Skip to main content
News Releases

Study IDs Visual Characteristics That Make People ‘Like’ Images on Instagram

the image contains three photographs: on the left is a sliced avocado; in the center is a collection of clothing in a store; on the right is a well-organized collection of tableware
The photo on the left has an optimum level of feature complexity. The photo on the right has an optimum level of design complexity. The center photo has neither.

For Immediate Release

William Rand

Marketing researchers have identified the combination of characteristics that make people “like” images on the social media platform Instagram. For example, the visual complexity of images has a significant effect on whether viewers choose to engage with a social media post.

“We are increasingly able to determine whether images included in social media messages are likely to garner interest from consumers,” says William Rand, co-author of a paper on the work and an associate professor of marketing at North Carolina State University. “But a lot of the variables that we know affect public interest have very little to do with the images themselves.

“For example, a brand’s strength, and the number of followers it has on Instagram, is the strongest predictor of whether consumers will engage with an image,” Rand says. “The text accompanying an image is also important.

“We wanted to look at the role the actual image plays, focusing specifically on how the complexity of an image drives consumer engagement. This is important information for the marketing community, because it can inform decisions about what sort of images to use to continuing building a brand.”

Previous work found there are two aspects of an image that people respond to: feature complexity and design complexity. Essentially, feature complexity refers to fundamental characteristics such as color and brightness. Design complexity refers to the actual things found in an image and how they are arranged.

To begin their analysis of how viewers respond to image complexity, the researchers identified six measures that can be used to assess various aspects of image complexity:

  • Distribution of color in the image;
  • Distribution of luminance in the image;
  • Number of edges within the image;
  • Number of objects in the image;
  • Regularity of objects, which is determined by whether objects share an orientation and the extent to which they overlap; and
  • How symmetrical the arrangement of objects is.

The researchers wrote a computer program to scan images and automatically generate scores for each of the six measures. The researchers also ran a validation experiment to ensure that the image assessment program was consistent with how humans perceive complexity in images.

The researchers then created a model to determine which combination of measures was most closely associated with generating ‘likes’ on Instagram. The model accounted for confounding variables, such as the number of followers a given Instagram feed has.

For the study, the researchers fed 147,963 Instagram images and related data into the model.

“We found that all six measures are important, but there were particular patterns in which images generated the most positive feedback,” Rand says.

When it comes to feature complexity, the researchers found that there is a sweet spot right in the middle. Consumers preferred images with some diversity of light and color, but not too much and not too little. The opposite is true for design complexity. People preferred images that were either very simple or very complex.

“In practical terms, we found that you could improve the number of likes of any given image by about 3% if you applied the appropriate filter to address issues related to feature complexity,” Rand says. “That’s nothing to sneeze at, particularly since applying a filter only takes a few seconds. What’s more, our model suggests that optimizing both feature and design complexity could improve consumer engagement by about 19% – after accounting for variables such as an account’s total number of followers.

“We’re putting this out there with the idea that it can be used to inform decisions made by design professionals in the marketing sector. But we’ve made the raw code for the model available. It’s not in a user-friendly format right now, but I’m sure the right tech-savvy people could use it to create a valuable tool for the industry.”

The paper, “Simplicity is not Key: Understanding Firm-Generated Social Media Images and Consumer Liking,” is published in the International Journal of Research in Marketing. The paper was co-authored by Gijs Overgoor of the Rochester Institute of Technology; Willemijn van Dolen of the University of Amsterdam; and Masoud Mazloom of the Ferdowsi University of Mashhad.


Note to Editors: The study abstract follows.

“Simplicity is not Key: Understanding Firm-Generated Social Media Images and Consumer Liking”

Authors: Gijs Overgoor, Rochester Institute of Technology; William Rand, North Carolina State University; Willemijn van Dolen, University of Amsterdam; and Masoud Mazloom, Ferdowsi University of Mashhad

Published: Dec. 29, 2021, International Journal of Research in Marketing

DOI: 10.1016/j.ijresmar.2021.12.005

Abstract: Social media platforms are becoming increasingly important marketing channels, and recently these channels are becoming dominated by content that is not textual, but visual in nature. In this paper, we explore the relationship between the visual complexity of firm-generated imagery (FGI) and consumer liking on social media. We use previously validated image mining methods, to automatically extract interpretable visual complexity measures from images. We construct a set of six interpretable measures that are categorized as either (1) feature complexity measures (i.e., unstructured pixel-level variation; color, luminance, and edges) or (2) design complexity measures (i.e., structured design-level variation; number of objects, irregularity of object arrangement, and asymmetry of object arrangement). These measures and their interpretability are validated using a human subject experiment. Subsequently, we relate these visual complexity measures to the number of likes. The results show an inverted u-shape between feature complexity and consumer liking and a regular u-shape relationship between design complexity and consumer liking. In addition, we demonstrate that using the six individual measures that constitute feature- and design complexity provides a more nuanced view of the relationship between the unique aspects of visual complexity and consumer liking of FGI on social media than observed in previous studies that used a more aggregated measure. Overall, the automated framework presented in this paper opens up a wide range of possibilities for studying the role of visual complexity in online content.