You stop handing over a file. You hand over the book.
Survey Coder produces the full tab book: banners, bases, nets and 95% significance, in an Excel your client can read without you in the room. With your open ends coded, or with none at all.
The problem is not the coding. It is the handover.
An agency workflow has a seam in the middle: someone codes the open ends, hands over a file, and the processing team rebuilds it to tabulate. That rebuild is where two days go, and where the codeframe and the tables stop agreeing with each other.
When coding and tabulation live in one pipeline there is no file to hand over. The codes you approved are literally the rows of your tables — there is no second interpretation of anything.
One pass
1. You upload the file
To code, your .sav, .xlsx or .csv in the shape it already has. To tabulate you need the .sav: it is the only one that carries value labels and variable types.
2. We code the open ends
Against your codeframe if you have one, or generating it and putting it through your review if you do not. If your study carries no open ends, this step simply does not happen.
3. The tab book comes out
Your banner applied to every question, with bases, nets and significance letters.
What if you have no open ends to code?
Step 2 is optional. If your study is all closed questions, or you coded the open ends yourself, you upload the file, define the banner and the book comes out the same: it is built from the variables your .sav already carries.
Coding adds tables to the book. It is not the toll for having one. Same platform, same engine and the same Excel at the end, with open ends in the mix or none at all.
You bring open ends
We code them and they enter the book as one more table, with their base, their nets where they apply and their significance letter. Nothing exported, nothing rebuilt.
You bring none
We tabulate the closed questions already in your file. The book comes out complete: banner, bases, nets, significance and weighting.
What you code enters the book as more tables
It's the one benefit no vendor can copy quickly: it requires coding and tabulating in the same place. In any other chain that happens in separate tools, with a manual handover in between — the same handover that costs the two days in the section above.
You finish coding an open end and there's nothing to export or rebuild to fit it into the banner: it shows up in the book exactly like any closed question, with its base, its nets where they apply, and its significance letter.
Every sheet declares its universe
A tab book is defensible or it is not the moment a client asks where a number came from. So every sheet prints the whole chain: who answered, who did not, and how many people each percentage was computed on.
| Base (N) | 10 | |
|---|---|---|
| 5 | 40.0% | |
| 6 | 20.0% | |
| 7 - Extremely satisfied | 40.0% | |
| Did not answer | 75 | |
| Total sample | 85 |
Look at that 40% "extremely satisfied". It reads like a headline. It is 4 people out of 85, because 75 were never asked the question. Before this the sheet said only "Base 10", and that real incidence had to be dug out of the .sav.
Grid sheets print their base too — one row per item, because each item has its own. And in weighted books the top and bottom boxes are computed with the weight, exactly like the mean: two different rules for the same column is the error nobody catches in time.
What the book contains
- One sheet per question, with your banner and its column legend.
- Declared bases: who answered, who did not, and the total sample.
- 95% significance testing in letter notation, suppressed below base 30.
- Automatic nets on scales (Top 2 Box, Bottom 2 Box) and means where they are legal.
- Weighting applied consistently across percentages, means and nets.
- The crosstab plan is saved to the project: which columns go in the banner, which questions go in and in what order, and multi-response batteries grouped. The next wave starts from there, not from a blank email.
- Weighting by target distributions: you say what percentage each group should have in your universe, and the system works out the weights — no need to bring a pre-computed weight column.
What a sheet looks like
One question per sheet, with the banner you define across the top and its column legend. The letters tell you where the difference between segments is real and where it is noise.
| Total | Age group | ||||
|---|---|---|---|---|---|
| 18-24A | 25-34B | 35-44C | 45-54D | ||
| Base (weighted) | 1246 | 202 | 299 | 362 | 258 |
| Base (unweighted) | 1000 | 165 | 237 | 294 | 206 |
| Very dissatisfied | 5.4% | 3.6% | 4.9% | 6.2% | 6.2% |
| Dissatisfied | 8.8% | 10.7% | 8.8% | 6.7% | 12.4% |
| Neutral | 20.5% | 11.4% | 23.9%A | 23.0%A | 23.4% |
| Satisfied | 35.6% | 37.9% | 35.8% | 33.6% | 32.0% |
| Very satisfied | 29.8% | 36.4% | 26.7% | 30.4% | 26.1% |
| Nets | |||||
| Top 2 Box | 65.4% | 74.2%D | 62.5% | 64.0% | 58.1% |
| Bottom 2 Box | 14.2% | 14.4% | 13.7% | 12.9% | 18.5% |
Example sheet. The letters mark which columns each percentage beats at 95% confidence.
Frequently asked questions
Try it on one wave
The honest way to evaluate it is with a study of your own, comparing the book that comes out against the one you deliver today.