How to Measure Marketing Campaign Effectiveness with Logo Detection

How to Measure Marketing Campaign Effectiveness with Logo Detection


How to Measure Marketing Campaign Effectiveness with Logo Detection

A 3-year-old child or even younger can recognize brands by their logos and react accordingly. At the same time, companies are competing against a shorter attention span and an ever-more-crowded advertising space. In these conditions, it becomes crucial to identify just how much impact the showing of the logo has, and if it is perceived and embraced.

Automating Brand Detection

There are several reasons for automatic logo detection. Some are linked to computing the effectiveness of marketing campaigns, others relate to brand curation and crisis prevention and management. Second, there are other marketing targets like finding the right associations, identifying brand ambassadors and growing a brand organically.

A Simple Use Case

Imagine your company sponsored an important sporting event, where all the players wore t-shirts bearing your logo. The game was featured on the national TV and local news had a few hundred invited press members and bloggers, and received positive appreciation from participants.

Now it’s time to draw the line and write the report for the management explaining the impact of the sponsorship. A simple question with a complicated answer is: how many times was your logo featured in the media? Another problem is, how many people were exposed to the logo? A subsequent question could be how many people noticed the logo once exposed to it. Of course, you could use estimates, but AI can offer a more realistic answer.

How Does Logo Detection Work?

Automating brand detection means that you let the machine do the counting for you, once it is trained to recognize the logo in various angles, light conditions, and even sizes. The technology is similar to Google’s image search or could be considered a variation of OCR.

There are two ways to perform this task: region-based methods and fully convolutional ones.

The first method looks at regions which can contain objects and then scan deeper in each of them. The algorithm first detects large objects like people, cars, then categorizes them and goes further and scans for details like your logo.

The convolutional approach is much faster since it looks at the full picture and tries to put labels on everything it finds, yet it requires more resources.

Logo detection is a step forward from text and hashtag analysis since most people won’t mention the brands in an image unless they have some real incentive to do so.

Top Uses for Logo Detection

This is an emerging technology, and most likely it will find more applications as it becomes an established tool. But so far, there are some clear uses that companies can start capitalizing on immediately.

Evaluating Sponsorship

As mentioned in the example above, measuring the exposure and impact of a brand awareness campaign during sponsored events is beyond the capacities of human marketers. An AI solution is a fast and reliable way to count the number of logo appearances, the exposure time to the audience and even the quality of the exposure.

Additional measures can be introduced to assign weights to each exposure. For example, if the logo is on the back of a player who just scored, the positive association with the brand would be stronger.

Sentiment Analysis

Consumers are most likely to tag brands when they express negative feelings to draw attention to them. However, when it comes to a positive emotion, they can post images of the products or the brand logo and write a caption that doesn’t mention the brand, leaving the association to be implied from the tone of the posting, the words used, or simply emoticons.

This can turn into a very negative-biased report if there is no logo detection technology in place when it comes to analyzing social media appearances of a company’s logo.

By identifying where the logo is displayed and analyzing the surrounding context, the marketing team can produce a comprehensive and well-balanced view of the brand in the online space.

Protecting the Brand

Maybe you are not in the position to worry that your brand will be misused by a cheap manufacturer hoping to sell more products, but any misuse can hurt it and create media crisis.

As experts from InData Labs state, detecting illegal uses of your logo or variations of it can help to protect your reputation and prevent misleading associations. Since the image damage in these cases is considerable, it is important to act fast and to warn customers about the misuses. This is also useful in the case of copycat or redesigned logos.

This tool is also good to have to prevent unwanted associations with causes you don’t support but which could benefit from your brand image by putting your logo on their materials.

Brand Associations and Ambassadors

The true test of a brand’s popularity is how it is used in the real world. The logo detection technology can identify the natural associations with the brand as the consumer sees them. These associations help the company create natural and fruitful partnerships, which the market will embrace and which will boost sales.

As part of the marketing team, you are interested in the context of your product use. This provides a good base for brand communication, creating commercials and speaking to the customer using things they are familiar with.

In the era of Instagram influencers and social media marketing, it has become challenging for companies to select the right ambassadors for their products, both from celebrities and micro-influencers. To retain credibility, it is best to pick those people who are your genuine customers and turn them into ambassadors. A simple way to do this is to identify social media accounts which display your brand without being paid.

Although these are just a few of the possible applications of logo detection tools and image processing by AI, we can expect more of these to become widely used. This trend is also powered by the rising popularity of video content on all platforms, which is harder to track by humans but is easier for AI algorithms.

Image by TeroVesalainen from Pixabay