Choosing your profile pictures
on the results you'll get on dating sites.
To us guys, meeting girls is important. It's a topic that deserves more than being the subject of superficial articles designed to get clicks and ad revenue. It's a topic that deserves to be rescued from the realm of guesswork and personal theories.
When it comes to online dating, there are some particularly puzzling pieces of advice out there. This excerpt from an article in which language professor and name researcher Christopher Kruken offers his insight is a good example:
"However, when it comes to choosing a partner, your name is very important. On dating sites, for instance, your name is probably the key to establishing contact, because of the connotations it will have."
Your name is the key to establishing contact on dating sites? I'll give the professor this; it is advisable to avoid nicknames with strong negative connotations. But on the other hand, choosing a name because you think it is going to say something positive about you will often make you look like you're bragging or trying to impress someone; "HandsomeDoctor" and "Bodybuilder79" are obviously a couple of douches! In the end, you're better off with something more neutral, and in any case, your nickname certainly isn't the key to anything; other factors, like your pictures, are infinitely more important. I'm sure the professor means well, but at the same time I'm willing to bet he's giving advice about something he's never really tried.
Another piece of common advice is that sites like hotornot.com are great tools for selecting which pictures to use in your dating profile. It may sound like a good idea at first, but try this simple experiment out for yourself: Upload the same image multiple times and see if it gets a consistent score. It won't. The voting is totally random and the uncertainties are huge. We can only conclude that someone started to recommend this method because it sounded like a good idea, without actually bothering to do any testing to confirm its real world usefulness. This is exactly the way of thinking we have to get rid of.
Guesswork has to be replaced with experience, and where possible, with the gold standard for knowledge: The combination of personal experience and objective, controlled studies.
An interesting aspect of online dating is that results are easily quantifiable. What percentage of the messages that you send are answered? How many conversations lead to dates? This makes measuring the efficiency of different approaches and strategies quite straightforward. I've researched online dating in this way for quite some time, and in this article I'm going to share the results of one of my studies.
My goal was to learn more about how guys and girls assess pictures of guys. Are there differences in what guys think is a good photo and what girls actually like? The material I've accumulated through working with clients here at Impressive Profile gives me a unique opportunity to examine this; after all, I have access to a ridiculous amount of pictures of each of them.
I selected three male clients and set my experiment up like this:
- Three male participants were to assess each client's photos
- Three female participants were to assess each client's photos
- Each client was to asses his own photos
All parties were tasked with:
- Selecting the five most attractive pictures
- Selecting the very best picture among these five
None of the participants had met the clients in real life; I wanted them to assess the clients' pictures as if they were strangers on a dating site. The clients obviously gave me their permission to use their pictures for this research, and they all received a small compensation in the form of a price discount. For male participants I used three colleagues from Good Vibe Dating, because I thought it would be interesting to see how they fared; they do, after all, have some serious skills when it comes to dating in general.
These are the questions I wanted answered:
- Do the girls agree with each other on what pictures are attractive?
- Do my colleagues agree with each other on what pictures are attractive?
- Do my colleagues and the girls agree on what pictures are attractive?
- Do the clients and the girls agree on what pictures are attractive?
- If we create three different profiles for each client, one with the girls' choices, one with my colleagues' choices and one with the clients' choices, will there be a difference in the results they get on a dating site?
Let's have a look at what I found:
Do the girls agree with each other on what pictures are attractive? Yes. That is to say, all three girls agreed on the best picture for all three clients, and they mostly agreed on the other pictures as well, even though there were a number of instances in which they had chosen different pictures from the same series, pictures that resembled each other rather closely. I had all the girls review each others choices after they had finished making them, and they would usually agree that the other girls' choices were good as well. In a few cases, though, one girl had made a selection that the other girls really didn't like.
Do my colleagues agree with each other on what pictures are attractive? There was more dissidence among my colleagues, but they still managed to a agree on a lot of the photos. Two of them agreed on the best picture, and the one who had made a different choice claimed he just as easily could have gone with their alternative.
In conclusion, opinions on which pictures were attractive seemed to be reasonably consistent within both groups. To me, this suggests that it's time to retire the myth that "everyone has a different taste"; what is attractive to one person tends to be attractive to others as well. This fits with what we see on dating sites; some guys get a lot of attention, others don't get any attention at all.
Let's move on to our next questions:
Do my colleagues agree with the girls on what pictures are attractive? The two colleagues who agreed on the best picture actually got it right; they picked the same one that the girls had already selected. In general, they did identify a fair number of the pictures that the girls had liked, but they also had a tendency to choose some pictures that the girls hated.
Do the clients themselves agree with the girls on what pictures are attractive? Short answer: No! The clients did a very poor job in identifying their own best pictures, as defined by the girls' choices. Furthermore, it almost seemed to me that they had a preference for images in which they didn't look like "themselves", but at this point that's just speculation.
And the thousand dollar question:
If we create three different profiles for each client, one with the girls' choices, one with my colleagues' choices and one with the clients' choices, will there be a difference in the results they get on a dating site?
This was the most important part of the experiment; it’s time to do some real life measurements and get some hard numbers! I chose a popular dating site, and over the course of a few weeks I created three profiles for client A; first one in Boston, then one in Seattle and finally one in New York. I used the girls’ choice of photos for the first one, my colleagues’ choice of photos for the second one and the clients own choice of photos for the last one. I gave all three profiles the same short profile text, and I filled out the personality test by answering it in the exact same way for each of them.
For each iteration of the client’s profile I made a search, using match value, geographical distance, recent activity on the site and age (client’s age – 10 to client’s age) as search parameters. I started sending messages, working my way down the resulting list until I had sent 80 messages that had been read. Here’s why I chose to do so:
- The site lets you see whether or not the recipient has read you message; I chose to omit messages that were never read from my data because I wanted to make sure I only included active profiles in the study.
- The number 80 was the maximum number of data points allowed by the statistics tool I was using for analysis.
- I wanted the average age and attractiveness of the girls contacted from each iteration of the profile to be the same, so I just worked my way down the list generated by my searches, messaging attractive and unattractive girls alike. I figured that this would make the average age and attractiveness reasonably constant as long as I messaged a sufficient number of women.
I calculated response ratios for all three iterations of client 1’s profile, defined as the number of replies received divided by the number of messages read within two weeks. The message I used was the very simple and neutral “Hi, I’m Mike:) What’s up?” I then repeated the whole process for client B and client C, making sure to switch up the cities and replacing the name Mike with John and Chris respectively.
Here are the average response ratios for each method of picture selection:
Perhaps not unexpectedly, the girls won. The real surprise lies in how much they won by! The profiles they put together did roughly twice as good as the profiles my colleagues put together on average, and almost trice as good as the profiles the clients would have chosen for themselves.
Are these results significant? In statistics, there is a value called p, and it tells you something about the probability that a result has occurred by pure chance. I used this tool to calculate a p-value for the difference between the girls' profiles and the client's profiles; a 1 in the matrix represented a message that was read and replied to, while a 0 represented a message that was read but not replied to. I found a p = 0.00025; in other words, the probability that the observed difference had occurred by chance was 0.025%(!). In medical science, for instance, one will often demand a p-value of 0.01 or less in order to accept a difference between two groups as real; in this experiment, we ended up quite a bit lower.
Discussion
There are, of course, weaknesses to this study. I've done it myself, without access to a relevant academic community; I would love for it to be peer reviewed, but frankly I don't think anyone else study these things in the same way. When online dating is the subject of "real" research, the goal never seems to be to identify what works and what doesn't. The OK Cupid-people may be an exception; I certainly respect their efforts, but they don't really do controlled studies or design experiments, they just analyze the traffic on their site.
An obvious objection to this study is that it's not very "blind"; ideally, I should not have known which group was which when sending messages to girls on the dating site or when analyzing the resulting data. However, this would just have been too hard for me to accomplish with the resources I had at my disposal, and I do hope that the standardized method I used for identifying and contacting girls made up for my lack of blindness.
I've also given response ratios as average values for each group, because I felt that would make this article easier to understand. Furthermore, there really wasn't that big of a difference between the three clients in each group anyway. However, I should note that the three of them were chosen because I felt they were very "average" guys; if they hadn't been, one could argue that the results may have been different.
One final point to be noted is that the pictures the participants had to choose from were taken by yours truly. At the risk of sounding arrogant, there are few, if any, out there with the same amount of experience in taking pictures for guys' dating profiles; I've been doing it for a living since 2009, and I have quite a few tricks up my sleeve. For an optimal picture to be selected, an optimal picture obviously has to be present in the source material, and this source material wasn't just some random Facebook album. If it had been, I suspect we would have seen a similar difference between the groups, but lower response ratios for all of them; I can't say that for sure, though.
In any case, and in all modesty, I still think studies like this one provide a reasonably sound foundation for giving advice, especially in light of all the guesswork and personal theories that's floating around the internet.
Conclusion
Studies and statistics are great sources of information, but they should always be combined with experience. In this case however, everything seems to be pointing in the same direction: You shouldn't be the one selecting which images to use in your dating profile! Even so, that's what most guys are doing; it's really quite interesting. They could probably have gotten results three times better than what they're getting today if they had done things differently.
One of the things that surprised me about this study was that the profiles my colleagues had put together did so poorly; after all, my colleagues did select a lot of the same pictures that the girls had chosen! However, they also tended to pick quite a few images that the girls really hated. I suspect this is why they failed; on a dating site, you will be judged on your worst picture! Think about it − you do the same thing when you look at a girl's album. If she has three good pictures and one bad one you’ll automatically assume the latter to be more representative of her real life looks, and it will affect your overall impression in a big way.
In the girl group, there were also a few cases in which one girl had made a selection that the other girls didn't care for; this seems to suggest that giving each girl the opportunity to veto the other girls' choices could be a smart move if you're assembling a group to select your photos for you. It would ensure that the selection made has the widest possible appeal.
About the author:
-
My name is Sondre and I'm an obsessive guy. I've spent the better part of a decade studying how attraction is kindled by text and images, and through interviews and experiments I have figured out how to create a profile that appeals to exactly the kind of women you want to meet. Through impressiveprofile.com I have helped regular guys get extraordinary results, and now it's finally your turn!