A finished survey leaves you with one data file and one big question: how do different groups answer? Banner tables are how market researchers answer it at scale. Instead of building a crosstab for each question, you define the groups you care about once and run the whole questionnaire against them. If tables are new to you, start with our guide to what a crosstab is, from one-way tables to banners. This guide covers the banner, the stub, the tab plan and a worked example.
Key takeaways
- Banner tables run every survey question (the stub) against the same subgroups (the banner), so the whole study reads the same way.
- Each table shows bases, column percentages, nets or summary rows, and significance letters.
- Design the banner from the decisions the study supports: 3 to 6 groups, with enough respondents in every column.
- The tab plan, or tab spec, defines the banners, stubs, bases, nets, weights and statistics.
- Banner tables give analysts the full read; charts tell the story.
What are banner tables in market research?
A banner table is a crosstab whose columns repeat across every question. The columns along the top are the banner: Total, then subgroups such as gender, age, region or customer segment. The rows down the side are the stub: the answer options of one question. Sixty questions against one banner give 60 tables with identical columns.
The name comes from print: column headings ran across the top of wide paper like a banner. “Tabs” is short for tabulations, so a request for “the tabs” or “the banners” means the full set of banner tables for a study. For how they fit into survey analysis more broadly, see our guide to survey crosstabs.
What are the parts of a banner table?
| Part | What it is |
|---|---|
| Banner | The set of columns across the top, applied to every table |
| Banner group | Related columns, such as the three age bands |
| Banner point | One column, such as “Age 18–34” |
| Column letter | A letter (A, B, C…) for each banner point, used by the significance tests |
| Stub | The rows down the side: the answer options |
| Base rows | Respondents behind each percentage: unweighted and, in weighted studies, weighted |
| Nets | Combined answer options, such as “Any brand” or “Satisfied (4–5)“ |
| Summary rows | Top-2-box, bottom-2-box or the mean of a rating scale |
| Significance letters | A letter in a cell means the value is significantly higher than that lettered column |
| Footnotes | Question wording, base definition, test used and confidence level |
The base row matters most: 62% of 380 people is a solid finding, 62% of 40 is not.
How do you design a banner?
- Start from the decisions the study supports. If the brand team is choosing between a younger and an older target, age belongs in the banner. Splits no one will act on stay out.
- Pick 3 to 6 banner groups. Total, gender, age, region, customer segment and brand user are typical.
- Keep each group mutually exclusive where possible. Each respondent falls into one age band. Label any group that overlaps, such as users of several brands.
- Check minimum base sizes. Merge or drop columns likely to fall below your reporting threshold.
- Order banners consistently. Keep groups, order and letters the same across banners and waves.
What does a banner table look like in practice?
The numbers below are illustrative. A survey of 1,200 respondents asks aided awareness of four brands. The banner has Total (A), three age bands (B to D) and two regions (E and F). People can know several brands, so columns add up to more than 100%.
| Aided awareness | Total (A) | 18–34 (B) | 35–54 (C) | 55+ (D) | East (E) | West (F) |
|---|---|---|---|---|---|---|
| Base: all respondents (unweighted) | 1,200 | 380 | 440 | 380 | 650 | 550 |
| Brand W | 80% | 79% | 80% | 82% | 81% | 79% |
| Brand X | 56% | 62% D | 56% | 50% | 55% | 58% |
| Brand Y | 43% | 42% | 43% | 44% | 47% F | 38% |
| Brand Z | 24% | 25% | 24% | 22% | 24% | 23% |
| Any of these brands (net) | 90% | 89% | 90% | 90% | 90% | 89% |
Column-proportion tests at 95% confidence, within each banner group. Total is not tested.
The arithmetic. Each cell is a column percentage: aware respondents divided by the column base. Brand X among 18–34s is 236 ÷ 380 = 62.1%, shown as 62%. Total must agree with its parts: for Brand X, 236 + 246 + 190 = 672 aware across the age bands and 355 + 317 = 672 across the regions, so Total is 672 ÷ 1,200 = 56%.
The letters. The “D” in column B means Brand X awareness among 18–34s (62%) is significantly higher than among those 55 and over (50%), a 12-point gap. The “F” in column E means Brand Y awareness in the East (47%) is significantly higher than in the West (38%). The 6-point gap between B and C on Brand X carries no letter: at these bases it falls short of 95% confidence. See how to read a crosstab and significance testing in crosstabs for more.
The net shows that 90% know at least one brand. It counts respondents; adding up the brand rows would double-count people who know several.
What goes in a tab plan?
The tab plan, also called the tab spec, is the written instruction set for producing the banner tables:
| Element | What the tab plan specifies |
|---|---|
| Banners | Groups, banner points, column definitions and letters |
| Stubs | Which questions get tables, row order and labels |
| Bases | Who each table covers, such as all respondents or only those aware of the brand |
| Filters | Subsets of respondents for specific tables |
| Nets | Which answer options combine, and their labels |
| Weights | Which weight applies, and which bases are shown |
| Statistics | Means, top-2-box, significance tests, confidence level, low-base rules |
| Sorting | Rows sorted by Total or kept in questionnaire order |
In agencies, the project lead usually writes the tab plan and a data-processing team builds the tables; in-house insights teams often do both. Write it before fieldwork closes so questionnaire and tables agree.
Should you use one banner or several?
One banner suits most studies. When the groups outgrow the page, split them into, say, a demographic banner (age, gender, region) and a behavioral banner (brand user, purchase frequency, segment). Tables stay readable, but output multiplies: two banners on 60 questions give 120 tables. Put Total in every banner as a reference point.
When do you use banner tables instead of charts?
Banner tables are for analysts and the full read: every answer, group, base and test, which is what you need to find the story and check it. Charts, slides and summary dashboards are for telling that story to people who will not read 120 tables. Use the tables to decide what matters, then chart the findings that change a decision.
How are banner tables delivered?
- Spreadsheet export. One table per sheet, ready to open in Excel or Google Sheets. Easy to share, but static.
- Interactive tables. Analysts change the banner, filter or base on screen and see the result, with tests rerun.
- Charts. Selected rows turned into visuals for reports and presentations.
If you build banner tables regularly, look for crosstabs and banner tables with weighting and significance testing, with charts and interactive report pages alongside. Some questions only come up after the tabs are read, and being able to ask your own research in plain language, with the source shown can save a new table request.
What are the common mistakes with banner tables?
- Too many banner points. Twenty columns are hard to read, and every extra column adds tests and chance letters.
- Overlapping groups presented as exclusive. Readers assume a group’s columns add up to Total. Label overlaps and don’t test them as independent.
- Bases not shown. Without base rows, no one can tell a solid 62% from a fragile one. Show unweighted bases always, and weighted bases when the data is weighted.
- Inconsistent nets across waves. If “Satisfied” means 4–5 in one wave and 3–5 in the next, the trend is an artifact. Fix net definitions in the tab plan and reuse them.
How do you get started?
Before your next study closes, draft the tab plan: list the decisions, choose the banner groups, estimate the base in every column and fix the nets. When the first tables arrive, check Total and the bases first. For teams running wide studies and global brand trackers, see how mTab works for insights and research teams.