A Peek Under The Algorithmic Hood

After months of constant changes, glitches, and so-called improvements to Instagram, last week’s crash (July 3rd, 2019) provided unique insight into the algorithm behind the platform and how its artificial intelligence detects and categorizes users’ shared content. Potentially more damaging than the crash itself is what it revealed to users across the globe:

Captured in the screenshot above is a desktop view (Chrome) of Instagram shortly after the platform crashed for several hours. For millions of users, uploaded imagery did not appear on the site. Instead, meta data from that imagery bled through the screen and enabled us to understand a bit more about the factors the platform picks up on.

Before the crash (and subsequent peek under Instagram’s hood), many believed the algorithm was mostly focused on colors (sunsets, bright tones, etc.). The image above shows a different landscape — one where the algorithm is searching for the number of people in a photo, the photo’s setting (indoors or outdoors), and even some unique identifiers, like trees or dogs.

While it may be a stretch to say this type of insight will change the way you post or plan your content, at least we now have one more piece of the never-ending puzzle to consider.

Do more people in a photo make a post perform better? Is there a sweet-spot performance target of 3-5 people? Indoor vs outdoor? As we test, we’ll keep you posted. Feel free to share your experiences as well!

Back to Thoughts