Services Offered

Workshops based on the use of Claude 3.7 Sonnet and Google NotebookLM to assist at every stage of the research process.

Two Training Workshops

Equipping participants to get the most out of the combination of Anthropic Claude 3.7 Sonnet and Google NotebookLM

Qualitative Analysis and Output

A 2 hour online course - practical and hands-on

based on a real qualitative report, together with the 30 interview transcript

• we will practice with Claude together, applying prompts and seeing the output

• the transcripts and report will be resources for you to draw on after the course

• Six participant maximum, to enable individual attention and focus on your projects/interest

Aim:

You will gain the skills, knowledge and confidence to use Claude 3.7 Sonnet and NotebookLM in your future projects

Quantitative Analysis and Output

A 2 hour online course - practical and hands-on

based on a synthetic, realistic quantitative data survey (with the option of supplying your own survey data) to practice with Claude together, applying prompts and seeing the output

Six participant maximum, to enable individual attention and focus on your projects/interests

Aim:

You will gain the skills, knowledge and confidence to use Claude 3.7 Sonnet in your future projects.

In-depth insights through interviews and focus groups with non-verbal metrics added

An example of adding non-verbal merics to a report

Capturing Non-Verbal Metrics

A key tool in testing marketing material is Active Engagement. When shown for 2 or 3 seconds, does the image or proposition grab attention? We can very quickly test multiple versions of taglines, or logos, in this way. Do they catch the eye, or more accurately catch the nonconscious brain. More questions will then tell us why they were eye-catching and engaging. In the real world where a brand must grab attention, this test is very valuable. We can get a positive, considered response in a discussion, but will it work out on the supermarket shelves or on the website?

Because no special setup, technology, or conditions are required, we are able to capture responses in all situations, whether someone is looking at their smart phone, scanning shelves in the supermarket, or watching a film in an auditorium. All that is required is reasonably good quality footage of the face and to have a view of where their eyes are looking. For video ads we can sync reactions second-by-second using the audio, when testing static material we rely on a verbal cue from the interviewer when showing each item. In many cases, video footage contains people simply talking about a brand or a service experience. In non-verbal testing, the respondent does not have to say they don’t enjoy using a busy department store or queuing at a pharmacy – we can literally see the emotions as they describe what they did. We humans are accustomed to providing narratives of what we did, but not moment-by-moment accounts of how we felt in situations. Verbal accounts of our emotions are, of course, influenced by what is acceptable, expected or tactful. In cases when respondents tell us they love a brand, for example, non-verbal cues tell us whether there is real feeling behind the verbal utterances, helping us to identify, not only preferences and negative and positive comments, but the strength of feeling behind them. Non-verbal cues represent a precise measure of whatever we are testing, whether it’s the ease of use of a website, how rewarding a TV ad is to watch, how well a brand catches our attention. The metrics from the data can be analysed and presented in a powerfully engaging way to complement verbal data.

Development of the Method

We conducted a systematic trial over the course of a year to identify the 55 subtle and mixed emotions. We used material designed to arouse a whole range of emotional responses, from positive emotions, to cognitive and mind/body split. To elicit strong negative emotions we read out a list of the most stressful life events and asked the respondents to describe the experience of divorce, bereavement etc. The trial underlined the importance of knowing the context when coding emotions. Knowing how the respondent is likely to react is a key advantage in interpreting signs. Studies have shown that an emotion out of context can be hard to identify (which is one of the weaknesses of AI technologies).

The method has been validated through hundreds of interviews and focus groups, generating thousands of hours of video footageWrite your text here...