For this report, our goal was to discover the 100 brands that were pictured most online. This would provide social benchmarks that could help inform any organization or industry using or invested in social media. In order to achieve this, we analyzed 300 separate logos and 100 million images within a four-month time period from Dec 1st, 2017 – April 1st, 2018.
Sentiment: Sentiment is evaluated using natural language processing (NLP) techniques. Brandwatch’s NLP algorithm is among the industry’s more conservative, aiming to qualify sentiment only when a certain confidence level is breached.
Gender, Interests & Profession: Gender, Interests, & Profession are evaluated learning user account or profile information as well as machine-learning techniques.
Rules & Categories: Brandwatch’s Rules, which rely on Boolean logic, allow users to separate specific conversations into specific categories. Rules can be understood as “Queries within Queries.”
Logo Detection: Brandwatch uses an adaptive learning engine to uncover logos within an image. The technology searches for a myriad of attributes to find every type of logo placement, including curved, blurred, small and partially covered detections.
Sampling: A statistically accurate 10% sample was used to collect the data. This sample was extrapolated by 10x in the report to give an accurate estimate on what the full coverage would be.
The Brand Visibility Report analyzed 300 logos in total. The 300 logos were selected through two processes. First, we examined revenue and output lists, industry literature, and social data to compile a list, then we cross-referenced with well-known brand ranks like the Fortune 500, the Interbrand Best Global Brands, and the Social Outlook Report.
Once those companies and agencies were collected, we ranked organizations according to the volume of images they were pictured in on both Twitter and Instagram. We then identified the top 100 brands and conducted additional analysis.
All revenue data was collected from annual reports or P&L statements from the last available full year. For example this annual statement from Adidas. Collecting kit manufacturing sponsorship data from European clubs was far more difficult. There’s no single source of truth available, so we were forced to gather data from a number of aggregating news articles like this. For context, we only monitored clubs if they were sponsored by Adidas, Nike, or Puma and if we could find the finances behind the sponsorship.
For that reason, Swansea is not included (as it is sponsored by Joma) and Borussia Dortmund is not included (as we couldn’t identify the sponsorship cost).
Comparison to 2017
Astute readers may have noticed differences in data volumes between the 2018 Brand Visibility Report and the 2017 edition. Specifically, in our 2017 the top brand – McDonald’s – generated an average of 900,000 images per month. This year, Adidas generate over 6 million. So, why the difference?
There’s two key reasons. The first is the increase in the volume of images published online. On average, the amount of images shared online increases by 50% each year. This partly explains our increased volume. The seconds reason is due to Brandwatch’s data collection. Over the last 12 months we’ve dramatically increased the volume of data we crawl and the frequency we crawl. This has subsequently increased the total volume of images collected.