Getting the most out of analytics
Published on 3 August 2023 by Andrew Owen (5 minutes)
Last week, I wrote about the social media platform formerly known as Twitter (TSMPFKAT). If the T and P are silent, like in tsar and pfennig, it can be pronounced sumf-kat. And I noted that I’m no longer using it to promote this blog, based on the analytics. So this week I thought I’d go into a bit more detail about the analytics I’m using and how you can use them.
I also previously mentioned that I haven’t had a like on a post to TSMPFKAT since May. I know that because the platform provides its own Tweet Analytics, which you can click on any post to get a breakdown. Additionally, these are available through the API, so you can also view them in social media management platforms like Buffer. It’s one reason the platform was so popular with advertisers, because it’s easy to measure engagement.
I recently replied to a former colleague’s post about people leaving TSMPFKAT ove the logo change. That post had:
Advertisers can use this data with a method called A/B testing to create content that’s more engaging by comparing the effectiveness of different approaches. On a darker note, when the venture capital money ran out and social networks needed to drive revenue, they used analytics to find out that highlighting content that made users angry was the best way to drive engagement.
Federated micro-blogging site Mastodon does not provide equivalent data, and this may explain why advertisers haven’t adopted it. That and a perception that it’s anti-corporate.
“Your home feed should be filled with what matters to you most, not what a corporation thinks you should see. Radically different social media, back in the hands of the people.” —Mastodon
Facebook parent company Meta has launched a TSMPFKAT-like service called Threads, tied to Instagram accounts. But due to a dispute with the EU, the platform is not available in the 27 country bloc, which may put advertisers off.
Meta provides analytics for Facebook and Instagram through the Meta Business Suite. Meta has a vast amount of detail about its users. As the saying goes: “If you’re not paying for the product, then you are the product.”
I have a Facebook page that I set up a long time ago for some of my personal projects. Looking at the analytics I can see the demographic breakdown and I can compare the reach year on year. I haven’t been posting to it lately, but it has over 300 followers, so I’ve linked my Instagram account. Now when I post with Buffer, it also gets shared to Facebook.
I’m not sure if developer relations is going to stick with TSMPFKAT, move to Mastodon or move to LinkedIn. But I think LinkedIn has got a chance to pick up some users. I suspect there’s more advanced functionality at the paid tier, but I can see likes, comments, clicks and reshares through Buffer’s interface.
But as you’ve probably guessed, the main tool I’m using is Google Analytics on the site itself. There have been privacy concerns raised about it, and alternatives are available. But I’ve got historical data that I don’t want to lose. I’ve set up my site so that it doesn’t collect personally identifiable information. I don’t display a cookie warning, but I have added a notice to the page footer.
As this blog is primarily about subjects of interest to people who work in developer relations, my target audience was the United States, India and Germany (all countries where developers mainly communicate in English). As an advocate for internationalization and localization, I always intended to add a second language to the site. I don’t have a team behind me and automated machine translation doesn’t replace links with alternatives in the target language. So I needed to pick one. Looking a the breakdown by nationality, I identified French as the language likely to add the most readers.
Looking at a snapshot of the last 28 days I can see that the top ten countries are:
The tricky part of analytics is working out how to interpret the data. I know that the top five countries all mainly use English for technical communication. I assume the same is true for South Korea. France is the main country that uses French for technical communication. Australia and Canada are also English speaking. I assume at least some technical content in the Netherlands is done in English. Which means the outlier is Japan.
But what all these countries have in common is that they are software development centers. So my takeaway is that my blog is reaching its intended audience. The other metric I look at most often is sessions. I try to post a new article at the same time on the same day every week, and the data shows that I get the most views on Thursday. That tends to support my view that putting out content every week on the same day at the same time is a good idea. But you do have to be careful about reading too much into the data.
After ensuring I’m reaching my intended audience, the next aim is to give that audience more of what it wants, based on how well received the articles are. However, because I don’t monetize the site and I’m not dependent on it for income, I have the editorial freedom to write about whatever I like, and I trust that so long as most of the content is relevant, my readers will tolerate the occasional off-topic post. Indeed, the curated list series has proven quite popular and I have some more ideas for future lists.
Also, apparently I am one of Buffer’s top users and so they have given me a referral link for anyone interested in a free account. I’m not sure if there are any perks associated with it. Maybe they just want to know where the referrals are coming from.