Work in your browser
Open CSV, Excel, or SPSS data. Compare groups, explore findings, and export tables and reports. Source rows stay on your device.
Open your dataSurvey analysis.
Clear answers.
Your way.
Turn survey responses into clear comparisons, statistical checks, and editable reports. Work in your browser or bring your AI. Free to use, with no Crosstabs account.
Or open your own CSV, Excel or SPSS file CROSSTABS MCPEXAMPLE
“Do customers recommend us more often than prospects?”
Brand trackerRecommendation by audience
SURVEYCUSTOMERS70% recommend
PROSPECTS35% recommend35 percentage points lower
THE READOUT
Customers lead by 35 percentage points. Check the sample, survey design, and uncertainty before generalizing.
Your analysis. Your choice.
Start with a working sample or your own survey. Keep the same focus on clear tables, checked numbers, and reports you can edit.
Work with your AI
Connect a compatible MCP client and ask in plain language. Use aggregate counts, or consent to temporary hosted uploads with 24-hour workspace access.
See how to connectFree access is subject to capacity limits. See limits · Browse calculators · 简体中文
Build on your work. Keep the context.
Make a comparison today, then carry the setup, decisions, and evidence into the next question.
Ask more without starting over.
Save a tab book with reusable question sets, audience banners, and filters. Keep table order and analysis choices together so the next run starts with a plan.
Explore reusable tab booksSee the size of a finding, and what supports it.
Read counts and percentages alongside statistical tests, effect sizes, and diagnostics. Check sparse cells, missing responses, and sample bases before drawing a conclusion.
Find the right statistical testKeep weights and multiple answers meaningful.
Use supported, explicitly defined weights and review raking balance and effective-base diagnostics. For multi-select questions, distinguish percentages of people from percentages of responses. Weighted inference requires a supported weight type or approved survey design.
Compare waves with fewer hidden surprises.
Compare compatible tables across saved waves. Detect schema drift, review changes to variables and categories, and approve reversible repairs. A matching schema still needs your judgment about question meaning.
Understand wave comparabilityPick up where you left off, with a record of why.
Saved projects retain analysis definitions, revisions, and audit history. Export a project package for reuse; include respondent data explicitly when you need a full-data handoff.
Explore local projects and portabilityBring open answers and outcomes into the picture.
Manually code open-ended responses into reviewed themes and apply them as project variables. Explore binary outcomes with an unweighted logistic model and holdout evidence. Predictions describe associations, not causal effects.
The right route for your data.
Browser and AI workflows share analysis foundations, but their controls, tools, and data boundaries differ.
Browser
Work visually with tables, saved projects, and exports on your device. Manual open-end coding and the outcome studio are browser workflows.
Open the browser workspaceLocal MCP
Install the local server for an AI-led project workflow: reusable tables, weighting, wave checks, reports, and audit history. It does not reproduce every browser control. Your AI client has its own data policy.
Read the local setup guideHosted MCP
Connect remotely for aggregate calculations or consent to file-based project analysis in a temporary workspace. Uploads leave your device; workspace access lasts 24 hours. Download your work before expiry.
See hosted connection detailsFor exact tools and requirements, see the MCP guide, capacity limits, and data policies. Browser coding and outcome-studio controls are not promised through MCP.
Ask in plain language. Get the full picture.
Not another dashboard to learn. A question, a table, and a result you can explain.
YOU ASK
Do customers recommend us more often than prospects? Show me the difference and what I should check before drawing a conclusion.
RESPONDENTS200 total100 customers + 100 prospects
DIFFERENCE+35 percentage points70% vs. 35% recommendation
RECOMMENDATION RATEUnweighted example
The difference is clear in this sample. 70 of 100 customers recommend the brand, compared with 35 of 100 prospects. This describes the observed responses, not a causal effect. Check sampling, weighting, independence, and uncertainty before making a wider claim.
Illustrative data, not a live AI response. Counts: customers 70 yes / 30 no; prospects 35 yes / 65 no.
Try these prompts
Tab books, wave comparisons, weighting, and report prompts need a project in local MCP or a consented hosted workspace. Available tools depend on your connection.
From survey data to something you can share.
From a clear comparison to an editable report, keep the numbers and the context together.
| Audience | Yes | No | Base |
|---|---|---|---|
| Customers | 70 | 30 | 100 |
| Prospects | 35 | 65 | 100 |
Bases, counts, and assumptions stay visible.
Tables you can explain
Compare audiences with counts, percentages, sample bases, and the method notes behind each result.
ILLUSTRATIVE REPORTReports you can make your own
Export editable PowerPoint (PPTX) and Word (DOCX) reports. Refresh compatible generated regions while preserving edits outside them and supported table styling. The browser refreshes the open table; changed table dimensions require regeneration.
Explore editable reportsBring survey intelligence into your AI.
For compatible MCP clients. Setup and availability vary by app.
Bring your own agent.
One endpoint. No Crosstabs account. Add it to a client that supports remote MCP.
Or connect manually with MCP
- 1
Open your agent's MCP settings
Use a client with Streamable HTTP support. Remote connector availability depends on your app, version, and plan.
- 2
Add Crosstabs
Add a remote MCP server or custom connector, then paste the URL. No Crosstabs sign-in or API key is needed to connect.
- 3
Choose how to use your data
Use aggregate counts, or explicitly consent to a temporary hosted workspace for files. Hosted uploads leave your device; access expires after 24 hours.
- 4
Ask a question. Keep the evidence.
Compare groups, build tables, and export results. Keep your workspace token private and download your work before access expires.
CONNECTION DETAILS
Crosstabs
- NAME
- Crosstabs
- URL
https://mcp.crosstabs.com/mcp- AUTH
- No sign-in to connect
These are setup instructions, not a one-click directory installation or a claim of client certification. Full connection guide
Open a temporary workspaceYour choice of data boundary. Hosted survey/report files: 5 MiB each; project files: 8 MiB; cumulative uploads: 10 MiB per workspace. Access expires 24 hours after creation; active cleanup and provider backup retention are separate. Prefer local MCP or the browser workspace when uploads are not permitted. Privacy · All limits
Survey analysis, explained.
Where should I start?
Try the browser sample to explore a working survey, or open your CSV, Excel, or SPSS file. New to cross-tabulation? Start with the crosstab guide and worked examples.
Do I need an account or an AI subscription?
No Crosstabs account is needed. The browser works without an AI client. For MCP, your chosen client may require a particular plan or version. Free Crosstabs access is subject to capacity limits.
Does my survey leave my device?
Browser analysis runs on your device. Local MCP runs locally, but anything shared with an AI client is subject to its policies. Hosted MCP file workflows require consent to upload and provide 24-hour workspace access. Read the privacy policy for cleanup and retention details.
Can I update a report after editing it?
Yes, for compatible Crosstabs PPTX and DOCX files. Refresh replaces selected generated content while preserving edits outside it and supported equal-shape table formatting. It does not preserve manual changes to generated values being refreshed. See refresh requirements and exceptions.
How do I know which results to trust?
Review the base, missingness, weight meaning, assumptions, and diagnostics alongside the result. Statistical significance does not establish causation. Explore statistical methods and validation evidence and limitations.
Can I use every browser feature through MCP?
No. MCP offers its own tool catalog and project workflows. Manual coding and outcome-studio interactions described here belong to the browser. Check the connection guide for your route, or visit support for help choosing.
READY WHEN YOU ARE
One product. Your way.
Explore a browser sample or take Crosstabs into your AI.
Prefer to keep files on your device?Install the local MCP