Or why AI doesn't shrink research - it multiplies its ambition
Every team is facing the same question. What will AI do to us by 2030?
For insights teams, the fear has a familiar tone. Will we need fewer researchers? Will AI replace analysts? Will survey platforms automate us out of relevance? Will synthetic respondents erase the need for human ones? The worry is that research becomes a machine. And the job becomes obsolete.
The reality is sharper, and far more interesting. By 2030, insight teams will be smaller in one dimension: production labour. But they'll be radically bigger in what they deliver. More tests, more iterations, more triangulation, more signals surfaced, more decision cycles supported, more influence at the table. Fewer people doing mechanical tasks. Ten times the output in strategic impact.
The shift begins with automation. By the end of the decade, the production stack will be nearly fully automated. Survey writing, programming, logic checks, translations, data cleaning, open-end coding, chart building, even first-pass reporting - none of this will require much manual labour. Not because the work goes away, but because it was never the point. These tasks existed to get the insight out. They were always the constraint. Every hour spent checking logic, fixing routing, formatting slides, or merging tables was an hour not spent shaping decisions. Now, the constraint is gone. And with it, the need to wrap a team around the bottleneck. The 2030 insights team has fewer people doing this work simply because there's almost no mechanical work left to do.
But the real shift isn't just speed - it's depth. Synthetic respondents fundamentally change the economics of exploration. Privately trained models, built on your own historical data, become the first layer of every research loop. Not a replacement for humans, but a filter. An exploration engine. Instead of testing ten concepts, you explore a hundred synthetically and validate the top ten. Instead of squeezing messages into five survey slots, you can pre-triage entire universes. You can stress-test combinations of price, feature, positioning, or framing. You can run the impossible surveys - messy, complex, high-dimensional - and only send the meaningful edge cases to human respondents. Synthetic models make early-stage testing cheap and scalable. Human validation sharpens what matters. That loop gives small teams real strategic bandwidth. Five people doing what used to require twenty.
This changes the centre of gravity of the insight function. It stops being about collecting and formatting data. It becomes about helping the business think clearly. Researchers shift from production to interpretation. They become pattern recognisers, signal integrators, scenario designers, risk translators, ambiguity handlers, strategic partners. The work becomes more human, not less. Automation handles speed and consistency. Insight handles judgement and meaning. The machine takes the noise away. The people are left with the signal.
And that signal moves fast. In 2030, research isn't operating on 12-week timelines. Marketing moves weekly. Product moves in sprints. Comms can shift by the hour. Media is optimised in real time. Most of the business is already operating on modern time. Research just hasn't caught up yet. By 2030, it will. Insight becomes embedded in the rhythm of the business. Always-on brand lift. Weekly message iteration. Daily creative checks. Sprint-aligned concept testing. Crisis-response testing that runs in hours. Continuous sentiment tracking. Rolling segmentation models that adapt with the market. Research stops being a blocker. It becomes infrastructure. Quiet, responsive, flexible. Not a report - an operating system.
That doesn't mean headcount grows. The opposite. The mechanical roles shrink - survey programmers, QA specialists, deck builders, manual coders, data cleaners. The new roles multiply. Insight strategists. Narrative synthesizers. Behavioural analysts. Scenario designers. Hybrid methodologists who know when to use synthetic, and when to go human. Business translators who know how to get a finding into a roadmap. Teams get smaller in task volume, but bigger in strategic range. And that's where the value lands.
The idea of "half the size, ten times the output" isn't hyperbole. It's arithmetic. Automation makes things faster and cheaper, so you can run more cycles. More studies, shorter timeframes, tighter alignment with decision-makers. Synthetic makes things broader and deeper - you test more, earlier, across wider ranges of possibilities. You find failure modes before you spend money chasing them. You explore edges, contradictions, and adjacencies that wouldn't fit in a traditional scope. Interpretation makes insight more actionable - real-time support, early warnings, scenario planning, tighter loops between learning and activation.
The only teams that shrink are the ones that don't move. The ones that stay anchored to mechanical process. The ones that think research is still a survey followed by a deck. The ones who treat AI as a threat instead of a multiplier. The ones who believe their job is to build charts, not shape choices. Those teams shrink because the work they're defending doesn't exist anymore.
The teams that grow are already shifting. Toward synthesis. Judgement. Creative testing. Empathetic understanding. Triangulation. Scenario planning. Faster cycles. Tighter loops. Real decisions, made with clarity. These are the skills AI makes more valuable, not less. Synthetic lets you explore more. Automation gets you there faster. But interpretation is still the lever. The human job is still to make it make sense.
By 2030, the insights team won't just look different. It'll think differently. Move differently. Act differently. And earn a different seat in the room. Half the size. Ten times the output. A hundred times the influence. Not because the team got smaller. Because it finally got free to do the work that matters.