On first hearing the term ‘image analysis’, it’s easy to think immediately of software that can identify objects in photographs. But while static image analysis is a vital application of computer vision, and one that offers immense value, it’s only a part of a bigger picture.
For marketers in particular, some of the most compelling use cases are centered on analyzing the living and breathing consumer, in order to improve marketing performance (and its measurement), create stronger personal connections between customers and brands, and reduce costs.
This article explores some marketing use cases for real-time visual analysis of people, along with the ones that involve analyzing photographic images. Each of these use cases offers direct benefits to brand owners and creative agencies by way of improved performance measurement, reduced costs, and the creation of more personalized, engaging experiences for consumers.
Making Marketing More Personal
Computer vision and image analysis technologies may have a way to go before they can even come close to the human levels of visual perception, but a fair degree of accuracy is already possible in certain scenarios, such as visual analysis of gender and age groups. That’s a level of precision sufficient to make new inroads into marketing personalization.
The McDonalds burger empire is just one enterprise investigating this marketing approach. The fast-food giant has plans to open self-service kiosks equipped with cameras and image analysis software. The kiosks will use computer vision to identify customers by their gender and approximate age and then recommend menu items based on this visual data.
Personalized Window Shopping
Aside from McDonalds, other retail companies are working with computer vision consultants and developers to increase personalization through image analysis. Camera-equipped screens in storefronts will soon be capturing and analyzing data about passers-by—or perhaps more specifically, window-shoppers.
When a window shopper stops to check out the screen displays, she will see the content change to show the items or products that the software perceives as relevant to that individual based on the analyzed visual data. In an apparel store, for example, a window-shopper may see images of garments that complement what she is wearing. Could there be any better way of grabbing window shoppers’ attention and turning them into impulse buyers?
Understanding Consumer Attention and Sentiment
Brand owners will soon be able to use image analysis to capture improved metrics about how advertising is perceived by viewers. Again, while the technology may not yet be sensitive enough to recognize all the subtleties of facial expressions and body language, computer vision can certainly discern between a smile, a blank expression or a frown.
This type of analysis can be a game-changer for enterprises wishing to assess TV and video advertising effectiveness. The conventional way to gather such information is via surveys, in which ad viewers express their sentiments about the content they see. The weakness of this approach though, is simply that humans are notoriously unreliable when it comes to articulating their impressions.
Smiles Don’t Lie
The reality of survey fallibility was evidenced when one particular advertisement, the subject of a 2016 study by Omnicom Media Group, was ranked 55th out of 63 advertisements on the 2016 USA Today Ad Meter List. That result was based on viewers’ stated opinions. However, the ad was then shown to a sample of viewers, whose emotional responses were measured using facial image analysis. The findings belied the USA Today ranking, as the ad raised more smiles among the viewers than did other productions ranked much higher on the Ad Meter list.
The takeaway from this study is that the capture of viewers’ spontaneous emotional reactions can provide more accurate data than self-reporting, at least when it comes to gauging the impact of audio-visual advertising. While it may not yet be clear how and where to utilize image analysis for this purpose, the possibilities are being taken seriously and it’s likely that the technology will soon become an important source of marketing performance data.
Accurate Analysis for Outdoor Advertising
Of all traditional methods of advertising, outdoor ads represent the only format still enjoying growth in terms of spending. In 2017, out-of-home advertising sales rose by 2.7%, according to the data from the Outdoor Advertising Association of America.
The gradual transformation of outdoor signage from static to dynamic digital displays is with no doubt helping to maintain public interest in this advertising format. The only problem with signs and billboards though, is the absence of clicks that can be counted to measure advertising performance.
Indeed, online marketing offers an infinitely greater degree of performance feedback than outdoor advertising, but image analysis could be the breakthrough to level this particular playing field. Small, inexpensive cameras mounted on or close to an outdoor signage, integrated with image analysis applications, could become the source of feedback on a range of performance factors, including:
- How many people walked past the ad signage
- How many people stopped to look at the advertisement
- How long people spent looking at the advertisement
- Demographics (such as age and gender) of people who stopped to look at the signage
According to a Marketing Land article, this outdoor equivalent of tracking click-through rates has already been deployed in tests by creative agencies such as M&C Saatchi, who believe it can support advertisers in making outdoor signage even more dynamic, perhaps enabling them to make real-time changes in the displayed content based on audience response.
No Time-Consuming Tagging
Another benefit that marketing teams might receive from image analysis is a reduction in the workload associated with displaying products on ecommerce websites. It currently takes plenty of effort to publish product images, since every image must be made identifiable by means of tags.
For many ecommerce enterprises, image tagging can add up to a substantial amount of time and therefore labor costs. It can also present semantic challenges for consumers when they search for products. By enabling visual search tools based on the deep learning technology, digital marketers will no longer need to assign text-based tags manually to product images, and consumers will no longer need to struggle to find the right words for their searches.
Seeing Beyond Semantics
As image analysis programs can scan images and identify not only the contents but also the context, visual searches powered by image analysis will allow online shoppers to search for products by selecting or uploading images similar to the items they wish to find. This is especially useful for products with subjective descriptions, which might be labeled with any out of a range of tags, depending on the jargon used by the merchant.
Ultimately then, this particular use case of image analysis will offer the double benefit of making searches easier for shoppers to conduct, and reducing the costs involved in creating and maintaining online product catalogs.
Image Analysis: A New Dimension in Data-Driven Marketing
The image analysis use cases described in this article are exciting. They may also be but a few of the more obvious possibilities awaiting marketers who set foot into this new dimension of physical/digital interaction.
In the same way that technologies like augmented reality, sensors, and beacons have blurred the lines between ecommerce and traditional retail, so cameras and image analysis may now play a similar role in merging digital and traditional forms of advertising media (which, let’s face it, are largely visual) to create a new landscape for data-driven marketing.
Time will tell if its influence will be that revolutionary, but it does seem likely that image analysis will reshape the way brands strengthen their identities, heighten consumer awareness, and attract new followers—all while measuring performance more accurately and reaching new levels of cost efficiency.