1. What is Verbatim Coding?
2. How to use Verbatim Coding
1. What is Verbatim Coding?
The Problem: Analysing the results of open-ended questions in surveys is time consuming. Researchers need to manually code, tag, aggregate and analyse results in Excel or similar tools.
Our Solution: a new module in Symanto Insights Platform called ‘Verbatim Coding’: it’s a software solution designed to automate the coding process for open-ended survey responses, turning qualitative feedback into quantitative insights. With AI text analytics, Verbatim Coding on Symanto Insights Platform reduces coding time from days to under an hour, offering multilingual support and additional features for comprehensive data analysis.
By removing the struggles associated with analysing large amounts of customer data and survey input, it provides actionable insights needed for success. Using the advancements in generative AI, such as automatic code book creation, code allocation, and sentiment analysis, businesses can unlock the full potential of open-end coding, to gain deeper understanding of their customers, better decision-making, and ultimately, competitive advantage.
2. How to use Verbatim Coding
- Login to https://app.symanto.com/
- Upload your data to the Data Hub (Excel or CSV file with data)
- Create new project or analyse dataset from Data Hub directly
- Once project is finished à Activate module ‘Verbatim Coding’ (no need to activate the other modules)
- Codebook Creation
Choose between:
Generate a codebook based on actual open-end responses from your data through OpenAI à creates 3 different codebook options (10 labels, 20 labels or 30 labels) for you to choose from, so that you can pick the one that works best for you. Subsequently, you can customize the codebooks to match your specific needs
Note: when choosing a codebook option you still have the option to delete or add labels
- Create your own codebook from scratch (adding codes and labels yourself)
- Upload an already existing codebook from an Excel file
- Run Analysis à Once done, go to ‘Verbatim Coding Results’
- Explore Results
- Understand distribution of codes (absolute and relative values) and the coverage of labels on the open-end-comments to determine which codes have appeared most; this helps to understand g. What are customers talking about? What is most important to customers?
- Adapt the coding rules: the calibration and coding rules allow you to inform the coding algorithm of your preferences. For example, if you want every comment to have a code, you could choose higher coverage, which means less accuracy, but more codes assigned to each comment. But if accuracy matters most and you're okay with some comments not getting codes, you can go for lower coverage, which gives fewer but more accurate codes. The coding rules help you decide how many codes the algorithm should assign to each comment.
- Filter for unlabelled posts or sort by text length
- Search for keywords or phrases or different filters to further split down your data
- Annotation Mode & Finetuning
- correct incorrectly assigned codes
- tell the system correctly assigned codes used to retrain the model
- retrain the model to achieve better results
- understand the quality increase of a retrained model
- Export the data