It’s always good to measure performance and try to improve obviously. The hard thing is finding WHAT to measure. I decided therefore analyzing performance of my posts and which title works best. Blog titles are easy to test and measure as you will see. The result however, was very unexpected.
I’ve decided to do an experiment with titles, just like suggested at the buffer blog. I pushed two titles in about an hour difference and see the results . This helped me choose the better title for my post.
To those of you who wonder what’s the link I use in the test, as the post is not published yet, but I do want to measure the clicks on it. So, I just used the link the SUN main page. Yes, people tell me that the link is broken, but I always
reply that it’s a post on the making, which is true. How else would I measure my clicks?
Here is how Belle from buffer described it:
A/B Test different headlines
Something we do with almost every post we share from the Buffer blog especially, is to test several different headlines to see what works best. Sharing the same post within a couple of hours using multiple headlines can help us work out which one matches the content and makes it seem the most interesting.
Here’s how we found this process to work best:
For Twitter, we’ve experimented with A/B testing the right headline. A/B testing on social is arguably very hard. Yet we’ve found it’s possible to still get reliable data that way. Here is how we approached this:
1.) Find 2 headlines for an article that you think will perform well.
2.) Tweet both of these headlines at roughly the same time, at least 1 hour apart. Here I’ve found that doing the 2 Tweets both in the AM or both in the PM works best – 9am is much more similar to 10am, then say 12pm is to 1pm. So going with clear “morning” or “afternoon” times is crucial.
3.) Compare the data for which headline to settle on.
Which network should you care most?
Here is where I stand. I keep discovering over and over again that my crowd is on linkedin. I have the most connections on linkedin and that helps me get to more and more people. Not to mention most newsletter registrations come from linkedin as well.
I’m not sure why I’m not happy about it though. Linkedin is a professional business network and therefore perceived as a serious, not fun social network (nobody puts cats videos on his feed on Linkedin) , which is not what I want Startup Noodle to be. Question if you can make fun branding on Linkedin….
Therefore I try to engage more with my audience on Twitter and Facebook as well, and most recently on Google+ and build the audience there. I still asking myself, should I just strengthen Linkedin and forget about the rest of the other social networks or keep building them all.
Anyway, bottom line, the statistics from Linkedin should be most important for me for now.
Here is the result from my experiment. Must say this is the most engagement I got on linkedin ever. But see the huge difference between Tweeter followers engagement and Linkedin?!
The weird thing is that my followers on linkedin are my connections. I have a group there but it’s not so big, and I only opened a business page a couple of days ago.
So you can argue that when you communicate with audience not through a group/page but just through your normal connections you can get much more, as it feels more authentic than interacting with people using a group/page.
My goal though is to communicate the same way in all my channels; Give value and help.
Publishing my post & real performance
A couple of days have passed and I published the post with the winning headline from linkedin: “Audio Books are Bad Marketing“.
Look! Surprising, no?! This is not even close to what I was expecting. Linkedin got slightly more engagement, but still not comparable from when the headline experiment was done.
I can’t explain why the engagement was significantly lower. You can argue that it was Christmas Eve. and therefore there was less traffic. But traffic to SUN on that day was only 25% lower than on December 19th (the day of the headlines experiment).
There was a different of 4-5 hours of the publishing times between the two dates: 1:34PM & 2:33PM VS. 9:33 AM. So, that might have caused the difference, I just didn’t think it would be that dramatic.
Conclusion & What I’ve learned?
Testing headlines is a good way to try to predict posts performance and indeed should be done. When doing so, you have to remember a couple of things:
1. Be consistent with the experiment and the real publishing time/days – If you want to get the same performance try to publish and promote your post at the same time and day of the week the experiment was done. It’s not always possible, but would give you best results.
2. Experiments predict success chances not future results – The results that you get for your experiments are actually saying what are the chances of a post to have a good readership. It doesn’t mean it will actually happen. Therefore you do the experiment; to increase your Chances of success. Of course, the more followers you have the more accurate your experiment is.
3. Try to aim to certain hours of the day you know most of your followers are online – I have this set up for Facebook but not to the other social networks yet. Some of this data is hard to find. It would be interesting to analyze back all publishing on these times( 1:34PM & 2:33PM) to see if I got the most engagement on linkedin on these certain hours, or it was just something happened by chance.
Now over to you?
Where are your followers? Where are they most engaged? On which channels? a page? a group? What kind of experiment you do to check what works and what doesn’t? I’m interested to hear and share with others. Add it in the comments.
Image credit by christopher.woo