2026 has certainly been a very busy year so far.
Now we’re into Q4 there has been enough activity (to say the least!) across the market research, insight, analytics and measurement space to see some clear 2026 themes.
In part 1 of a 2 part blog series, we look at 5 key areas:
In part two, we'll look at 'AI as a consumer channel that brands need to understand and measure ', what M&A tells us about where businesses see value, and a review of 2026 funding activity across the sector.
A year ago plenty of people were googling what "Agentic" was. In 2026 agentic launches and upgrades have come thick and fast, across a wide range of use cases.
Some are focused on helping researchers make sense of existing information. E.g. GlobalData has continued developing Ava into a more capable AI research analyst, able to work through multi-step questions.
Stravito's Deep Research Agent is aimed at complex questions involving large amounts of material, while Pulsar's Saga brings an agentic approach to social intelligence.
Others are moving further into actually running research. Bilendi's Discuss can manage, analyse and report on multi-country projects in one workflow. Forsta has added agents covering tasks such as survey setup, verbatim analysis and reporting.
Outset Agent goes further across the process, allowing researchers to create studies, configure recruitment, launch fieldwork and analyse findings from environments including Claude and ChatGPT.
Then there are products looking to connect different stages of research. Voxpopme's Compass can look across existing research, identify gaps and launch new research to fill them.
Bolt Insight's - Bolt Intelligence brings qual, quant, UX research and social listening into one workflow, while Fuel Cycle's Research Harness is designed to use AI across the research lifecycle.
We're also seeing more specialist applications, with NIQ's launching Optiq it's 'insight agent' and Ai Palette's Arya aimed at food and CPG innovation.
AI is moving from answering one question at a time towards systems that can complete several connected tasks towards a business outcome. In some cases that means working across different tools and platforms, with agents increasingly interacting with other agents!
The more work AI takes on the more human judgement matters - we’ve heard phrases like "human in the loop" or "human oversight" - a human still needs to know whether the research question is right, whether the method makes sense and whether the output stands up to client and stakeholder scrutiny.
AI-moderated qualitative research isn't necessarily the same as a fully agentic workflow but it's clearly part of the same wider AI change in the industry.
Ipsos launched AI-Augmented Moderation globally at the end of April, under its Creative|Labs brand, combining human expertise with AI for same day quant and scaled qual on early-stage ad evaluation.
In May Conveo combined MaxDiff with AI-moderated qualitative probing, so researchers can see not only which options people preferred but why.
In June Attest added Explore, bringing AI-moderated interviews to its quant platform and QuestionPro followed with ListenAI.
The format is also moving beyond one-off interviews too. In July Conveo launched StoryLines, a continuous AI-led qualitative programme designed to track how views change over successive waves. Outset has gone in a similar direction with Diary Studies, using its AI moderator across repeated sessions rather than a single interview.
That move towards continuous research is one we're following. A lot of traditional research is still built around a project that starts, finishes and produces an output. Some of these tools make research more persistent, with repeated interviews and successive waves building up a picture over time.
If I'd written this article at the start of the year, MCP probably wouldn't have had its own section. Now it feels like another business announces an MCP integration almost every week.
Model Context Protocol (MCP) gives compatible AI applications a standard way of connecting to external tools and data.
In research terms, that could mean accessing research data, launching fieldwork or interrogating previous studies from an AI environment (e.g ChatGPT, Claude etc) you already use.
GWI were an early mover in this space launching Spark MCP on Salesforce AgentExchange in June 2025 so Agentforce could pull consumer insights directly from GWI's data. GWI followed with MCP connections to ChatGPT and Claude in November 2025. Audiense were also early launching an MCP product in July 2025.
Since then there have been ~50+ launches, expansions and previews in 2026 so far. Some of the earlier research names were PureSpectrum, QuestionPro and YouGov.
More recently we've seen Voxpopme, Meltwater, Caplena, Rep Data, HFS Research and Dscout. Cint and Potloc are developing a connection, SurveyMonkey has launched a ChatGPT plugin and Euromonitor has made Passport available through Microsoft 365 Copilot.
Rep Data's Research Desk MCP Server is another interesting example - researchers can manage fieldwork through natural-language prompts, while keeping its existing fraud checks.
What's potentially interesting here isn't just the number of connections being launched, but what this could mean for how clients access research & data.
If a client can access research data, ask questions and potentially launch fieldwork through an AI assistant they already use, do they still need to log into the research supplier's platform?
Synthetic respondents, personas and digital twins have continued to attract attention throughout 2026.
We've seen some really interesting products launched from NewtonX, StatSocial, Human8, Outset, Zappi and others. YouGov launched Parallax, which combines AI-generated simulation with validation from real consumers.
Zappi's Synthetic Idea Pre-Screener sits earlier in the innovation process, helping teams prioritise ideas before committing to deeper research.
Much of the discussion around these synthetic tools is how reliable they are.
Burke published its FAR Framework for evaluating synthetic data quality and decision reliability, and THG Fluently has launched synthetic market research validation services.
A Columbia Business School team tested LLM-built digital twins across 19 pre-registered studies and 164 outcomes. The twins showed lower within-group variation than the real people they were modelling in 154 of them.
Real people disagree, contradict themselves and behave in ways that can be hard to model. If a synthetic audience gives answers that are too consistent, that needs to be understood before anyone relies on the findings.
The businesses that build credibility here will probably be the ones that are clear about where synthetic approaches are useful, how they're validated and where real human research still has a role.
Data quality and survey fraud is something that's a key talking point for the industry. The old saying "Shit data in, shit data" out still applies.
Data Quality Co-op introduced its Data Trust Score earlier in the year. Rep Data has kept investing in fraud detection and data cleaning.
Sample Ninja has integrated Verisoul, and Sago launched SagoVerify in September. aytm also made an interesting move publishing a Data Quality Benchmark Report covering 1.1 million survey attempts on its own panel.
As automation makes research easier to produce, clients need more confidence in where the underlying evidence came from.
We're likely to see more suppliers competing on speed, however being able to show that the data behind the answer is reliable may end up mattering just as much.
Summary
The pace of launches in 2026 has been pretty incredible.
There are obvious opportunities to make research faster, more accessible and more continuous there are also important questions about quality, reliability and where human expertise fits.
In part two, we'll look at AI as a consumer channel brands need to understand and measure, what M&A shows about where businesses see value, and a review of 2026 funding across the sector.
Hiring in the insight & data space, or thinking about your next move? Feel free to get in touch with me at james@talenthatch.co.uk
--
👋 I’m James Webb the founder of Talent Hatch.