Auto-coding is our platform’s newest feature that’s set to simplify your market research experience even further, courtesy of the Zappi development team. It’s currently a ‘Beta’ feature, which means we’re looking to refine it with your feedback.

This feature is only available for ‘likes’ and ‘dislikes’ responses on Zappi Tools for now, but we will roll out the feature to cover more products and open-ended question types in the near future.

Open-ended responses to survey questions are free text inputted by a respondent and offer an alternative to single or multi-choice responses.

These responses can contain invaluable insight as to what consumers think of the stimuli you’re testing. Typically, making sense of this data has been difficult, requiring hours spent categorizing responses to try and make sense of what’s been said.

With deep learning algorithms, everything changes.

Auto-coding: The Basics

What It Does

Our new feature extracts insight from open-ended responses without manual organization and analysis. At the click of a button, it reads through open responses and categorizes them into relevant topics, allowing for aggregated reporting on open-end responses and easy navigation through the things people are saying about certain topics.

How It’s Organized

After reviewing large amounts of open ended data, Zappi has curated a vast list of possible topic areas. These topics are categorized into a three tier hierarchy: super topics, topics, and subtopics. Our chart allows you to navigate through these different tiers with ease.

Why It’s Flexible

We know that some verbatim responses do not provide too much actionable insight, so we’ve built functionality that allows for topics to be removed and re-added. When this happens, the proportions of the chart are automatically recalculated.

Digging Deeper

This feature is interactive, allowing users to navigate through topic areas and take a closer look at the individual responses within them.

Knowing More: What Are The Possibilities?

Scale questions cannot truly represent the minutiae of human opinion. Closed questions hide a lot of useful information about why people like an advertisement and what exactly they liked about it.

Our solution, however, introduces structure to data that would otherwise be entirely undefined. It’s important not to underestimate the analytical opportunities open to us with this technology.

Crucially, the ability to ask more open-ended questions also sends the right message to our respondents:

“Open-ended questions demonstrate to someone that you actually care what they think. It makes it obvious that your agenda is to learn, not to convince.”

Diana Kander, Author and Keynote Speaker

But you might be wondering: ‘What’s all this fuss over free text and why is it preferable?’ One example of open-ended questions showing real value is when they reveal unintended consequences.

A respondent might say: “I really liked the story but the song lyrics are offensive.” Without an open-ended response, this negative feedback may have been totally sidelined in favour of the ‘liked’ message.

There are countless examples of when this unexpected insight may show up. Seeing more of it will equip brands with a deeper understanding of their audiences more regularly and with less effort.

What Does the Future Look Like?

The future

Machines have been incapable of understanding the intricacies of human language due to how varied and complex our references can seem. Building further frameworks around ‘context’ is crucial.

The trend for consumer insights so far has been to look back, assess the damage, and apply whatever changes to future projects. We think the increased speed of modern business demands a more immediate solution.


If you have any questions on autocoded open ends, please reach out to your customer success team member.

A James Hodges

Posted by

A James Hodges

Loading Disqus Comments ...