Among other things, we sell job listings through our Careers 2.0 service, and we thought it might be helpful to determine some of the factors that impact the success of a listing. So we crunched through 6 months of data and these are some of the things we found.
On getting seen
In order to get people to apply for your job, you will have to get them to your listing first. There are a few places that will help with that: our text ads on Stack Overflow (see image), various places on our web site and our tweets. Due to space considerations there isn’t a whole lot for people to base their decision on (to click or not to click?). They’ll see:
- The title
- The employer
- The location
- Whether the job is telecommutable
- Tags (during the research period only on our website)
That’s all you have to sell your position with, so you have to make it count. While you (probably) can’t do much about your name or location, the telecommute status, title and tags you choose can make a difference. We provide roughly 60 characters of title space when displaying jobs on Stack Overflow, and yet a lot of jobs simply have “Developer” or some such as a title. It’s like the seller of luscious, beautiful, high piled, soft shag rugs made with the wool of virgin sheep fed nothing but the finest ambrosia taking out an AdSense ad that says: “Rugs for sale”. Wouldn’t you rather click on “C# developer; work on massive, scalable social solutions” (if you were into such things)? Something as simple as mentioning a technology in your title can improve the percentage of people that apply to your job after seeing the listing (apply rate) by about 25%. While it doesn’t necessarily improve the number of views, it improves the relevance of the viewers, leading to more applications. If you can also give some sense of what type of work people will do, all the better. In January we did something new: we added the ability to tag your job listing. As it turned out, this was a good thing. Listings with tags get on average between 40 to 60% more views than jobs without tags. Based on this finding we now also show the tags in our ads on Stack Overflow. It’s still too early to tell if this will make a difference, but we are thinking it will. For telecommute jobs, the numbers are even more dramatic. Jobs that are marked “telecommute” receive on average 2.25 times more views and 2.125 times more applicants.
On getting applicants
Eyeballs, while important, are at best a measure of how successful you are at attracting people to your listing. The real objective is to get the right people to apply and it turns out there are certain things you can do to improve the number of applicants (just like there are things that will drive most applicants away). To determine what these factors might be we looked at some of the best performing listings and some of the worst ones as measured by apply rate. The difference between the two groups was quite dramatic: The average apply rate for the high performing group was 30.9%, and the average for the lower was 3.2%. For both these groups we looked at a number of factors that might influence a listing’s performance, both positively and negatively, and scored listings accordingly (+1 for positive factors, -1 for negative ones, 0 if not applicable). On average, the well performing listings had twice the score of the low performing ones. Some of the things we found:
The three biggest factors associated with a high apply rate are:
- Culture description (5 times more prevalent among the well performing listings)
- Does the work advertised sound cool? (As measured by the admittedly somewhat arbitrary measure of: “would we like to do this?” – 3 times more prevalent).
- Few bullets (seen in 20% of the high response listings and none of the low response listings).
With regards to culture description, we should note that where the culture was mentioned it always was a good one, which may be the real reason this has a positive effect (we’re fairly certain that describing an average or downright sucky culture would not do much to help). So the Do should really read: If you have a great culture, list it. If you don’t, create one, then list it.
Not all work is inherently cool. But even then, the way it is described matters. Do you give an example of the type of problems candidates will be working on? Of what their work might mean to others? In short, do you tell potential candidates why they should care? (Wrong answer: to make us more money and keep the shareholders happy – we’re paying you a salary after all) Other things we looked at included company description, whether the listing company was well known, position description, telecommute status, salary range posted, and willingness to sponsor H1Bs, but these either didn’t differ from one group to the other or there were too few listings that had these to say something conclusive about them.
We also looked at some potentially negative influencers:
- A plethora of bullets (46% of the low apply rate listings vs. 7% of the top)
- TL;DR (31% vs. 0%)
- Generic title (46% vs. 40%)
We are talking lots of bullets, 10-15+. There were a even few listings with over 25 bullets between the various sections. The worst offenders had multilevel bullets with no descriptive text.
TL;DR really indicated our inability to finish reading the listing, either because of excessive length, dryness or marketing speak. Clear, to the point descriptions of your company and the work the candidate will be doing are good, copying your PR department’s latest press release, not so much. You’re courting here, save your life story for the 3rd date.
When you are hiring a developer you enter a highly competitive market, and to attract stand-out candidates, you need a stand-out listing. While your mileage may vary, the above could help you increase the number of applicants by a factor of (almost) 10. We hope this will help both employers (by getting them more applicants) and programmers (by having better listings to choose from). That's what we found, but we would love to hear from developers and potential employers -- what works for you?