Everyday Data Vis — Instagram Ads
Last month, I recorded every compliment I got on an aspect of my appearance. This time, though, we’re going to be looking at something a bit less ego-centric. We’re going to be tackling Instagram ads.
The methodology
During October, I made an effort to see all the posts in my feed so that I could be exposed to, theoretically, all the ads! Instagram lets you know when you’ve caught up, so I would scroll until I hit this little icon.
During the month of October, I was following about 480 people. I made a conscious effort to not do mass follow/unfollow sprees and I would like posts as normal. This time, though, I screenshotted every single ad I saw in my feed or between stories.
Originally, I set up a spreadsheet so that I could record information I was interested in, which was: date, time of day, whether it was a feed or story ad, whether it was an image or a video, whether it was a single or carousel of ads, and the categories and sub-categories. I very quickly learned that this would be exhausting, since I was seeing around 100 ads per day.
In the end, I imported all the images into Eagle, which is a program I use to collate all my design inspiration and used their mass tagging functionality to tackle everything. Still slow, but significantly less painful!
The prediction
Going into this, I actually was pretty fond of Instagram ads. I mean, they usually were pretty well-targeted, and half the time they were so nice that I just thought they were normal posts in my feed!
Like a lot of others, I did have slight suspicions that Instagram was eavesdropping on me a bit. I’d open up the app soon after having a conversation and find an ad about that exact same thing — that can’t be a coincidence, right?
Over 31 days, I was shown 2,749 ads from 1,255 companies. This averages out to over 88 ads per day, and 2 ads per company.
The overwhelming majority of ads were in my feed, but that’s not particularly surprising since I’ve muted stories for almost everyone that I follow.
I also had a look at which format (story vs. feed) was more likely to show multiple ads in one — a carousel. In feed, 733 ads had multiple tiles, or 28%, compared to just 11% of stories.
My biggest targeter… was Instagram.
Yeah. I got the most ads on Instagram for… Instagram.
20 companies showed me ads more than 10 times in the month, with Instagram topping the list at 29 ads — pretty much one per day!
4 of the brands were for mobile games, and 11 were in fashion or beauty categories. And out of the 20 brands, 4 of them were companies I already follow!
So after looking at all of the ads, I’m not sure Instagram knows me as well as I thought it did. Maybe it’s not monitoring my browsing activity very closely, and maybe it’s not listening to me — or maybe it’s just doing a really poor job of it. But really, I think it actually might just be the Baader-Meinhof phenomenon. Also known as Frequency Bias, this is the idea that once you learn about something, you start to see it ‘everywhere.’ And it basically boils down to two things:
- We’re good at tuning out information that’s not relevant — so we may have been exposed to something before and just not made a conscious note of it.
- We tend to prioritise information that confirms our own beliefs.
A few years ago, I found out about Deliveroo — and suddenly the next day I started seeing the takeaway bags everywhere. While there is a chance that it didn’t take off until the day after I discovered them… it’s probably more likely that I was just tuning them out before.
So done with this.
Coolest finding
The sheer number of ads was really intense — I expected a lot, but not quite this many. I’m now a lot more aware of the frequency when I use the app now and honestly it’s a lot less fun to browse now! Possibly a good thing…
Findings I didn’t get to
I ran out of time while processing this data (I mean, it’s almost time for November’s data by the time this is out!) and so didn’t get to find all the insights I was hoping to.
Some things I thought would be cool:
- Looking for trends for themes (eg. wedding, environmentally friendly, LGBT)
- Measuring video vs static image ratios
- Finding dominant colours across ads
Most difficult thing
Oh man, I knew that Instagram was ad-heavy, but I didn’t realise it would be seriously this much effort. Not only was catching up on my feed every day quite a bit of work, but processing all that data manually was insanely time-consuming!
I talked to a few pals about ways to automate the tagging process for ads, but unfortunately we didn’t have any amazing machine learning skills. Plus, a lot of the ads I saw wouldn’t even mention the type of product in the post, or the image would be unrelated, so I probably would’ve ended up having to check everything anyway.
Another thing I wanted to do at the end of this experiment was look at the way I’d been profiled for advertising on Instagram, to see if I could figure out what would cause the types of ads I was seeing. Unfortunately, my account says I don’t have any ad interests, so we may never know.
And a side effect of this experiment is that using Instagram became such a chore. Opening it to post something spontaneously was something I couldn’t really do, since I’d need to then scroll down to catch up on my feed!
What’s next?
Hopefully something that’s a bit less tiring to process, but I must admit I do love having all this information! I do have a bit of a different perception of Instagram ads after this.
Thanks for reading! Give this post a 👏 if you enjoyed it, and feel free to say hi to me on LinkedIn 🎉
