ViSenze Blog

Making images shoppable with visual search

Posted by Oliver Tan 26-November-2015

This article written by Oliver Tan, Co-Founder and CEO at ViSenze, was originally published on CampaignAsia on 13th November 2015. 

The world of “Buyable Pins” and “Shop Now” buttons just got more exciting this week with social discovery engine, Pinterest, announcing their new “crazy-fun” visual search.

Years after visual search was implemented in ecommerce, this visual commerce feature finally comes to Pinterest and its 100 million monthly active users after its first announcement in 2014.


[Pinterest visual search now allows users to crop an object in a pinned image and find visually similar products, their information and where they can be bought.]

In the exploding visual web, this play by Pinterest underlines the growing trend towards less text search and more visual search in ecommerce. It shows the heavy significance players place on social engagement and content to drive shopping. What is increasingly clear is that the relationship between retailers, brands and consumers has fundamentally changed forever. Where it used to be brands and retailers pushing ads to reach consumers, now they are using images on social media as shoppable content.

Whilst it is still early days to see how consumers will react to visual search and Buy buttons on social media, there is evidence that in the ecommerce space visual search is quickly gaining traction.

For example, ecommerce players like Flipkart and Zalora have started using visual search as far back as 12 months ago to help online shoppers search and find products more efficiently. In a booming ecommerce world still dominated by inefficient guessing keywords, this comes as a game changer for both consumers, retailers and brands. General results across have shown better search engagement and uplift of up to 50% in conversion rate compared with keyword search, especially in fashion ecommerce.

[Flipkart allows users to take photos of products around them and find visually similar ones ready to buy within the app]

So what are the implications for marketing and advertising practitioners?

Brands have now the opportunity to make content shoppable

Content is a major driver of shopping interest. This is especially for visual domains like fashion and home decor. Brands have now the opportunity to make content shoppable. How? By identifying content that inspires interest and encourages discovery to user-generated inspirational photography. One smart innovation is using visual search on leveraged content to facilitate discovery shopping.

Visual content can provide insights about consumers

Data scientists have long leveraged text data to gain intelligence. Further data-mining visual content can provide even more valuable insights on what consumers are looking for, that text engines cannot penetrate. Image recognition engines understand what consumers are looking for in the shopping process, and associate this data with demographics and other elements to provide valuable intelligence. Imagine an intelligent algorithm that not only processes images to surface visually similar ones, but also associates which type of people like which type of products better.

Content creators can now have an avenue of monetising their media assets

Publishers and content owners can also get into the game. The advertising industry benefits now from highly engaging and relevant ads: shoppable content itself. Visual search is that new bridge that connects the content and commerce.

What are the challenges?

While companies like Pinterest develop their own machine intelligence to power such technology, there are also vendors happy to provide their APIs to companies that don’t want to roll up their sleeves and do their own R&D. Being an industry in its nascency, most vendors focus on a few verticals at a time. Therefore, it is quite challenging to provide great visual search results across all product categories at the moment. Then, the mobile and web user experience of using such technology is yet to be polished.

What about video?

Why stop at image content? Why can’t video content be a source? Afterall, CTR on video Ads are generally between 2.4x - 3.5x higher than standard banner ads. At ViSenze we bet not only on searching products in images, but also in videos. Video content is rapidly booming on various platforms and visual search built with machine intelligence can also make video content shoppable. From products in fashion shows streamings to those in user generated videos, nothing will be unsearchable.

We believe we will see massive transformations in the visual web space in the near future. Technology is rapidly evolving and more and more popular platforms are familiarising consumers with visual search. Bridging the world of content with commerce is happening and we can all benefit from it.


New Call-to-action

  Industry news,   Opinion pieces