The modern research workflow
Editor's note: This sponsored article is based on findings from “The Modern Research Workflow: How AI Will Transform the Research Tech Stack in 2026,” Marvin’s guide featuring interviews with research leaders from Microsoft, Bentley University, Netwrix and other organizations on modern research workflows, operational friction points and the evolving role of AI in research.
R
esearch teams are under pressure from every direction. They’re asked to do more than ever. Move faster. Support more stakeholders. Adopt AI. Prove impact.
But there's a growing gap between what's expected of research teams and the reality of how most research workflows operate. Many teams are still working with fragmented tools, manual processes and workflows that weren't designed for today's pace of product development.
The result? You finished a report on Friday night. By Monday, stakeholders already made the call without your research. The insights were good, but the timing wasn't.
This is one of the most common breakdowns in modern research workflows.
AI can help solve this, but only if you know where to apply it.
That tension is exactly what Marvin's new guide is built around. “The Modern Research Workflow: How AI Will Transform the Research Tech Stack in 2026” maps five hidden frictions that slow most research teams down, as well as a seven-stage workflow to address them.
The problem with "just add AI"
The instinct (and pressure) to automate everything is understandable. But one of the five hidden frictions the guide identifies is over-reliance on AI itself. Automating too much (especially during the interview and analysis stages) risks losing the human nuance, creativity and connection that make research worth doing.
"The human in the loop is really important, especially for qualitative research that can be very subjective," said Janelle Estes, experience design platform director and faculty lecturer at Bentley University.
Good AI adoption is about removing friction from the workflow and keeping researchers focused on the work that requires their human experience.
The 5 hidden frictions (and why they compound)
The compounding effect of all five frictions is the real problem. When recruitment runs slow, fieldwork gets delayed. Delayed fieldwork compresses analysis time. Rushed analysis produces shallower synthesis. Shallower synthesis erodes stakeholder trust and makes it harder to get buy-in for the next project. Each friction feeds the next.
AI, applied at the right moments across the workflow, can interrupt that chain before it runs its course.
The guide identifies five operational friction points that slow most research teams down in their workflows:
- Research silos that fragment data across departments.
- Workflows that can't keep pace with agile product teams.
- Insights that get buried in reports or arrive after decisions have already been made.
- Over-automation that strips out the human judgment that makes research credible.
- Recruitment bottlenecks that slow down research and create delays across the workflow.
The 7 stages of the ideal research workflow
The guide maps a seven-stage workflow from scoping the problem through implementation and impact tracking. Each stage gets a concrete before/after: what the manual version looked like, where AI changes the equation and which tools are worth considering.
The synthesis stage is a good example. Before AI, many researchers were building a new Word document for every interview, then manually constructing an Excel matrix to map themes across respondents. Now, AI handles thematic coding in minutes.
"It's literally saving me days of work," said Morgan Koufos, former lead UX researcher at User Interviews.
Partner with AI as a junior colleague
The researchers getting the most value treat AI like a partner. It’s useful for delivering a first draft, pushing your thinking or pressure-testing a screener for bias. But like a junior colleague, it needs direction and context from someone who can spot when the output is off.
"What I found with my team was that we had to learn how to iterate on prompts to get the quality of output we wanted," said Lauren Nitta, director of pricing strategy and market research at Netwrix Corporation. "It was a new skill we had to develop."
Teams that invest in these collaboration skills find they can do more meaningful work, faster.
Download the guide
Every research team has workflow friction. The challenge is knowing where it's slowing you down the most and where AI can have the biggest impact.
“The Modern Research Workflow” covers the complete seven-stage workflow, the five hidden frictions and a framework for building a future-ready research tech stack.