AI is the reason, not the threat.

What AI for market research agencies does not make safe

None of that makes you safe, because AI does not hit a strategic agency where the demand is. It hits four places that are less visible from outside.

The work that is disappearing is the work on your invoices. Production hours are the largest line, and the line about to get much cheaper. As soon as a credible competitor can deliver the same study in fewer days, “why does this take twelve days?” stops being rhetorical.

Your pricing model turns efficiency against you. Most strategic agencies price like contractors, so when market research AI halves the production days, the whole benefit passes to the client automatically. The financial case works through the fixes.

The route that made your seniors is closing. Juniors learned judgement by doing the work AI now does. PwC’s 2026 Global AI Jobs Barometer found AI-exposed entry-level roles in the US seven times more likely to require senior skills such as judgement and leadership.

Plausible has stopped being worth anything. A client’s own team with a generic AI assistant can produce a credible-looking deck in an afternoon. Plausible is not the same as warranted, and what clients will pay for is findings they can stand behind: validated methods, real respondents, a trail from brief to debrief, a named person who signed it off.

Read together, those are one problem with four faces: everything that can be produced is deflating towards free, and what is left is warranted thinking. The response is one response too. Cut the production time. Warrant the work. Price the thinking rather than the days. Rebuild how you recruit and train.

The upgrade, in one line

The old trade-off in our industry was depth against speed. You could have the senior thinking or the fast turnaround, rarely both, because production consumed the budget. AI for market research agencies ends that trade-off when it is adopted properly, with the researcher in the lead and the machine on production. The brawn goes to the machine. The brains were always the point.

Insightology is the specialist AI workflow for market research agencies: one golden thread from brief to debrief, with your method in the machine and humans in the lead. To see the upgrade on a live project, see how it works or book a demo.

FAQs

Will AI reduce the number of market research agencies?

Probably, at the margins, but the agencies most at risk are those that resist AI or compete on manual labour at the commodity end. AI-enabled agencies doing strategic work tend to pick up the work others leave behind, and total industry demand is growing in cash terms.

What is the future of market research agencies?

Leaner, more senior and more selective. Production work moves to AI market research software, which resets the cost base; the agency’s value concentrates in framing, interpretation and recommendations a client can stand behind, priced as thinking rather than days.

Which market research industry trends matter most right now?

Two patterns underlie all seven changes: the market is splitting into commodity tactical work and higher-value strategic work, and total demand is growing because cheaper research plus more data means more research gets bought.

What should an agency do first?

Count your production hours on the last ten projects, split them into production and thinking, and find out how much of your revenue sits in work clients can now do themselves. Those two numbers decide which of the seven changes hits you first.