For over a year, our team has been working with select users and marketing agencies to discern how media is received when published online. Our focus relies on end user engagement with photos and videos taken from our app, noting the number of views, likes, clicks, and re-shares from Instagram, Facebook, Twitter, and Snapchat. We believe strongly in visual media as a critical tool in marketing and engagement, but feel that companies and brands do not often take the time to research how their media choices can impact engagement.
As a brief summary, we surveyed media produced during 1,427 unique events during the time period of March 2018 - March 2019. The result of this study encompassed analyzing engagement from over 20,000 posts made to social media. For posterity, we leveraged in-house APIs to continually track engagement, as well as utilized custom trained machine learning models built with TensorFlow, Caffe, and CreateML, which allowed us to focus those APIs on posts with the highest likelihood of engagement (for reference, we'll be publishing the technical details of our machine learning models in an upcoming article).
As you read on, you will find our detailed efforts related to media type. Succinctly, we found that end-user engagement increased by 22.884% when a brand opted to publish a GIF, by 31.582% when publishing a "Boomerang," and by 41.992% when publishing a slow motion video. End-user engagement decreased by 11.048% when publishing a single still image. The factors to motivate the change in engagement are complicated, but can be broken down into relevant categories to best optimize your marketing needs.
Unsurprisingly, image quality plays a role in determining the likelihood of a user engaging with content posted to social media. Our app is iPad-based, most often leveraging the front-facing "selfie" camera, which, universally, is not known for image quality that could rival that of professional cameras. Media taken on higher-end iPads (the 2017 and 2018 iPad Pro models) resulted in the highest likelihood of engagement, while media taken with the rear camera on iPad actually decreased engagement.
How is this possible? As we began to survey our partners at marketing agencies who worked alongside us to track this engagement, we found that posts that had the highest image quality, most ideal lighting conditions, and professional-looking environments resulted in end-users finding the content "over produced" and "unauthentic." Similarly, media taken with less-capable iPad models (non iPad Pro and older than 2017 models) appeared under-produced and was frequently overlooked in a crowded feed.
Many users have become familiar with the GIF format; an animated sequence of photos showed at varying intervals. While GIF media improves the likelihood of social media engagement, we have found there to be many often overlooked factors impacting engagement with media. Our focus has led us to a generalized conclusion; GIFs with 4 pictures, a delay of 0.8 seconds, and seamless loops, resulted in the highest likelihood of engagement. Seamlessly looping GIFs is not an easy feat; platforms that natively support GIFs, like Facebook and Twitter, result in the highest engagement. Instagram and Snapchat, not supporting native .gif files, result in the need to convert a GIF to a movie file first, which can create inconsistent looping experiences.
In all, our findings indicate that social media engagement increases with a GIF, but most noticeably on platforms such as Facebook and Twitter that natively support GIF files.
We refer to Boomerangs as "Burst" media in our app, but the idea is the same; a video or series of images looping forwards and backwards. Boomerangs have become incredibly popularized by Instagram, and their native approach certainly demonstrates the willingness of users to engage with such media form. Much like the above GIF data, we have found that Boomerang media receives the highest engagement on its native platform; Instagram. Similarly, the speed and length of a Boomerang can impact the user engagement. Our findings lead us to recognize that Instagram's Boomerang app creates an experience that users closely align with. The Boomerang app records a 1 second video at a set frame rate, looping the video forwards and backwards. Deviating from this experience can still result in higher engagement than still images or GIFs, but engagement only increases when sticking with the speed and length Instagram has demonstrated.
In all, our findings indicate Boomerangs receive higher social media engagement on Instagram than any other platform, with a 1 second video at a set frame rate being the most engaged.
Slow Motion Video
Across the board, slow motion video proves to result in the highest likelihood of social media engagement on all platforms. The results are not all that surprising; people love video. The GIF and the Boomerang have been around for many years; slow motion video is still gaining adoption. Most users are not even aware that their smartphones are typically capable of recording slow motion video. The newness of slow motion video, coupled with a more cinematic look, results in a medium that end-users continue to find new and modern. Brands looking to leverage the latest in popular formats tend to find the greatest success working with video, especially slow motion. Our findings have demonstrated that a roughly 10-second video (or 4-5 seconds, slowed down for slow motion) generates the highest likelihood of social engagements across all platforms.
In all, we believe slow motion video is still new and exciting for end-users. Slow motion video creates a premium aesthetic while still offering an accessible form that users can relate to. Our experience with other video types (video recordings with no further effects of processing) have garnered significantly less engagement than slow motion, encouraging the need for accessible videos that users can both relate to but do not seem over-produced.
The final factor strongly influencing user engagement relates to image size. End-users prefer to interact with media native to their chosen platform; the Instagram feed displays images in a "square" layout. Displaying images with borders, pillars, or empty space results in considerably less engagement. Similarly, platforms such as Instagram Stories display media in a 1080x1920 size; media that does not appear native to this size is often "skipped" or noted as unremarkable in a crowded space.
In all, we believe that image size should remain native to the platform being viewed. Attempts to "cut corners" or cross-post may result in less engagement on a user's preferred platform, despite best efforts to reach as wide a customer base as possible.
When working with a visual medium, we believe the above demonstrates a series of best practices to optimize user engagement. Within the scope, specifically, of visual capture experiences (I.E. "photo booths,") we find the methods of minimizing static imagery (limiting "still images") and moving toward more visually stimulating media to be most effective.
Brands should keep the expected user experience in mind and make best efforts to remain true to this form. Optimize your media for the exact size and format that the relevant platform recommends. If working with a "Boomerang," stick to the speed and length that users already have familiarity with on Instagram. If working with video, optimize your video for square aspect ratios, or 9:16 aspect ratios for Instagram Stories.
As a final note, our machine learning models started to pick up an interesting trend as we surveyed GIF and Boomerang media; minimize the use of gimmicks. We started to notice brands leveraging humorous visual effects and digital accessories (hats, glasses, necklaces, etc.) that began to decrease the likelihood of overall engagement. Why? End users found this content to be "dated" and lacking in technical capabilities as compared to more popularized apps (Snapchat, Tik Tok, Instagram, etc). As a final point, we suggest limiting use of gimmicks if increasing visual engagement is a top priority for you or your clients.
As social media trends change and new visual imagery becomes available, we plan to continue to re-train and improve our models to track online behavior. Gathering enough data to develop a behavioral trend analysis has taken over a year; we continue to do so and plan to add more partners to our study. We take user privacy very seriously, using the best methods of differential privacy to track media we know came from our software, but using randomized and unique identifiers to not connect behavior with a particular user.
If you would like to learn more about this study, or join our program for future studies, send us an e-mail at firstname.lastname@example.org. We love talking about how visual media impacts social sharing and how we can help brands create the strongest engagement.