When designing the landing page for CXL Institute, we conducted an experiment regarding our explainer video.
We wanted to find out how “trustworthy” and “attractive” different voices were perceived. In this CXL Institute study, we tested four different voices, which differed by gender and whether they were professional voice actors or not.
Question is, did it make a different in how people perceived our video content? Yes, and the results were somewhat surprising.
- Viewers trusted the videos with female narration significantly more than the videos with male narration.
- Viewers generally found the video less attractive when presented by a male, but the professional male voice significantly decreased perceptions of attractiveness.
- Viewers overwhelmingly preferred the professional female voice-over.
- Preferences were influenced by which voice they originally heard, which led to an increased likelihood that the viewer would prefer the original voice heard.
How do I apply this research?The big takeaway here is that the choice of video narrator or voice-over matters. It seems that the choice of gender can affect the trustworthiness of the product and that the choice of voice in general can affect users' perceptions of product attractiveness. So the take home here is to test different voice-over professionals for your videos...and try a female voice first.
The argument for hiring a professional voice actor is that they are experts at setting the tone for the video, inform viewers of what’s important and why they’re watching, and generally spark interest in the video content.
But most of these arguments come from companies who sell voice-overs. There’s not much objective information out there.
On the other hand, letting Alex or Susie from CXL do the voice over is cheap and fast. And perhaps just as effective.
As we started crafting our own explainer video, we questioned how compelling paid voice-overs really are. Is there even a difference between a paid, professional voice-over and an amateur one (someone who’s never done a voice-over, someone like Susie, the CXL videographer)?
Moreover, what kind of voice do people want to hear? A woman’s voice or a man’s? How old should they be? Should they have an accent?
One research study later, here’s what we found.
Full Study Report: Effects of Gender and Professionalism in Voice-Overs
We recruited 202 participants from the United States using an online panel provider. Participants completed a survey based on the CXL promotional video with one of four possible voice-overs:
Voice 1: Professional female voice-over (top-rated seller with >2000 ratings on Fiverr).
Voice 2: “Amateur” female voice-over (Susie Cansler from CXL; has never done voice-over work before).
Voice 3: “Amateur” male voice-over (Alex Birkett from CXL; has never done voice-over work before).
Voice 4: Professional male voice-over (top-rated seller with >2000 ratings on Fiverr).
Participants were randomly assigned to a voice.
Professional female voice-over.
“Amateur” female voice-over.
“Amateur” male voice-over.
Professional male voice-over.
After viewing the video, we asked them two questions:
1. “How trustworthy would you rate this video?” (On a scale of one to seven.)
2. “How would you rate the visual appeal of this video?” (On a scale of one to seven.)
Participants had the option to share additional feedback on the trustworthiness and visual appeal in the video. This provided us with some qualitative feedback data to improve the video.
3. A preference question in which they selected which voice (from 10-second clips) they liked most.
The same snippet was used for each voice, producing a total of four clips (one for each voice-over).
Here are the sound clips (note a 2 second delay to start):
Male Amateur (Alex):
Female Amateur (Susie):
1. People trusted female voices more. Professionalism didn’t matter.
Analysis of variance indicates that there were differences among the voice treatments [F(3, 194) = 6.71, p = 0.00025] in how users viewed the trustworthiness of the video message. Tukey’s post hoc tests revealed that female voices are significantly more trustworthy than male voices, while ‘professionalism’ didn’t influence viewers’ trust perception.
2. The professional male voice-over was rated as the least attractive.
Analysis of variance indicates that there were differences among the voice treatments [F(3, 194) = 6.21, p = 0.00048] in how users viewed the attractiveness of the video message. Tukey’s post hoc tests revealed that the primary cause of this is the professional male’s voice, which caused a significant decrease in viewers’ attractiveness perception compared to the professional female (p-value < 0.01) and the amateur female (p-value < 0.01). The decrease was nearly significant compared to the amateur male (p-value = 0.074) as well.
3. People generally preferred the professional female voice-over, however, they preferred voices differently depending on which voice they originally heard.
People overwhelmingly preferred the pro female voice-over. Professional female voice-over aside, participants were more likely to pick the voice from the original video they watched (e.g. if a participant heard the amateur male voice-over in the original video, they were more likely to prefer that voice when comparing it to the others).
Even with this pattern, the professional female was the clear favorite. A Chi-square goodness of fit test indicates a significant difference in mean preference among users who watched the different original videos [X2 (9, N = 195) = 48.139, p < 0.001].
4. Qualitative insights are useful for product improvement.
For your enjoyment, here is a sampling of some of the feedback we got on the video… which will help us improve it for the next version.
Nothing at all was bad. Liked the experiment graphics best.
Everything flowed very smoothly. It is a great informative video.
I didn’t like how it was so fast, and plain colored. It didn’t draw my attention at all.
I like the graphics showing what the website looked like and the tools available on the website, as well as the colors showing heat maps where users were looking.
It was too vague
The best part was showing an actual person for a second or so in the video. Maybe if it actually showed the narrator every now and then it would liven up the video. Looked sort of amateurish in much of the vide, just copying a webpage.
It seemed to move much too fast to be able to fully comprehend all of the information.
I liked that it was clean, focused, gave details and examples, the speaker was easy to understand, I don’t know what A/B is so it might need to be dumbed down a pit or taken into consideration that I am not the target audience
- We tested four different voices, but there are billions out there. Further research should test more than two variations of each female and male. We should also investigate preference patterns for voices among specific demographics. Zanbaka et. al. (2006) found that there is a cross-preference for voices between genders. In other words, males prefer female voices and females prefer male voices. We should have asked about the respondents’ gender!
- Another interesting tangent might be exploring the effect of foreign accents. Perhaps some accents convey particular messages, moods, or tones (with the British accent being most attractive, of course).
- It could be insightful to test this study on a web optimization demographic instead of the general public. Branching off Zanbaka’s work, there’s a possibility that this predominately male demographic prefers a female voice.
Alex Birkett’s voice was rated as the least trustworthy, so don’t hire him to do a voice-over for your promotional videos.
Other than that, we found that people overwhelmingly preferred the professional female voice-over actress, and that, in general, people rated female voices as more trustworthy than males’ voices.
Of course, our study may have been highly contextual – and we only tested four voices – but there were significant differences in how people perceived the video and those results are surprising.