# How to increase traffic to the site?

First, a congratulatory news: at some point today, Mathematics.StackExchange became the 3rd largest site on SE network, by the number of questions posted. This is the first time an SE 2.0 site overtook a member of the original trilogy.

But the 4th and 5th columns are low, especially the fifth one: traffic to the site. By that parameter we are merely $\mathbf{14}^{th}$ on the network, behind sites such as TeX.

Update. The current state of the StackExchange Universe (excluding StackOverflow, which is way off the chart) is pictured below, in the Questions-Visits coordinate system:

The universe is expanding from the Big Bang (Area 51) zone in bottom left corner. Different galaxies are headed in very different directions. Someone should make a Hertzsprung–Russell diagram out of this.

Low traffic means few people read the content. Of the 155K questions on Math, only 2725 have 1000 views. Thus, we have fewer widely-read question/answers than other, smaller sites:

• Android: 3526 questions with 1000+ views
• English: 4749
• MathOverflow: 5038
• Programmers: 5192
• TeX: 5827
• Ask Different: 7032
• Arqade: 12517

What can we do to make the content more widely read? Some ideas:

1. Write descriptive titles to questions, and use some actual, Google-able words in questions and answers. It's not unusual to see titles and posts that are essentially MathJax with some prepositions, articles and punctuations. Such content isn't going to be found easily.

2. Create canonical Q&A pairs; e.g., if you find yourself repeating the same argument in several answers, consider isolating it in a self-contained question with self-answer. (I admit not doing this myself, but I'll try to improve.)

3. Finally get the blog started. This was discussed and many people volunteered to contribute. Is there any reason the blog doesn't exist yet?

If you have more or better ideas, please share them.

• We can pass legislations which forces everyone everywhere to take math courses all the time. Surely the traffic will increase tenfold! – Asaf Karagila Jul 14 '13 at 19:46
• @AsafKaragila "Simply" getting more of existing math students to use the site would have the same effect... the current number of users is a tiny fraction of the number of students taking math. – 40 votes Jul 14 '13 at 19:51
• View from the other side: Is ServerFault doomed? – 40 votes Jul 14 '13 at 20:01
• Should we post comments on that thread in the form of "neener neener"? :-P – Asaf Karagila Jul 14 '13 at 20:03
• Every time someone posts a question like "Prove blah, blah." Write the comment, "We won't answer your question until you view at least ten questions that have more than one hundred views." Make them work for their answers! – Jay Jul 14 '13 at 21:18
• Maybe SE in general needs some publicity. I found out about this website a few months ago from a math undergrad student, but before that I had never heard of it before. – Ovi Jul 14 '13 at 22:03
• Concerning (2): canonical answers for lower-level questions, and that would in all likelihood be the bulk of them, are rarely worth the energy. Textbooks are full of the equivalent, and one more is unlikely to make much difference. Typically it’s specific feedback on a specific question that’s needed. – Brian M. Scott Jul 15 '13 at 10:13
• I don’t think that it’s at all surprising that we have fewer questions with 1000+ views than English: it’s inherent in the natures of the two subjects and in people’s attitudes and exposure to them. – Brian M. Scott Jul 15 '13 at 10:31
• Perhaps we should consider whether we have too much traffic already. – Gerry Myerson Jul 15 '13 at 13:09
• @GerryMyerson You may be referring to the number of questions asked per day. I agree that the number of questions is not too low. In SE terms, traffic is the number of visits to the site per day; it's quite low here relative to other statistics. Most of the traffic consists of non-users arriving via search. – 40 votes Jul 15 '13 at 13:20
• @40 , mostly I was being a mathematician, and not accepting that there is a problem without first considering other possibilities. – Gerry Myerson Jul 15 '13 at 13:39
• @GerryMyerson That's fine. Most mathematicians I know aren't very willing to consider the possibility that their work is read by too many people, rather than too few. – 40 votes Jul 15 '13 at 15:12
• Why don't you post a question on meta titled "Very easy way to get a lot of gold badges" and tell people about the publicist badge and how to get it? – Brian Rushton Jul 16 '13 at 0:11
• @BrianRushton Because I have zero credibility in the matter, lacking even the Announcer. You have two Publicist badges -- share the secret! – 40 votes Jul 16 '13 at 0:13
• I propose you are asking the wrong question: the question-to-visit ratio suggests that the internet believes the average value of a math.stackexchange.com question is much lower than other stackexchange sites. I argue that we shouldn't be looking to increase traffic, but raise the average value of a question, even if it means fewer questions overall. – user14972 May 3 '14 at 20:16

I do not consider this to be an issue. Mathematics is a Q&A site which just so happens to have an archive. However I use the archive in TeX, for example, more often than I would be willing to admit, and I think that this is where the high traffic is coming from.

I think a greater use of the archives of other sites is because it is easier to ask a Linear Algebra question than to try and find it in an archive (lots of people know the answer and are willing to help, but searching your specific problem which involves vectors and matrices and the like takes forever!). On the other hand, off the top of my head I can't think of anyone I know who can help me to make precisely one of the ten enumerate lists I have in my TeX file be numbered using Roman numerals. TeX knows though, and I found it by google-ing "numbering using roman numerals enumerate". So using the TeX archive is dead easy, much easier than asking the question myself.

My point is that I do not think we should be loosing sleep over this. On the other hand, I do not think going out of our way to make this site more appealing is a bad thing!

• To be clear, I wasn't thinking of homework-type questions as candidates for 1000+ views. Take Can every group be represented by a group of matrices?, for example: good question, good answers, <1000 views... – 40 votes Jul 15 '13 at 15:20
• @40votes My point is not about getting specific questions with high views, but rather why other sites have a larger number of visitors (although clearly this will increase the number of views questions get across the board, so it is related). – user1729 Jul 15 '13 at 15:40
• losing* (not loosing) – Dan Moore Apr 26 '19 at 20:22

Given that I collect network-wide statistics for fun (the data from this site alone gives over 700 numbers to play with, and I enjoy playing around with statistics so I collect that and other data on a daily basis), and I've been involved in other SE sites besides this one with very different cultures, I thought I'd contribute my 2 cents.

Here's a plot of all SE sites. Each data point is an individual site (104 points in total). The horizontal axis is site age (this is not so important, but I need a parameter to plot things against and age happens to be convenient). The vertical axis is the average visits per question per day on a logscale. The point corresponding to MSE is colored red.

As you can see, it's near the bottom; in fact, the only sites below MSE are Judaism and Meta Stack Overflow. Other sites which are low on this scale are Theoretical Computer Science, Genealogy, Stack Apps, MO, and Startups. The common feature here is that the posts on these sites tend to only be useful to a small number of people, while the top-trafficked sites (on a per-question basis) like gaming and cooking have questions which are useful to very broad groups of people.

The only legitimate way to get more traffic is for more people to find the pages on the site which answer questions they have. For SE sites, usually over 90% of the traffic is search traffic. Moderators have tools to determine how much of the traffic is coming from searches, but I don't think they're allowed to share the numbers. Of course, if there's a really cool question that is of interest to a lot of people, you can share it on various other sites, but this is never going to be a big source of traffic.

Conversely, low traffic is only a problem if it's because people aren't finding the pages which answer their questions. If this is the case (which I'm not sure of) then there are several things that can be done. The first, and most obvious, thing is to improve question titles. Titles get a lot of weight in search engine results (far more than the body and content of answers). The title needs to be phrased so that someone searching will hit all of the keywords, and other than that as brief as possible. Of course, this is probably impossible for PSQs, which are usually too specific to be searched for by lots of people, but for common questions about understanding concepts a better title can easily bring it from unlisted on relevant searches to one of the top results on Google. It's likely that avoiding MathJax in the title would probably help also (I don't know for sure on this point), though this can obviously only be done for some questions. Beyond titles, the only other big thing is to make sure that the answers actually do fully and completely answer the question, as there are plenty of alternative sites to go to if the answers here aren't good, and they'll end up as the ones getting most of the traffic.

Given a particular question, you can always check how well it's ranked by search engines by going to Google and searching for what the question is asking (though you have to be careful, as the phrasing of the question can bias you towards a specific way of asking the question). If you find the question in the first few results from Google, it's probably doing alright. If you find lots of other sites first, the question should probably be edited to have a better title. Also, if a particular question is frequently getting duplicated, it's an indication that question is probably of a lot of interest, and people aren't finding it.

Ultimately, though, I think MSE has a different approach than most SE sites, and the low traffic isn't such a big deal. The goal here isn't so much to disseminate useful information for future viewers as to be a tool to assist in learning mathematics. To that end, the questions are mostly useful to the original askers, which is typically not the case on most SE sites. This could lead to lower traffic, but that's not really a big deal if questions are getting answered.

• I sometimes suspect that Google takes my browsing history (in Chrome) into account and biases search results in favor of MSE, figuring that I must like the site where I spend so much of my time... By the way, how did you grab data from that page? Its HTML markup does not look data-extraction-friendly. – 40 votes Jul 23 '13 at 2:30
• @40votes I've noticed that as well from time to time. If you go to private browsing mode (in Chrome use Ctrl+Shift+N) then whatever biases it applies seem to go away. As for extracting the data, I actually didn't write the code which does that, but I think all it does is look for the strings like "<input type="hidden" name="questions-per-day" value="7049.35714285714" />" and copy the "value" portions (in this case, this is SO's Q/D). These occur right at the end of the section which defines each site. The rest of it is just skipped over. – Logan M Jul 23 '13 at 2:48
• I agree very strongly with your final paragraph. – Brian M. Scott Jul 23 '13 at 2:56

Having read "Is serverfault doomed?" (From the comments) my experience of late around the authors main point, to vote more is relevant to the discussion.

While still relatively new (almost a year) user, I have been offput from answering some questions. Whilst I don't expect $1/3$ of my $(136)$ answers to be bad (via downvoting/commenting), they are neither acknowledged as good (via upvoting). This makes it difficult to guage what is an appropriate level of quality within an answer, as well a being unsure if it is even being read.

Both of these reasons limit my activity on the site. I do not know how these reasons transfer to other users or particularly new users.

• The site has 10 Unsung Heroes, and you are one of them... I completely agree that voting activity on the site is insufficient... For what it's worth, I always use my – 40 votes Jul 14 '13 at 23:14
• Assuming the score of a question/answer is the best and absolute way of determining its quality, here is the best 2013 MSE production so far. – Julien Jul 15 '13 at 0:56
• @julien Yes, but this is orthogonal to Daryl's point, which was: a lot of good answers get exactly zero votes. I don't really care if some post goes viral and acquires tons of votes from other SE sites (this is at worst mildly annoying), but zeros next to nontrivial contributions look really bad. (This point is also made in that ServerFault discussion). – 40 votes Jul 15 '13 at 0:58
• @40votes Sure (you should know by now that my comments are often biased). I actually don't find viral posts annoying, it's often entertaining. Like you, I find it more annoying when nontrivial contributions don't receive the upvotes they deserve. The person I learn the most from here does not have a single nice answer badge and about 70% of answers with 2 upvotes at most. – Julien Jul 15 '13 at 1:10
• @julien My point is not that I don't get huge numbers of votes, but rather that many of my posts, from my perspective, seem as though they are ignored. Whether they are or not is a different story, but I do get that impression at times. Also, I'm sure that I have a number of bad/very bad answers, but I don't think it is a third of them. – Daryl Jul 15 '13 at 1:19
• @Daryl It seems that you often answer questions that I have not really seen or looked carefully. All I can say is that I will pay special attention in the future if I come across an answer of yours. – Julien Jul 15 '13 at 1:32

Second 1,2,3. I'd like to try to write blog at my level hopefully will be able to help beginners like myself. Even on blogs from outside directed to mathSE would increase the flow.

An idea: add subscribe feature for the site. There are many people I'd like to follow, for example, @40 votes. Others like @Jack Lee, @Matt E, @Ted Shifrin, @Zev Chonoles, @nik etc. Adding such feature not only provides convenience, but reminds me to return to mathSE for content I absolutely don't want to miss.

Create archive type articles - an overview of related topics. An article that directs people to outstanding posts for a deeper and wider understanding of the topic is really helpful, and it increases the hits. To refer to @Jay's comment above - such reference ("work out the questions XXX, XXX...") would be really helpful! So I would like to suggest to encourage such posts, and create a corresponding tag (like "archive").

• Re: "subscribe feature" -- every user profile has "user feed" RSS link in the bottom right corner. – 40 votes Jul 14 '13 at 23:09
• @40votes If you are not already getting a salary from SE, you should negociate one. – Julien Jul 14 '13 at 23:30
• Great, thanks @40votes. Negotiate a salary! Or ask for a raise if you already have one.. – 1LiterTears Jul 14 '13 at 23:40
• @julien Somehow I am thinking 40votes is 5pm, whose post I have learned a LOT from. – Shuhao Cao Jul 15 '13 at 4:52

Simple suggestion regarding the 4th column, I don't know whether this is appropriate or not, but it is still an idea. It might lower the average post quality, but any significant increase in traffic may have similar results.

I think that users of Math.SE represent a huge number of academic institutions world-wide. To attract more users, volunteers could hang posters provided by SE (those could be sent via mail, or maybe even printed at destination). Poster could have excerpts of best questions/answers from the site. Depending on funds available, all users with reputation $>X$ that would send a photo of hanged poster might get a T-shirt.

The thing is to make communities aware of existence of Math.SE. When someone needs help, any mild effort (like registration, writing a question, etc.) is worth it. Of course, summer holidays is not a good time to put a poster.

What do you think?

PS, some funny observation: it was not specific knowledge/problem that get me to know stackoverflow, but this funny topic. After reading that I was more inclined to search for answers there than somewhere else.

• Mug, t-shirts, etc is good idea. A customized t-shirt would be nice. Like MSE logo in front, punchline "solving your problems since 2010" in the back. – Shuhao Cao Jul 15 '13 at 13:39

Perhaps there should be some badges for "archive management" - for example a gold badge for editing 500 titles of questions which already have an answer and are associated to a particular tag so that they are consistent and helpful and show up easily in appropriate searches.

Also retagging the archive for consistency and search ability.

[Obviously the problem is that such edits could be unhelpful]

• Sadly, the existing badges Strunk & White and Copy Editor don't seem to stimulate people enough. Only 184 of the former and 39 of the latter have been awarded. Ridiculously low numbers. I'm aware that mass editing of old posts can be annoying to some people, but there is no downside to editing a question one has just answered. Maybe I'm going to put a sticky note next to the screen to remind myself: answered? hit "edit" next! – 40 votes Jul 24 '13 at 15:15