As brands are starting to realize that their social activities tie to the larger business impact, there is an ever-growing focus on social media spend. Brands want to connect and engage with their consumers in real time and to understand who their customers are and how they see their brand.
According to Gartner, in 2013 the average digital marketing spend for companies was a little under 25%. However, the rise in social media sharing and the engagement through content has increased this spend to 35% in 2016 and this figure is steadily rising. But how are marketers monitoring this spend?
Social media monitoring has exploded and companies in this area are saturating the market, giving brands the opportunity to choose like never before. But are brands choosing their provider wisely and is the data they are receiving enough to make those ever-crucial marketing decisions?
Here are three ways you could be throwing your money away on social media monitoring.
1) Focusing on text-only social mentions
In a world where over 80% of social media posts contain images or video, we must evolve our monitoring tools to work with this crucial aspect of social.
Monitoring only text is like looking at one minor part of your audience and deciding that they’re the voice for thousands or even millions of others.
Instagram, Pinterest and Snapchat have changed how we communicate with each other and visual content is key. With over 400 million monthly active users on Instagram, brands are starting to realise that the opportunities to connect and engage with their customers using these social media platforms are enormous.
However, brands are missing out on learning what is being said about them. Over 85% of logos that appear in images do not include any mention of the brand in the accompanying text.
2) Relying on “good enough” visual analytics technologies
So you’ve chosen to monitor your brand through text and image, but what do you need to look out for when choosing the tools to do this?
There are two important measures to look for when assessing a visual listening tool: precision and recall (explained below). If these measures are insufficiently high – or worse, if the vendor does not report them at all – you will likely not receive useful data from the tool.
High precision means that when a visual listening tool reports that it has detected a logo, it is very likely that the logo is indeed present in the image. Conversely, low precision means that you will be plagued by “false positives”.
A high hit rate, also referred to as high recall, tells you that the visual listening software will detect your logo almost every time it appears on social media. This is arguably more important than high precision because it allows you to discover the popularity of your brand’s logo compared to those of your competitors.
3) Measuring using biased data
You’ve chosen to monitor your brand by both text and image. You’ve chosen a social analytics company that offer the best in hit rate and precision. What else could possibly be needed to ensure that the data you receive is not only accurate but also as unbiased as possible?
To reduce bias, raw data must first be collected from platforms such as Instagram, Twitter and Pinterest. The raw data should be just that, raw, and it should be filtered as little as possible but gathered in a cost-effective way.
Once we understand that, we should look at the factors that can affect the retrieval of the data so that we can maximise the size of our sample set.
Visual must always be the primary source when using visual listening, i.e., we must not rely on text searches to find relevant images. This sounds obvious, but often keywords are used to filter data before the images are retrieved. This can lead to a failure to find all images which include your brand’s logo, resulting in incomplete data.
All logos should have an equal chance of being detected, regardless of what the logo looks like, or whether it is from a small local brand or a multinational giant. ‘Tricky’ logos such asthe Nike “swoosh” or the text-only Staples logo shouldn’t be treated any differently to that of a common Starbucks logo.
To ensure the most cost-effective way to monitor your brand, a decision should be made as to where the brand is most active and the audience that is most responsive to engagement. Brands should be able to monitor city-wide or even country-wide easily.
The final factor that is important when ensuring the data is unbiased is the timeframe in which the data is monitored, for example monitoring a sponsorship event. The event should not just be monitored on the day in question, but also in the days before and after.
Recently, we monitored brands during the Super Bowl. Unlike other visual listening companies, LogoGrab monitored this event without the use of keywords. Instead of choosing just the day of the event, LogoGrab measured brand activity in social media during the run-up, at the time of the game, and during the wind-down. This is a useful way in which brands can measure not only their brand ROI of the event sponsorship, but also their brand’s share of voice versus their competitors’ surrounding the event.
With brands investing more in visual content in 2016 than ever before, and the rise of the influencer taking over previous methods of product endorsement, brands can now not only connect and engage with their audience, but they can also monitor customer moments that may otherwise have been missed.