What to Ask: 5 Ways to Get Started with MCP for Your Research Repository

Imagine you've just joined a new research team. There is a vast amount of research in the repository. Hundreds of interviews, usability tests, surveys, and notes. Everyone tells you, "We've probably researched that already," but nobody knows when or where.

‍

Instead of manually searching projects or asking around, Model Context Protocol (MCP) lets you ask questions about studies conducted in Optimal and instantly surface the evidence you need.

‍

Here are practical ways UX researchers can use MCP with Optimal to understand past research, accelerate new studies, and uncover insights across their repository.

‍

1. Get up to speed on past and current research

‍

One of the best ways to use MCP is understanding what's already known. Instead of combing through different studies, ask MCP to summarize existing knowledge before planning your next study.

‍

Try asking:

• Based on the research I’ve run in Optimal, what are the biggest UX opportunities for our product?
• Summarize the key findings from checkout research over the past year.
• What usability issues have been identified most frequently?
• What research should I read first to understand this project?

‍

2. Define your next research study

‍

Before creating your next study, writing discussion guides or recruiting participants, check what questions have already been answered and which gaps remain.

‍

MCP can help identify opportunities for follow-up research and prevent unnecessary duplication.

‍

Try asking:

• What questions about account creation are still unanswered?
• What themes need further investigation?
• Based on previous studies, what should our next usability test focus on?
• What hypotheses should we validate next?

‍

3. Find supporting evidence faster

‍

Whether you're preparing a presentation or writing a report, MCP can help to surface quotes, observations, participant metadata, and findings in seconds.

‍

Try asking:

• Find participant quotes describing frustration during onboarding.
• Show examples of navigation issues from recent usability tests.
• How many participants completed this study on mobile?
• Which sessions mentioned difficulty finding pricing information?
• Pull task completion rates from all prototype tests in the Dashboard project as a CSV.

‍

4. Discover research before starting from scratch

‍

One of the easiest ways to waste research effort is repeating work that's already been done. Use MCP to explore what's already in your repository before creating a new study.

‍

Try asking:

• What research already exists about navigation?
• Have we previously tested this feature?
• What have we already learned about search?
• Which studies relate to account settings?

‍

5. Identify patterns across multiple studies

‍

The biggest insights often emerge when you zoom out. Instead of reviewing studies individually, MCP can synthesize findings across projects to reveal recurring themes, behaviours, and pain points.

‍

Try asking:

• What pain points appear consistently across checkout studies?
• Compare findings from our last five usability tests.
• What themes have become more common over the past six months?
• Which usability issues keep appearing regardless of product area?

‍

6. Create stakeholder-ready summaries

‍

Research is most valuable when it's easy to share. Use MCP to turn large volumes of research into concise summaries or visualizations tailored to your audience.

‍

Try asking:

• Summarize this quarter's most important customer insights.
• Create an executive summary for leadership.
• Create a pie chart with a breakdown of onboarding studies by study method. 
• What are the three biggest opportunities we should prioritize?
• Write a summary suitable for our product team.

‍

7. Bring research into your existing workflows

‍

Connect it with the AI tools and platforms your team already uses so research becomes part of everyday decision-making.

‍

Examples include:

• Ask research questions and post insights directly into Slack.
• Create a new page in Notion summarizing research findings.
• Draft insight summaries for Google Docs and Confluence.

‍

Best practices for getting the best answers from MCP

‍

Like any AI assistant, the quality of the output depends on the context you provide. A few simple habits can make a big difference.

‍

Start with a clear goal

‍

Rather than asking broad questions, explain what you're trying to achieve. Instead of Tell me about onboarding.

‍

Try: I'm planning a usability study on onboarding. What problems have previous research uncovered that we should investigate further?

‍

Narrow your search when appropriate

‍

Large repositories can contain a wealth of research.

‍

Specify:

• Study tool and/or project
• Research method
• Time period
• Team
• Participant segment

‍

For example: Summarize usability studies about checkout conducted during the past 12 months.

‍

Decide whether you need one study or many

‍

Sometimes you need detailed findings from a single study. Other times you're looking for patterns across dozens of studies. Tell MCP which perspective you want.

‍

Ask follow-up questions

‍

Treat MCP like a research partner rather than a search engine. For example:

‍

• Can you show supporting participant quotes?
• Which studies contributed to this finding?
• Are there conflicting findings?
• What evidence supports this recommendation?

‍

Tell MCP how you want the answer

‍

Different audiences need different outputs.

‍

Ask for:

• Bullet-point summaries
• Executive briefings
• Presentation-ready insights
• Charts
• Tables
• Research reports
• Action items
• Product recommendations

‍

MCP helps researchers spend less time hunting for information and more time generating insights that move products forward. 

‍

How will you use MCP with your Optimal data? Whether you're uncovering past insights, planning new studies, or connecting research with the rest of your tools, we'd love to hear how you're putting it to work.

Walk into any product company and you'll hear the same language about being user-centric. So why does the average person still rage-quit apps weekly?  If everyone's truly putting users at the center, something's not adding up. Most companies aren't lying when they say they're user-centric. They genuinely believe it. Leadership believes it. Product teams believe it. The values are probably painted on the office wall. But there's a massive gap between claiming to be user-centric and actually building products that work for users.

‍

‍

‍

Where user insights go to die

‍

‍

Let's trace what actually happens to user research in most organizations. A researcher conducts a brilliant study. They talk to real users, identify genuine pain points, uncover surprising insights about how people actually use the product. The researcher creates a comprehensive report with clear recommendations. They present to stakeholders and then…nothing happens.

‍

Product moves forward based on roadmap commitments made three months ago. Design tackles the projects already scoped. Engineering builds what's already been spec'd and by the time the next planning cycle rolls around, the research is old news. The context has shifted. New priorities have emerged. The research was great. It just never actually changed anything.

‍

The issue here isn't that teams don't value research. It's that the systems aren't set up to close the gap between insight and implementation. Research lives in one place and product decisions happen somewhere else. There's a handoff required. Someone needs to take the insights from the research and translate them into the specific contexts where decisions get made. But who owns that handoff? Usually nobody explicitly. The researcher assumes product managers will incorporate the findings. Product managers assume researchers will flag critical issues. Designers are caught in the middle, trying to advocate for users while also hitting delivery dates. So insights leak at every transition point.

‍

Even when research does make it into decision-making conversations, it's often in the wrong format for action. Researchers are trained to present thorough findings and all of that is valuable. But it's not always actionable. The gap between what users said and what should be done about it, is a lot bigger than most teams realize, but an even bigger problem is that product decisions can’t wait for perfect information. Product decisions need to happen fast, often in real-time, in contexts where pulling up a research report isn't practical. If research insights aren't immediately accessible in the moment decisions are being made, they effectively don't exist.

‍

‍

‍

So…What actually works?

‍

‍

The companies that successfully put users at the center do something different. They build research into the decision-making flow rather than treating it as a separate activity that informs decisions. They make sure that research insights live where decisions happen, integrated into project management tools, design files, and product specs.

‍

Increasingly, they're also using MCP (Model Context Protocol) to connect AI tools like Claude, ChatGPT, and Cursor directly to their research repository – here's how. Instead of searching through reports, teams can ask questions in natural language to surface relevant findings, user insights, and context exactly when they need them, or generate artifacts and workflows grounded in real user research. Insights are accessible in the moment someone needs them.

‍

At Optimal, our design team embeds research findings directly in their Figma files. Not as links to external reports, but as context right next to the designs. When someone's reviewing navigation options, the tree test results are right there. When they're looking at prototype variations, the usability data is embedded in the file. Decisions and insights exist in the same place.

‍

Another great step is to make sure that all research insights are framed as recommendations, not just observations. Instead of "Users struggled with the checkout flow," the insight could be "Simplify checkout to 3 steps instead of 5, based on usability testing where users abandoned at step 4." The researcher isn't just reporting what happened. They're translating it into clear next steps.

‍

We also firmly believe that research should be continuous instead of episodic. Instead of big studies that take weeks and produce comprehensive reports, teams run smaller tests constantly. This doesn't replace deep foundational research. But it supplements it with rapid feedback loops that keep user input flowing into active work.

‍

And lastly - research should be accessible to all. When only dedicated researchers can conduct studies, research becomes a bottleneck. When product managers and designers can run quick tests themselves, with appropriate guidance and quality standards, research becomes embedded in how the team works rather than something teams have to request and wait for.

‍

Being genuinely user-centric requires infrastructure, not just values. You need tools that make it easy to share findings where people actually work. You need workflows that integrate research into product development rather than treating it as a separate track. You need the ability to run research at the speed decisions happen.

‍

Most importantly, you need to make the gap between insight and action as small as possible. The best research in the world doesn't help users if it never influences what gets built.

You've invested time, budget, and effort into your research. But when you or someone else needs an answer, finding the right insight often means searching through projects, remembering which study covered the topic, and piecing insights together manually.

‍
The Model Context Protocol (MCP) changes that. It connects AI tools like Claude, ChatGPT, and Cursor securely to your research in Optimal, so your team can ask questions in plain language and get insightful answers or connect your research to AI-powered workflows.

‍
Below, we cover what MCP makes possible and how to connect it, so you finally get full value from the research you've already done and the repository you’ve built.

‍
Getting started: What can MCP do for research teams?

‍
Once connected, MCP turns your repository into something you can ask directly, so all that past and current research is always instantly accessible. Here are 5 ways to use it:

‍

1. Research assistance

‍

Ask "What usability issues have we found recently?" or "What have we learned about onboarding?" Pull participant data and metadata e.g. “How many participants took this study on mobile?”

‍

2. Discovery

‍

‍Ask "What research already exists on navigation?" so work isn't duplicated. You can also use MCP to pull quotes or review transcript data. “Surface participant quotes that highlight points of friction when navigating the homepage.”

‍
3. Executive summaries

‍

Ask your AI tool to summarize the most important themes from research this quarter or format findings into charts or graphs.

‍
4. Cross-study synthesis

Surface recurring participant pain points across multiple usability studies at once.

‍
5. AI assistants & workflows

Connect Optimal research with the tools and workflows your team already integrated with your AI tools, like Zapier, Slack, Jira, and Notion.

‍
What can you ask? Real questions, by research method

‍
Here are some practical examples of the kinds of questions teams can ask.
‍

Across studies

• What themes appear across multiple usability studies?
• What are recurring participant pain points this quarter?
• Which studies were conducted around onboarding in the past year?
• Summarize all checkout-related findings from studies this quarter.
  ‍

Interviews

• Can you summarize the key pain points for participants who have downloaded and used the mobile app?
  ‍

‍Prototype testing

• Which task had the lowest success rate in the latest prototype test, and what usability issues contributed to it?
• What usability issues contributed to task failure?
• Pull task completion rates from all prototype tests in the Dashboard project as a CSV.

‍
Tree testing‍

• What % of users found the checkout successfully in last week's tree test, and where did the rest drop off?
• Where did users drop off?

‍
Card sorting

• Which categories did participants consistently group together in the navigation card sort?

‍
First-click testing

• Where did users first click when asked to find the Pricing page in the first click test?

‍

‍
How do you set up MCP with Optimal?

‍
‍
Step 1: Sign in to your AI tool
Log into your preferred AI assistant (e.g. Claude, ChatGPT, or Cursor).
‍
Step 2: Connect your Optimal account
Go to your AI tool’s settings and add a new MCP connection. Authenticate your Optimal account via OAuth 2.0 to securely grant access to your Optimal data.
‍
Step 3: Start asking questions
Return to your AI tool and begin with simple, high-value questions grounded in your research.

‍
Step 4: Embed it into your workflow‍
Use MCP regularly to explore insights, synthesize findings, and support decision-making.
The most effective MCP implementations are not standalone tools; they are embedded into daily decision-making.

‍
If your AI tool is already connected to tools like Slack, Jira, Notion, or Zapier, you can use your Optimal research to trigger workflows, such as:

‍

• Sending Slack alerts when key findings are uncovered
• Creating tickets in Jira when usability issues are detected
• Feeding insights into product documentation tools
• Connecting findings to internal AI assistants used by product and design teams

‍
You've already done the hard part: running the studies and capturing the findings. The value is sitting in your repository. MCP helps you unlock what's already in your repository, making it easy to discover, reuse, and turn into action.

‍

Whether a study was conducted yesterday or months ago, you’ll be able to gather insights with MCP to make faster, more informed decisions today.