Skip to main content
An analysis plan organizes the questions, comparisons and outputs needed for your research decision. Each plan folder becomes a sub-folder in Charts. Review the plan before generating and verify the resulting bases before sharing.

Open the plan

Open Analysis → Analysis plan. On the first visit, ask the assistant to build a plan, upload an Excel analysis plan, or start from scratch. Generate one folder per survey section builds outputs using the questionnaire’s sections without saving a custom plan. Editors and Owners can change and generate the plan; readers can inspect it. A Draft study can hold a plan, while generation requires eligible responses and a suitable study state. See Generate and update analysis results.

Organize folders and questions

1

Name the plan

Enter Plan name. The generated parent chart folder is named after this plan unless you subsequently rename it.
2

Add questions and folders

Use the Add menu to add questions, a folder or a variable. Arrange questions through drag and drop, their menus or multi-selection. A question can occur in several folders with different comparisons.
3

Review each question's splits

A folder shows one line per question. Each chip is one split with its own chart/table settings. Totals identifies a split with no comparison or distinguishing override. Use Split to add another.
4

Inspect the outputs

Hover a split to inspect its comparison, base, answer edits, statistics and generated outputs. Grey chips follow inherited settings; cyan chips have their own. Click a chip to open its first generated chart, or its settings if no output exists. A chart opened here provides Back to plan.
Split settings let you switch between a question’s splits, add or remove one, or move it to another folder. The saved order keeps the question’s splits together. Questions listed under No charts remain available as comparisons and filters.

Set comparisons and the response base

Open Plan settings for defaults. A folder inherits the plan’s comparisons and can override them; a split can then override its folder. Review which level supplies the setting before changing it. A deliberately empty comparison means totals rather than inheritance. Response filters on a folder and a split restrict the plan’s base further. They do not replace the plan’s filters or lower the response threshold for open-ended analysis. Check the resulting bases for every planned comparison. Choose Unweighted or a reviewed weighting scheme and use Manage weighting to inspect its targets and diagnostics. The generation gives its charts the selected weighting. To show counts instead of percentages, use the Numbers setting; an individual split can override it.

Choose outputs and table settings

Set the output language and whether each split makes charts, tables or both. Each split can configure its base, hidden/merged answers, nets, statistics and currency. A categorical grid can have a summary with a mean score or a percentage on an answer or net; a numeric grid is tabulated by cell, with per-row statistics and range nets. The split’s statistics override the plan’s. Otherwise, the plan’s statistics appear under its chart and as rows in its table. An in-chart variable belongs to that split’s outputs and is not added to the questionnaire. See Choose chart metrics and statistics. Only keep tables with a significant difference is optional and off by default, including generation without a saved custom plan. When enabled, it narrows comparisons using the primary-level column test and raw column bases of at least 30. A tables-only split whose table is dropped gets its chart instead. Repeated-record tables that cannot be tested are retained; absence of a test is not absence of a difference. See Read statistical significance. Table settings also control cells, decimals, additional rows, confidence levels, significance letters, shading and other tests. New plans use count and column percentage, one decimal, a base row and 95% primary column tests. Enable secondary and 99% levels deliberately. Exactly two comparisons can be nested when that layout is appropriate.

Work with the plan assistant

Save or undo manual edits before asking the assistant; the chat is locked while the plan has unsaved changes. Manual plan editing is held while the assistant answers. Review proposed variables and answer scores before confirming their cards. Variables can be added to the study, and an existing variable proposal offers an update. A replaced proposal cannot be saved. Bulk creation is atomic: if the study refuses a variable, inspect the card’s reason, correct it and retry rather than assuming the other variables were saved. Bulk creation does not overwrite edits you made to existing saved variables. A saved proposal can be placed in the folder or split the assistant proposed. Wait for calculated values to finish backfilling before checking bases or generating outputs. Keep weighting-only variables out of the displayed outputs unless the analysis requires them.

Import a tabulation plan

Attach an Excel, CSV or Word plan to the Plan assistant, or choose the Excel upload route on the first visit. The assistant reads workbook sheets, cells and their layout, including settings instructions. It checks its plan against the file and reports unmapped lines or confirms that all lines are represented. Review the folder structure, comparisons, bases, nets, score cards, weighting and statistical settings against the actual file. A workbook sheet may define a folder, several sections or banner groups, or shared settings; do not assume each sheet becomes a folder. A generated output carries its uploaded source line, such as PDT!R12, under its heading and in Excel. Search that line in Charts to find the corresponding outputs. A table-only line can remain table-only when a chart would add no information. Resolve reported omissions and ambiguous mappings before generating.

Save and generate

Save draft saves without generating and can retain an incomplete plan. Undo changes restores the saved state. If saving fails, keep the page open: your edits remain there. Select Save and generate N folders, or Generate N folders when there are no pending edits. The footer estimates charts and tables; “up to” applies when significance filtering can remove tables. Theme-chart counts depend on the generated themes. Review the large-output confirmation if it appears. When a chart folder already exists, Where should the charts go? offers:
Before replacing a folder, move or duplicate anything you need to retain, or create a new folder. Editable plan charts are also replaced.
The run opens Charts. The generated folders follow plan order, including each question’s splits. They and their charts are ordinary editable library items. A failed save starts no generation; a failed run leaves the plan saved for correction and retry.

Check the result

Open the outputs from their split chips or the Charts library. Check source lines, selected measures, filters, weighting, comparisons and bases. Trace themes and findings to their answers. A missing table needs its settings and eligibility checked before any interpretation about the groups.
The plan groups questions and split chips into folders.

The plan groups questions and split chips into folders.