The questions researchers ask.

The full set of crosstabs for a study: every question run against every banner point (age, market, household and so on), with weighted bases, nets and significance letters. In FlexTabs the tab book is live — recut it, chart it, export it, ask it questions.

Yes — crosstab, cross-tabulation and crosstabulation software all describe the same thing: tables of one survey question broken by another. FlexTabs builds them from your tab spec, weighted and significance-tested, and adds charts, exports and an AI layer on top.

b³ programs and fields it. The Wincross tab spec becomes your live book automatically — every question, banner point, and weight already mapped the day fieldwork closes — so there is no setup and nothing to rebuild. Raw data and custom tables are yours to build on top.

The book exists the moment fieldwork closes — it was built from the tab spec while the survey was still in field. The honest bottleneck becomes your banner plan, not the software.

Yes — weighted bases run through the entire book, and the toolbar says so on every table. Significance tests run on the design-corrected base.

Column tests against your banner points at the confidence levels your study specifies — set per project, anywhere from 80% to 99.9%, and usually two at once, shown as uppercase and lowercase significance letters inside the cells. Weighted studies test on the effective base. The book arrives already tested — there is no separate “sig run”.

Banners and filters are drag-and-drop with real AND/OR logic, and the label writes itself — “East Canada AND Male”, not a nest of parentheses. Custom tables let you compose exactly the view a client asks for.

No — and this is the product’s spine. The engine does the math; the AI narrates. Every AI answer and every Stories card carries a verbatim machine stat line and a deep link to the exact table and row. The AI quotes figures; it never invents one.

A ranked, narrated feed of the survey’s strongest findings. It profiles every table, scores every finding for interestingness × reliability, corrects for false discovery (Benjamini–Hochberg, q ≤ .05), applies effect-size floors — and reports what it deliberately did not say, instead of padding the feed.

Yes — FlexTabs speaks MCP (early access). Point Claude, Claude Code, or any MCP-capable agent at your workspace with a scoped API key, and it queries the same engine the app uses — so your AI cites weighted, sig-tested, table-traceable numbers instead of inventing them. That’s the point of the split: the AI talks, the engine counts, whoever’s asking.

Low-base findings are flagged on the card itself, not in a footnote — and results that are “significant” but too small to matter are held back by the effect floor.

The whole tab book as formatted Excel (sig letters included) and the findings as PowerPoint — charted in your theme. Nothing is retyped on the way out.

FlexTabs runs on Microsoft Azure. Access is organization-scoped with user roles and permissions, secrets live in Azure Key Vault, and activity is captured in an audit trail.

Book a demo. We’ll run it on a live book — tables, charts, chat, and Stories — and answer your team’s follow-ups in the room, because that’s the fastest way to judge it.