Skip to main content
Industry Contributor 4 May 2021 - 3 min read

Visual search: How do people see your brand?  

By Dan Benton - Neo Media World

Dan Benton, GM of GroupM performance marketing agency Neo Media World, casts an eye over the vast opportunities for marketers in the emerging world of visual search.

What you need to know:

  • If they are not already, all marketers should be getting to grips with visual search.
  • Younger audiences are driving growth.
  • It's the one search area not tied up by Google.
  • Key players and market opportunities outlined below.

Visual search is a rapidly growing category that should be on the radar of all marketers. Visual search differs from “traditional” keyword driven image searches as the input for the query is an image uploaded by a user – typically a photo from a smart phone.

This is made possible by platforms’ access to cheap and powerful computer vision and ML models which take an uploaded image input and then return images of “similar” objects.  The results are surprisingly accurate and are only set to improve as the providers continually train their models.  

Visual is popular with younger audiences with one piece of research showing the majority of Gen Z and millennials want visual search options when shopping. Unsurprisingly visual search is particularly relevant for categories that have an aesthetic, think, fashion, home décor and furnishings, food, travel and auto.

Unlike voice search in Australia, which is largely sewn up by Google, who have a stronghold on core search, a dominant position in smart speakers and own the Android o/s, visual search is still wide open. There are currently several players jostling to capture market share.

The visual players:

  • Pinterest who’ve previously reported receiving over 600 million+ visual searches per month via its “lens” feature and can identify a staggering 2.5 billion objects. With a monthly Australian audience of almost 8m users, Pinterest is a clear contender.  
  • Google with its lens solution which is found in a standalone lens app for android, within the Google photos app for android and in the Google app for iOS. The lens app has capabilities beyond product search including the ability to translate text, identify plants and animals, help solve homework problems and provide information on places (pulling in images and reviews from maps data). Google have purported to have categorised over a billion images.
  • Snapchat with its camera search app connecting users to product search on Amazon.
  • Amazon StyleSnap that allows users to upload an image and get outfit inspiration to shop from the fashion items listed on Amazon.  

For marketers thinking about trying to tap into this latent and growing opportunity, there are a couple of potential paths to take.

Seeing the marketing opportunities

  • Testing media investment in visually driven ad formats. For instance, uploading a product catalogue to Pinterest and Google and running shopping ads. Beyond driving sales, this likely has a secondary benefit of providing structured ingestible data to the platforms to help them train their models (which may aid organic discovery).
  • Optimising the images that are hosted on a brands site to try and improve the organic ranking in visual search. The optimisation approaches vary slightly for each platform but at a high level the objective is make it easier for a platform to “understand” each image by optimising the image, the image attributes, and other relevant data elements.
  • Building a visual search capability as part of a brand site or app. A local example of this is General Pants who have set up the ability to upload a photo to try and “shop the look”. While the results are currently limited due to the local inventory it is a clear example of how visual search can assist the user journey.

The future of visual search is exciting, as the major platforms train and refine their models and assemble even larger databases of objects we will likely see visual search embedded in more applications.

Visualising use cases

  • AR where image recognition can pull in data on anything a user is looking / pointing their device at; think social profiles, plants, animals, buildings, reviews, transport timetables etc. As a sidenote Microsoft’s recent $22 billion deal to supply AR headsets to the US Army hints at the military use cases of visual.
  • Visual making almost any image “shoppable” as users are served an endless aisle of relevant products.
  • Medical, industrial, and agricultural applications; think symptom diagnosis, manufacturing quality inspection and using visual search to identify which crops that are perfectly ripe to harvest.

The scope and applications for visual search are already wide ranging. It’s an exciting evolving field and the marketers should be thinking about the role of visual in their consumer journeys and ultimately what their brand looks like.

 

What do you think?

Search Mi3 Articles