Insight Examples
Here are three examples of insights you could get from a study like this one. There are also some suggestions for follow-up studies and analyses for your own inspiration.
Contents
- Biometrics add nuance to Survey Responses
- How to go beyond surveys for A/B testing
- Segment biometric data with Survey Responses
Biometrics add nuance to Survey Responses
Surveys help us understand participants and consumer’s intentions and preferences. They give us explicit information about the respondents’ thoughts and feelings. Comparing participant responses to their behavior gives more nuanced insights into their experience.
Example
In this study, one participant reported that they preferred the video ad that highlighted convenience and ingredients.
“More about the making pizza process, which I found more entertaining and tempting than the ‘lets all chill together’ vibe of the first one.”
In that particular video ad, the scenes showing the pizza making process (rolling dough, chopping onion, grating cheese, and spreading sauce) comprised only 35% of the total ad, but made an impression on this participant.
The top yellow wave shows the inhalation and exhalation of the participant. The middle purple line indicates the respiration rate (respiration cycles per minute). The higher the rate, the faster the breathing, the more aroused the person is. At the bottom, you can see the video scenes. The green bars indicate when there was food prep occurring in the ad. The light green rectangles across all three visualizations serve as a visual aid.
Looking at this participant’s respiration data, we can see that this participant had an increased respiration rate that coincides with food prep scenes, starting with chopping an onion and continuing through the rest of the ad.
This respiration data supports the respondent’s answer that they preferred scenes about making pizza. If this study was looking to brainstorm ideas for an effective ad, including more scenes related to making pizza might be an effective strategy for customers similar to this participant.
How to go beyond surveys for A/B testing
Amongst our participants, our survey results show that there was a clear favorite ad!
Because Il Martello designed this ad to focus on more social themes, they may assume that what people liked about the ad was the social aspect.
Why did people like this ad more?
These are examples of research questions that could be explored with remote data collection.
- Maybe these participants liked seeing people enjoying eating pizza together. We could look at eye tracking and facial expression analysis to see if participant’s faces show engagement when they are looking at faces in each ad.
- This ad showed nine pizzas and the other showed one (and a few pizza boxes at the end). We could look at eye tracking data in this ad to see whether viewers paid attention to the pizzas in this ad.
- This ad also had many scenes with people drinking beer. Perhaps participants were paying attention to the beer rather than the pizza. Again, we could use eye tracking to test this.
- Perhaps this ad had more scenes that affected participants’ arousal. We could also look at respiration data in both ads and see which scenes induced changes in arousal, similar to the individual shown above with the other ad. We could then compare to see which ad had more arousing scenes.
Why not just ask them?
Don’t worry, we did! In the survey, we gave them a free text field to tell us why they preferred their favorite ad.
We categorized the responses as reflections or evaluations based on the words they used in their survey response. Responses were considered reflections if they referred to themselves in their answer. They were considered evaluations if they did not include this kind of language. Below are some samples of actual responses.
Reflection
- “It shows made pizza, I like that”
- “I could relate to the first ad better. I could see myself more in those situations than being rushed in the second ad. Pizza is usually a planned meal for my family.”
- “I think they both work well.”
- “It relates more to me – I always buy pizza when I don’t feel like cooking, and not for any of the more “positive” motivators described in the first ad.”
Evaluations
- “Both of the ads have something missing and are very generic rather than connected to the restaurant (besides the logo at the end).”
- “Showed cooked pizza not just ingredients”
- “Happy smiling people eating pizza”
- “More relatable and the imagery chosen was also more relatable- like younger groups of people, POC…”
Many people that preferred the social ad did not provide reflections, but rather gave evaluations. They did not share which ad had a bigger effect on them personally, but rather reported which ad they thought would perform best generally.
This is where nonconscious metrics are very useful.
- Respiration rates could give us an idea of which ads or which scenes were the most arousing.
- Facial expression analysis data could be used to see which scenes were most engaging.
- We could have also used voice analysis rather than the survey question to gather information about their tone of voice when talking about each ad!
Segment Biometric Data with Survey Responses
By “segment the data”, we mean that we can divide our participants into groups and compare their biometric data. For example, we could compare data from people who grew up in one place to people who grew up in another. We could compare people that live with children to people who do not live with children.
A good strategy is to align the demographic questions with the ad targeting options you have. If you have the ability to target your ads by gender, ask respondents their gender so you can look at your data divided by gender. This way you could use your results to influence your strategy.
Below is a summary of some of the demographics survey data from this study. This shows some categories we could have used to segment the data.
Example: Children vs No Children
Let’s say Il Martello was looking to target consumers with children. They want to know if participants with children pay more attention to children in scenes where there are children. They are considering developing an ad which will feature more scenes with children and families.
We took two different scenes from the social ad that have children, adults, and pizza in them. The graphs show comparisons of what participants who reported living with children look at (orange) compared to participants that do not live with children (blue). This is an example of segmenting the data by survey response. The data presented represents the average of the group.
Problem: From this data, Il Martello has a mixed picture.
In the first scene, participants with children looked at Child1 more than Child2. Participants without children spent more time looking at the children compared to participants without children. Participants with children were more interested in looking at the pizza on the table compared to participants with no children.
In the second scene, participants with children looked more at the child than participants with no children. Participants with no children looked more at the pizza than participants without children, the inverse of what we saw in the previous scene.
Solution: Choose how far you want to go
Il Martello could decide that it is not worth making a separate ad to target potential customers with children since there is not clear evidence that these participants consistently respond differently to the scenes with children and families.
They could also choose to investigate further to see why the reactions to the scenes were different.
- The audio: The narrator mentions children explicitly in scene 1. “To make memories with our kids”. In scene two, the narrator says “Just Order Pizza”. One follow-up experiment would be to edit the video so that the scenes are switched but the audio is the same.
- The previous scene: For scene 1, the previous scene features ballerinas sitting on the floor eating pizza. For scene 2, the previous scene shows men drinking beer and eating pizza in front of a television. Switching the scenes could help us know more about how the previous scene contributes to the reaction to the chosen scenes.
- Viewing order: This data is from the first time the participants saw this advertisement. It would be relatively simple to look at only the second viewing or an average of the two viewings to see if the pattern is similar.
- A different comparison: To really get to the bottom of this question, they might also want to know how participants with children responded to scenes without children compared to scenes with children.
Conclusion
Surveys and biometrics together make for a more nuanced and powerful analysis. Biometrics add nuance to survey responses and help researchers dig a little deeper into their data. Surveys help researchers extract actionable insights from their biometric data.
Explore The Sample Study
From here, you can navigate to see examples of insights from this study or go behind the scenes to see how the study was built or how the analysis was conducted.
- Back to Sample Study Landing Page
- Behind the Scenes: Building the Study
- Behind the Scenes: Analyzing the Study (coming soon)
