The challenges of coding brands in Market Research

Companies are often very keen to measure the saliency of the brands that they own.

In the area of market research, this is commonly measured using a spontaneous awareness question. For example, "When you think about car brands, which ones come to mind first?". Respondents will then list several brands that come into their heads spontaneously. This kind of question is the acid test of brand salience because it measures top-of-mind awareness.

All quite simple and straightforward so far – but how do you analyze the data collected from a question like this?

Surely, it’s easy to just count the number of mentions of each brand, perhaps using a pivot table in Excel? Unfortunately, in the real world, things aren’t that straightforward. In this blog, we outline the common challenges when it comes to brand coding and how codeit can help you handle this kind of data properly.

Common Challenges in Brand Coding

1. Misspellings of Brand Names

People filling out your survey will make often mistakes when typing responses. Sometimes this is due to lack of care and attention, but some brands are just hard to spell. “Lamborghini”, “Schwarzkopf” and “Häagen-Dazs” are good examples.

When analyzing brand data, it’s important that all mentions are coded and counted correctly – and that includes the misspelled versions.

The dedicated “Brand Coding” mode in codeit contains a specialized AI that finds misspellings and groups them together with distinct brand mentions. This makes it easy to review and allows you to quickly code many verbatims all in one go.

2. Ambiguity, Vagueness and Slang in Survey Data

Sometimes the responses you collect might not be the clear-cut brand names you’d expect. Responses can sometimes be ambiguous (e.g. “Galaxy”), vague (e.g. “Virgin”) or use slang or colloquialisms (e.g. “Mickey D’s”).

Resolving and coding responses like this is usually a judgement call. It requires a real person to adjudicate - based on the research objectives, subject matter and the purpose of the survey. codeit’s human-led ethos means that you are always in control of decisions like this. It’s quick and easy for a human coder to review items that can’t be automatically coded by the AI and take the necessary action.

3. Multiple Brand Mentions in Survey Responses

If your survey question asks respondents to name multiple brands, ideally you would want each brand to appear as a separate verbatim in your data. Usually, you would encourage respondents to separate their answers by including a text box in your questionnaire for each response.

Regardless, it’s quite common for some respondents to simply list a set of brands in one long verbatim, e.g. “McDonalds, KFC, Burger King”. Clearly a longer verbatim like this is more awkward to work with and more difficult to code than single brand mentions.

For this reason, codeit’s Brand Coding mode can automatically split multiple brand mentions into individual items. The coder can then work with each brand separately, making the task much quicker and more efficient.

Can generative AI solve brand coding challenges?

With all the hype around generative AI these days, you might be wondering whether it can solve some of the challenges of coding brands. While Large Language Models are undeniably very useful, they perform quite poorly when applied to brand coding, precisely because of the issues above.  We believe that our specialized AI, designed for brand coding under the careful guidance of a human coder is the fastest way to achieve high quality data when coding brands.

So, if you are struggling with brand coding on your surveys and you want to try the fastest and most efficient way to code brands, sign up for our free 30-day trial.