Upload your CSV or Excel file, choose the column to code, and receive a code frame, an analysis-ready variable matrix, an Excel summary, and a follow-up prompt for any answer engine. All by email. Themes and nets you can hand straight to a stakeholder, plus a way to keep asking questions of your data afterwards.
Self-serve coding, delivered as themes and nets
Coding aims to fit at least 85% of substantive responses to a theme. Up to ~15% uncoded is normal and matches the human-coding benchmark.
Upload, choose the column, brief us in plain English, and pay. Results arrive by email.
A single CSV or Excel (.xlsx) file up to 200MB. You will choose which column to code in the next step.
For Excel files we read the first sheet and expect a single table starting at row 1. Older .xls files should be saved as .xlsx or CSV first.
All selectors are populated from the headers in your file.
Two short fields. The clearer the context, the cleaner the theme frame.
Themes will be built from your data. If you already have a fixed frame you need to apply across a tracking wave, email greg@knowsis.net and we will scope it as a custom project.
The estimate above is approximate. The final price is confirmed from your file when you submit.
Secure payment via Paystack. By continuing you agree to our Terms of Service and Refund Policy.
Coding jobs above 10,000 responses need a custom scope. Leave your details and we will come back with a quote, usually within one working day.
Your code frame, variable matrix, Excel summary, and answer-engine follow-up prompt will arrive at by email.
Three steps from upload to themes in your inbox.
Upload your CSV or Excel file and pick the column to code. Optional respondent ID and break columns for subgroup splits.
Price is set by your response volume, with the tiers below. South African orders include 15% VAT shown at checkout.
A code frame, an analysis-ready variable matrix, an Excel summary, and a follow-up prompt for any answer engine all arrive in your inbox.
Three deliverables you can act on immediately, plus a few promises we hold ourselves to.
The themes and nets your responses were coded to, delivered as a DOCX for marking up and a PDF for sharing.
One row per respondent, with the themes coded out as variables. Plain CSV, ready to load into any analysis tool alongside the rest of your survey data.
Counts and percentages by theme and net, ready to drop into a deck or share with a stakeholder.
A ready-to-use prompt (interrogation_prompt.md) you paste into ChatGPT, Claude, Gemini, or any other answer engine to ask plain-English follow-up questions of your coded data.
Responses that mention more than one thing are coded to every theme that applies, not forced into a single bucket.
Coding aims to fit at least 85% of substantive responses to a theme. Up to ~15% uncoded is normal and matches the human-coding benchmark.
Self-serve, pay-and-go. No account, no platform. POPIA compliant and files deleted within 7 days of delivery.
The variable matrix is plain CSV, ready for any analysis tool. If you want to run driver analysis to see which themes shift a key outcome, build segmentations from the coded data, or layer it with the rest of your survey, the Data Alchemy Platform does all of that in one connected workflow. Free 14-day trial, no credit card needed.
Self-serve in delivery. Practitioner-grounded in method.
Knowsis is an analytics consultancy founded by a researcher who has been writing and coding open-ended survey questions for over twenty years across consumer, B2B, financial services, FMCG, and tobacco categories. The coding standards behind this tool come from that practice.
The coding standards behind this tool are the same ones in use with our clients today.
JTI · Capitec · FT Longitude · dunnhumby · NielsenIQ · Shoprite · Trade Intelligence · Embark Insight
Your file is processed in confidence and used only to produce your results. Uploaded files are deleted within 7 days of delivery, outputs retained 90 days for support, then deleted. Not used as training data. Not shared with third parties.
One flat price per coding order, set by the number of responses in your chosen column.
| Responses | Price (VAT inclusive) |
|---|---|
| Up to 2,000 | R500 |
| 2,001 to 5,000 | R1,000 |
| 5,001 to 10,000 | R1,500 |
| Above 10,000 | Enquiry |
South African orders include 15% VAT, shown split out at checkout. Orders outside South Africa are zero-rated. The estimate on screen is approximate; the final count and price are confirmed from your file when you submit.
Honest answers to the questions most people ask before they upload.
It covers the processing cost and signals genuine intent, which lets us deliver high-quality coding without limit on volume. Coding 2,000 verbatim responses by hand takes a research executive a full day; here it costs R500 and arrives in your inbox.
A CSV or Excel (.xlsx) file up to 200MB with one column of open-ended responses. After upload you choose which column to code from a dropdown populated by your file's own headers.
For Excel files we read the first sheet and expect a single table starting at row 1. Older .xls files should be saved as .xlsx or CSV first.
If your file has several open-ended columns, that is fine, code one at a time as separate orders, one per question.
By the number of responses detected in your chosen column, per the published tiers (up to 2,000 for R500, 2,001 to 5,000 for R1,000, 5,001 to 10,000 for R1,500). South African orders include 15% VAT shown at checkout; orders outside South Africa are zero-rated.
The estimate shown on screen is approximate. The final count and price are confirmed from your file when you submit.
Coding aims to fit at least 85% of substantive responses to a theme. Up to ~15% uncoded is normal and matches the human-coding benchmark.
Genuine "other" responses, blank fields, and one-off mentions sit in an uncoded bucket so you can see exactly what was excluded, and decide whether to refine the frame for the next wave.
The theme frame is built from your data. Reading the responses produces a set of themes and nets that fit them, rather than forcing the data into a pre-existing frame.
If you already have a fixed frame you need to apply across a tracking wave or category framework, email greg@knowsis.net and we will scope it as a custom project.
Yes. Your file is used only to produce your results. Uploaded files are deleted from our systems within 7 days of delivery. Outputs are retained for 90 days for support purposes (so we can re-send them if you lose the email), then deleted.
Not used as training data. Not shared with any third party. Knowsis is POPIA compliant.
Larger coding jobs need a custom scope. Upload your file and the tool will switch to a short enquiry form, or email greg@knowsis.net directly. We typically come back with a quote within one working day.
Results are delivered immediately on payment and the work is done as soon as it lands in your inbox, so we do not offer refunds on this specific product.
If something is clearly wrong, email greg@knowsis.net and Greg will come back to you personally. For higher-stakes studies or recurring tracking, you can also book a conversation with Greg via the follow-up email.
Upload, choose your column, brief us in plain English, and pay. Results arrive by email.
Start codingCoding aims to fit at least 85% of substantive responses to a theme. Up to ~15% uncoded is normal and matches the human-coding benchmark.
