Welcome!
Online Dating Insider delivers cutting-edge insight and commentary on all aspects of the online dating industry. Topics include industry news, site reviews, emerging trends, analysis of dating site features, discussion about safety safety, finance and other issues important to the online dating market. Don't miss our Tools & Services directory, useful to anyone running dating or social networking sites. Subscribe to the RSS feed (you can subscribe via email as well). Your comments and suggestions for stories are welcomed.
Dating site search engines and the algorithms used to display results have always mystified me. Often there seems to be no rhyme or reason to the results. Are members who log in frequently listed first? Are people who don’t log in often effectively punished by showing up farther down the list of results? Match and some other sites allow you to sort by activity and new members. Being able to search based on age of profile or last time user logged in would be useful and straightforward to implement. All sites should offer this.
Many articles have been written on how to improve a site’s Google PageRank, but where is the disucssion about dating site search results ranking? What are the unofficial actions members can perform to show up more prominently in search results? Do profiles that contain certain keywords or phrases get preferential ranking?
To shed some light into how dating site search works, reader mauricev sent me details about his experiments with Match.com’s search, which we was nice enough to allow me to repost.
Do any open-ended search in a big-city (e.g., every woman 33-50 within 50 miles) and Match returns a grid of 16 matches per page and 32 pages. That’s 512 total profiles. But such a search will undoubtedly have more matches than that. So where is everyone else?
Once the search somehow gets the first 512, it just stops and leaves out the rest. How a profile gets into the rest is somewhat unclear. It almost seems random, although the returned matches are supposed to match in order of decreasing criteria. I’ve demonstrated this by searching for myself and watching my result get pushed out beyond the 512 mark as I progressively enlarge the search radius in miles. I believe once I got beyond 10 miles, I no longer show up in the first 512 and therefore don’t show up. With even slightly restricted criteria, there are enough women on match to require searches be limited to about a 5-year range; otherwise, you risk missing profiles!
Except when one does keyword searches from a very large starting pool, the reverse seems to happen. It does the initial search on the match criteria, which here is really just age, distance and gender, then it applies the keyword on those that turn up in the set beyond the first 512. As an example, search for the keyword “engineer” for women aged 33-50 in a 50-mile radius around New York. The search will only return a few hits, but if one does the search repeatedly for small age ranges, there are way more matches returned. That and the fact that hidden profiles have always shown up in all keyword and matches searches round out the major bugs.
Try hiding matches you don’t want in keyword or matchwords searches and they come back the next time you log on.
This is very interesting stuff, indeed. He’s right about the cutoff at 512 results. I noticed that I had to restrict my search criteria to 5 miles from my zip code and sort on activity date to get my profile to show up on the first page of search results. How many people stick 50-mile search radius into their saved searches and forget about it?
I have 10 or so saved search patters on Match. Now I have to go back into each one and make sure I’ve cranked down the age range and radius.
Side note: I never use the “don’t show this person in search results” feature because it would take forever to hide the older profiles that I’ve seen on the site for years. Do people really use this feature and fit it useful?
Scores of people are constantly poking under the hood on social networking, to learn more about how they work an also to exploit vulnerabilities. I’ve seen very little in-depth research into how dating sites operate. Hopefully this will change as the industry matures and dating and social networking sites continue to work together.
Category:UncategorizedBlog reactions
9 responses so far ↓
1
meir (Check me out!)
// Dec 23, 2007 at 1:15 pm
mauricev - very cool analysis.
Have you ever done a comparison between different sites search? While Google is reputedly the best search engine, I’m wondering how you would compare the search engines from the various major dating sites.
2
Markus (Check me out!)
// Dec 24, 2007 at 2:49 pm
The results are most likely based on the probably that they will respond to you. The higher the probability the higher that the recieving user will convert to a paid member.
3
mauricev (Check me out!)
// Dec 24, 2007 at 10:08 pm
Yes, I have compared other sites. Yahoo generally has the best search of all because 1) it didn’t really have serious bugs like match and the keyword searches searched on word roots, so you a search on “geek” would pick up “geeky”. For whatever odd reason, the word root searching stopping working a while ago and the recent update makes it difficult to do repeated keyword searches. Yahoo’s biggest problem is that they don’t have that many profiles (at least for women in NYC).
Most other sites don’t even have keyword searching. I think keyword searching is very useful because most of the other parameters are just clerical; they ignore the meat of the profile.
Another problem with many sites’ search is that they don’t take into account missing answers to profile questions. For example, if one searches for “not sure” or “definitely not” to wanting children, the search will skip profiles that leave the answer blank.
4
may (Check me out!)
// Dec 25, 2007 at 1:58 am
Once I have heard that the website :www.tallhub.com ,is a good place for singles to make friends and dating,maybe you can go and have a try.
5
Hubbs (Check me out!)
// Jan 1, 2008 at 5:14 pm
Match.com is a scam… I tried it and found no one worth responding to. I tried to cancel my subscription after the first two weeks of seeing what they had to offer - then I got an email that said I had to reactivate my account in order to read it - when I did - unbeknownst to me - it reupped my renewal and I just got a deduction from my checking account for $60. I protested the charge with Match.com and was told “Match.com does not have a refund policy” so in other words, I was tricked into renewing and even though as soon as I noticed it, I cancelled again, they still charged my account 3 months later. THIS IS A TOTAL SCAM - DON’T FALL FOR IT…
6
Tom (Check me out!)
// Jan 2, 2008 at 3:04 pm
Sorting an arbitrarily large set of results is a potentially resource intensive process, so every site probably has a limit. If you search on OKCupid, you’ll see that they explicitly say something to the effect of “Your search returned thousands of results; here is a random selection of 200 of them.”
The issue isn’t the limit, it’s the lack of transparency (combined with the ever changing nature) in how search works on the various sites.
The comment about probability of replying (Markus) doesn’t really apply to Match since I’m pretty sure that they don’t even let you send email unless you pay, but it does point out that the services and their customers don’t necessarily have the same priorities in optimizing the search algorithms.
btw - the formatting on the original article makes it impossible to tell which parts are David’s and which parts are quotes from Maurice.
7
mauricev (Check me out!)
// Jan 2, 2008 at 7:53 pm
Actually, the issue is the limit. It’s actually searching from the pool of everyone who has ever been on match! Sort it by date and by the time you scroll halfway through the set of 512, the most recent login is more than three weeks ago. I’d many of those profiles aren’t active any longer! Remove those profiles from the searchable database and that would fix two of the three bugs. BTW, I believe OKCupid does this; they call it “refrigerating” the account.
8
Dr. Joel Block (Check me out!)
// Jan 3, 2008 at 11:07 am
Since the biggest complaint of dating site subscribers is misrepresentation, a partial solution is a LEGITIMATE compatibility test that discourages lying.
Wanna see one that takes 15 minutes and is more engaging and powerful than e-harmony’s test? Go to http://www.ButterfliesAgain.com, check it out and talk to me…
9
Bob Gaines (Check me out!)
// Jul 7, 2008 at 12:43 pm
Just went to check out new dating program that everyone seems to be discussing, ButterfliesAgain.com. Very Impressive. I’ve enrolled in several, this one took only 15 minutes.
Leave a Comment