Prepare messy survey data for analysis
Survey exports often arrive messy—split multi-select columns, coded headers, rating grids, hidden metadata, weights, separate wave files, and columns that need combining before a banner makes sense.
This guide covers the common fixes in AddMaple for exports from Typeform, SurveyMonkey, Qualtrics, Google Forms, SPSS, Excel, and client survey vendors. It is the overview before the detailed preparation docs.
AddMaple lets you fix structure, labels, and categories in the same project you analyze, then pivot against the updated fields immediately.
Upload the raw export
You do not need a perfect analysis template before uploading. AddMaple supports CSV, Excel (.xlsx, .xls), SPSS (.sav), and exports from most survey tools.
When you load a file, AddMaple scans columns, detects likely data types, and creates an initial set of charts and tables. For more detail on upload, see Importing and preparing data.
Review what AddMaple detected
After upload, check whether columns look right before you build reports or dashboards. In Manage Column(s), review and adjust how AddMaple interpreted your data.
Common checks:
- Are long question titles readable?
- Did checkbox or select-all-that-apply questions become multi-select columns?
- Are rating or grid questions grouped together where useful?
- Are numeric codes, dates, and categories treated as the right type?
- Are metadata fields, test fields, or duplicate columns hidden from the dashboard?
Changes in Manage Column(s) affect project views, charts, pivots, and exports, but the raw source data is not modified. See Manage Columns.
Fix multi-select and split answer columns
Many survey tools export one multi-select question as several columns—one per answer option. For example, "Which brands have you used?" may become separate columns for Brand A, Brand B, Brand C, and so on.
AddMaple can often detect this automatically and show it as one multi-select variable. If not, combine the related columns manually in Manage Column(s).
This covers formats such as one column per checkbox option, 1/0 or Yes/No flags, or blank cells for unselected options.
See Multi-Select and How to combine columns.
Group rating grids and repeated scales
Survey exports often split matrix or grid questions into one column per row. For example, a satisfaction grid might have separate columns for ease of use, value for money, support, and speed on the same scale.
In AddMaple, group related items so they are easier to compare while keeping each item as its own column:
- Combine when several source columns represent one question, such as a multi-select.
- Group when several separate questions belong together, such as rating statements on the same scale.
Grouping is especially useful for Likert scales, brand attribute ratings, CX drivers, and repeated question sets. See Group columns.
Clean labels, categories, and coded values
Messy exports often contain labels that are technically correct but painful in charts—question codes like Q12_3, category values like 1, 2, 3, long repeated question stems, or inconsistent wording.
In Manage Column(s), rename columns and grouped items, change column types, and hide columns you do not want on the dashboard. Category labels are cleaned in the Legend—see How to work with legends.
For column settings and types, see Manage Columns.
Clean categories with AI
Exports frequently include spelling variants, inconsistent capitalization, near-duplicates like "Yes" and "Y", numeric codes instead of answer text, Likert scales in the wrong order, or a long tail of one-off responses that clutter charts.
Clean with AI in the Legend can standardize these in one pass. Open the legend from a pivot chart (Legend tab on the right) or from Manage Column(s) (open the column and scroll to Legend). Click Clean categories with AI.
In the dialog:
- Allow Category Merging is off by default. Turn it on for near-duplicates—for example "U.S.", "USA", and "United States", or variants of the same open-text answer.
- Use Additional instructions to guide the AI, such as "Keep age categories in ascending order" or "Do not merge 'Other' with anything else".
Click Start, review the suggested renames, order, and merges, then Apply and Save. If categories are merged, click View Mappings to see how original values map to cleaned ones.
By default, Clean with AI suggests renames and ordering only. For a smaller set of duplicates, select categories in the legend and click Merge instead.
For one-off fixes, edit the legend directly. See Clean with AI, Merging categories in the legend, and How do I clean or recode response options?.
Handle header rows correctly
Some survey platforms export more than one header row—question codes in one row, full question text or answer labels in another.
AddMaple usually detects multi-row headers automatically. Override the setting if your first charts show symptoms such as question text appearing as data, the first respondent row missing, columns named only with technical codes, or row counts off by one.
Changing the header row setting reloads the project and clears existing column configuration—fix this early. See Header Rows.
Combine waves, trackers, and split files
If your study arrives in multiple files, use Combine files before analysis. Common when each tracker wave is exported separately, column names changed slightly between waves, or new questions appear in later waves.
AddMaple supports stacked combines for wave files and side-by-side combines when joining on a common ID. After combining, analyze by wave, round, country, segment, or other fields in the dataset.
See Combine files (waves and trackers) and Analyzing tracker studies.
Set up weights if your file already has them
If your survey has a weight column, choose it in Manage Project so charts, pivot tables, statistical tests, and exports use weighted results.
The weight column must be numeric. AddMaple applies an existing weight column but does not currently calculate new weights.
See Weighting.
Quick prep checklist
Work through this before you start cutting banners:
- Upload the raw export and check the automatically generated charts.
- Confirm header rows are correct before other column changes.
- Open Manage Column(s) and review names, types, groups, and combined columns.
- Combine split multi-select or checkbox columns that should be one question.
- Group related rating, grid, or repeated-scale columns where comparison helps.
- Rename unclear labels, clean messy categories in the legend, and hide clutter columns.
- Combine separate files first if the study has waves, split files, or tracker rounds.
- Select a weight column if analysis needs weighting.
Once structure looks right, filtering, pivoting, dashboards, insights, and exports use the prepared version of your project.