AI Powered Analysis, Human Driven Insights

At Bolt, we’re always looking for ways to revolutionise market research and maximise impact and efficiencies for our clients. That’s why we’re now harnessing the latest in AI and Natural Language Processing tech to power our open-ended response analysis and coding – as well as in our Creative-Neuro analysis, where we’re utilising Emotion AI technologies for our Facial Coding and Eye tracking capabilities.

AI & Open-Ended Responses: A Powerful Combination

Marketers and researchers have long known that open-ended responses are powerful weapons in any businesses’ market research arsenal. Whereas structured answers in surveys offer participants a closed list of predefined options, open-ended responses can provide unprecedented, un-primed insight into the real thoughts and opinions of our clients’ customers – and a window into the minds of consumers.

With open-ended answers, we have the potential to discover what we didn’t even know we didn’t know. Allowing participants space to share their unique and spontaneous opinions, thoughts, and suggestions can provide unexpected insights and value, offer new and previously unexplored perspectives, and open ways of thinking and approaching concepts that just aren’t possible with traditional structured survey answers.

But until now, the analysis and coding of open-ended responses has traditionally come with challenges around their efficiency and expense, particularly when scaling up to anything truly meaningful.

With manual coding, categorisation, and analysis of responses so often being slow, inefficient, and expensive to run, their value has often been largely offset by their costs and resource intensive lead times. There’s also been the issue of consistency – where two people may not interpret and code responses in the same way or context, leading to analytical inconsistencies at source.

That is – until now…

AI: A Research Revolution

Through utilising the recent revolutions in AI, we’re now able to circumvent these traditional problems, and offer open-ended coding at scales, speeds, and costs that were previously just not possible.

Our new AI solution for coding and analysing open-ended responses not only costs a fraction of the cost, we’re now able to turn around results in hours rather than days.

Not only does it speed up delivery of the all-important insights, it also frees up our skilled analysts time to focus on what really matters – getting behind the steering wheel of the AI engine and getting to uncover and understand the data and insights that are going to really make the difference to our clients outcomes.

Creative Neuro: AI Powered Creative Analysis

Another area where we’re using AI to supercharge our analytics capabilities is in our Creative-Neuro solutions.

Whether looking to optimise upcoming campaigns, measure reaction to a new concept or product design, or understand how to best place a claim or product, our digital Creative Neuro solutions help our clients gain unprecedented insight into the unconscious reactions their ad or campaign is having on their consumers.

Through utilising Emotion AI to power eye tracking, facial coding algorithms, and measure implicit reaction time, we can automatically track and record people’s eye movements and where their attention is directed, as well as analyse their facial expressions and unconscious reactions, to uncover and decode their real, unfiltered emotional responses when exposed to our clients’ ad or creatives.

Once the test is run, we offer powerful frame by frame dashboard insights to help finetune precisely where our clients need to – to optimise results, ensure success, and fuel ongoing growth.

FAQs:

What are the risks of running AI data analysis for open-ended responses?

There are occasions where the AI will misunderstand the meaning and context of a code/tag and end up assigning it to responses that the user didn’t intend.

In cases like these, there are options that our AI solution provides:

• Reword the code so that the AI better understand it and run the process again
• Make bulk edits based on contextual searches, so that the code can be applied to responses that associate with a particular theme

Are the AI tools specialised in analysing market research data?

The AI models that our AI uses are pre-trained on billions of rows of generalist data from around the digital space, so they can be applied to a multitude of use cases, including market research in various industries.

In the case of the AI not being trained to recognise or understand very high-level domain-specific terms or themes, the user can create rule-based edits to define a “dictionary” of themes to guide the AI to a more accurate result, and apply this logic to future datasets.

Is the solution safe to use with proprietary and sensitive data?

Our AI solution is SOC 2 Type II compliant, meaning that it implements the highest standard of data handling security that can be expected in the industry. It is also GDPR compliant.

The platform uses a multi-tenant AWS environment that can be limited to the region of the customer’s operations, i.e. EU customers can opt to have their data stored only on the EU instance of Relevance AI’s environment.

Does it work across multiple languages?

There are in-built translation workflows that allow users to analyse surveys that include multiple languages. There are over 50 languages supported.