Algorithms are already highlighting products to us based on data collected about our purchasing and browsing habits, whether individually or as a demographic segment. In future, the way we discover products will be both more reflective of natural shopping behaviour and more technologically dependent than ever.
Product discovery means someone coming across or being shown a product she’s interested in buying. It isn’t the same as finding a product she’s looking for through search engines (although search engines are often used as product discovery tools) – because in that case she already knows what she wants. We’re talking about discovery, the moment of “ooh, I want that”.
This has happened in plenty of different ways in traditional retail. Perhaps most obviously, part of the enjoyment of going out shopping is in discovering something new and exciting to own. Additionally, print media and advertisements were a primary form of product discovery for decades and are still relevant places for shoppers to find new things to want. Finally, word of mouth and seeing a friend or peer having/using/wearing a product is and will remain an essential way in which we find new things to buy.
Digital product discovery
All of the above are still common routes to discovering a product, but to them we can add digital options. Browsing the high street or a shopping centre is increasingly being replaced or supplemented with the use of marketplaces – both are environments where consumers expect a majority of the relevant brands to exist to represent their products. Social media has transformed the way we recommend, review, and share products, and digital advertising that tracks users across the web is more personal (and potentially creepy) than any magazine ad.
Digital product discovery has a problem though – it’s not as fun. Browsing ten pages of search results on Amazon (where 55% of product searches now start) doesn’t have the same charm as looking through the racks in a handful of my favourite high street shops. It’s so arduous that for all its dominance in product searches, Amazon shoppers almost never get past one page of search results – only 30% of shoppers make it to page 2.
So people using Amazon aren’t typically discovering products – they’re looking for something they already want. This is obviously a generalisation, but one supported by the data. Where do we as digital shoppers go for actual product discovery, then?
Finding products on social media
The likely suspect is Instagram. The social image sharing site is saving Facebook some major user loss and data breach-related blushes at the moment, thanks to its strong performance and growth. However, while Instagram is experimenting with product tagging in photos and shoppable links, it’s hard to assess the success and power of these – is shopping really what audiences want to do here?
The likely answer is yes – anecdotally, many brands have built cult followings on the platform and, through that community relationship, they’ve introduced their products to millions. This is the first major shift we’re seeing in product discovery – in the social commerce world, it’s a shared experience first and foremost. Relationship building and community interaction lays the groundwork for products to be placed into users feeds.
Machine learning creates live recommendations
On the other end of the spectrum, Amazon is experimenting with a technical way to help users discover products they’re interested in. Amazon Scout has shoppers like or dislike product images and draws on its catalogue to populate and refresh an array of 8 options, with options changing every time the user likes or dislikes an image.
The tool currently only functions for a subset of Amazon’s catalogue, but it’s not hard to imagine the potential it has to shape product discovery in all sorts of interesting ways. Shoppers don’t want to play a guessing game about the right keyword to deploy to get to the product they want, and they don’t want to crawl through results pages looking either. An interactive, visual and customisable discovery tool helps Amazon return some of the fun of shopping into the experience.
In terms of visually oriented discovery and curation, Amazon is borrowing philosophically from Pinterest. This makes sense, as we wrote earlier in the week:
93% of Pinterest users use the platform to plan purchases
2,000,000 people pin product pins daily
Amazon’s part-automation is the clever part – by suggesting based on the increasingly deep data a user provides as they like and dislike, Scout is converting a massive dataset (Amazon’s catalogue) into a much smaller, tailored catalogue for the user. Conceivably Amazon could introduce incentives or gamification into this experience, in order to get more consumer behaviour data to refine the machine learning behind Scout.
There’s also no reason why this type of innovation would be limited to Amazon. What’s required is to train a machine learning algorithm to understand visual product attributes and weight different attributes intelligently as it learns more about the preference of the user. Product data quality and structure is absolutely essential to this new form of product discovery.