Data Weighting is available on Enterprise plans. To learn more about what's included in your plan, reach out to our team.
Data Weighting allows researchers to rebalance their sample results to more evenly distribute responses across a population. You build a weight for each respondent outside the platform, upload it as a custom variable, and switch weighting on from the Weighting panel in your Results.
In This Article
- Preparing to construct your custom schema
- Building your custom weight file
- Weighting your results
- What weighting applies to
- Limits and troubleshooting
Preparing to construct your custom schema
You will need:
- your desired metric or variable for weighting
- the exact target percentage for each option or value
- the sampled percentage for each option
- a raw data file from your survey with the individual values to be weighted
The industry standard for weighting is Random Iterative Method (RIM) weighting, also known as raking, which balances a sample against several variables at once. aytm applies one weight column per report, so if you want a RIM scheme across multiple variables, calculate the final per-respondent weight offline and upload it as a single column. If you have questions about developing an appropriate weighting scheme for your data, our Research Services team can help.
Building your custom weight file
- Starting at the Results page, open Response level data and export from your Original - as fielded results. Choose Excel as the file type, and Numerically coded with labels as the data type.
- Follow the rules of a custom variable upload, deleting all columns EXCEPT the respondent ID column and the original variable column (for example, gender).
- Add a column for the custom weight with a title that starts with aytm: followed by a word or phrase to help you remember what the variable is. Example: aytm:wt_gender or aytm:age_weight.
- Calculate the value of each group by dividing the target percentage by the sampled percentage. Use exact, non-rounded percentages in your calculations.
- Populate the aytm: column with the appropriate weighted value for each respondent.
- Save the file in Excel format.
⚠️ Note: Save as Excel (.xlsx), not CSV. A raw export labels the respondent column Response ID, and the CSV upload path only recognizes a column named id or uid. Uploading that CSV unchanged matches zero rows and the upload fails with "No meaningful data was found in the uploaded file." If you must use CSV, rename the header to id first.
💡 Tip: Weighted value = target % ÷ actual (sample) %. To populate the column quickly, use a filter and copy and paste the values, or an if/then formula. Keep values roughly between 0.2 and 5—the platform does not enforce this, but larger swings distort your data.
Column naming rules
| Rule | What happens |
|---|---|
| aytm: prefix | Required. Columns without it are ignored entirely and will not appear as custom variables. |
| wt / weight | Convention only, not a requirement. Any numeric custom variable can be used as a weight, so a clear name helps you pick the right one. |
| Formatting | Names are lowercased, non-alphanumeric characters become underscores, and the name is truncated to 25 characters. |
| Reserved names | A name that collides with an existing platform variable (date, gender, age, and so on) is not rejected—it is renamed to custom_<name>. Look for it under that name in the dropdown. |
| Limit | 100 custom variables per survey. |
Weighting your results
- Click the Filters tab and scroll past any pre-populated variables to the Uploaded Variables group at the bottom.
- Click Upload CSV/XLSX (or Upload more variables) and select your saved file. You can also drag and drop the file onto the panel.
- Open the Weighting tab in the left-hand menu.
- Click the Choose weighting variable dropdown and select Weight by Custom variable.
- In the second dropdown that appears, Choose custom variable options, select the variable you want to weight by. The first variable alphabetically is pre-selected, so confirm this even if you only uploaded one file.
- Toggle weighting on. The caption changes from No weighting applied to Weighted by custom variable, and the data set reloads with your scheme applied. If you are changing an already-active weighting, click Apply.
- Go to the Reports tab and click Save as to save a new report with the weighted data.
⚠️ Note: Respondents with no weight in your file are not dropped—they are kept and assigned a weight of 1.0. The panel displays a warning when this happens. Because this changes your numbers, make sure every respondent you intend to weight has a value in the column.
You can upload as many weighting variables as you like and save a separate report for each scheme, but only one weighting variable can be active at a time.
4. What weighting applies to
| Weighted | Not weighted |
|---|---|
| Platform charts | Response-level (raw) exports, by design |
| Banner Table exports | |
| Significance testing | |
| Summary data | |
| TURF and conjoint exports |
Anyone opening a share link to a saved weighted report sees the weighted figures, though the Weighting tab itself is not available to them.
Limits and troubleshooting
- My upload failed. Check that the file is .xlsx, or that the CSV's respondent column is named id, and that your weight column carries the aytm: prefix.
- I can't find my variable in the dropdown. The dropdown lists numeric custom variables only. If you used a reserved name, look for it as custom_<name>.
- My saved report is no longer weighted. Deleting the custom variable a report was weighted by removes the weighting from that report. Keep the variable in place for as long as you need the report.
- The Weighting tab isn't there. Weighting is unavailable on virtual reports, special reports, and in QA mode, and it applies only in Charts mode.