How is the "seasonal behaviour" computed?

The graph displayed in each Trend Details page shows how a trend’s visibility has evolved on social media since 2017.

To make this data easier to interpret, each point on the graph is normalized. This normalization is handled by an internal module developed by Heuritech’s data science team to correct for four common biases:

• Shifts in posting habits
  For example, users might post more often over time, creating an illusion of growth not related to the trend itself.

• Changes in our panels
  If certain accounts stop posting or leave a platform, it may cause artificial drops in volume that don’t reflect real trend movement.

• Platform seasonality
  People tend to post more in certain seasons (such as during summer), which can affect visibility patterns.

• Other social media biases
  For instance, an increase in selfie-style content can reduce the number of visible clothing items like shoes.

By correcting for these factors, we ensure the seasonal behaviour graph reflects the trend’s actual visibility more accurately over time.