Columnist and Microsoftie Christi Olson discusses the company’s personal shopping assistant and provides some thoughts on how marketers might adapt to this new technology.
This year, I decided to up my Black Friday and Cyber Monday game by researching my purchases in advance, so I knew exactly what deals to be on the lookout for based on what I planned to buy for my friends and family. To keep myself organized, I tested a Shopping Assistant (developed by my employer as a Microsoft Garage project) to help me track the products I was researching and to organize my potential purchases.
On Black Friday, instead of getting up at the crack of dawn to fight the crowds for discounts and deals, I opted to stay at home and fill my shopping cart online. According to a survey by the National Retail Federation (NRF), I was among the 44 percent of consumers who shopped online versus the 40 percent who shopped in a physical store.
In 2016, both Black Friday and Cyber Monday smashed previous years’ online sales records, with each day respectively generating over $3 billion in online sales and hitting a new record of over $1 billion in mobile sales, according to Adobe. I joined the crowds online to gobble up the savings.
After using the personal shopping assistant throughout November, I sat down with the team that developed the assistant to learn more about how it works today, and the team’s plans for incorporating the bot framework to create a more intelligent shopping bot that can interact with consumers throughout their purchase journey.
Meanwhile, others are experimenting with this type of service, as well, with eBay beta testing its ShopBot, which works with Facebook Messenger — though ShopBot only surfaces eBay products, rather than items from across the web.
Why use a shopping assistant?
The concept behind the personal shopping assistant is simple: create a tool to help consumers track and navigate the products they are reviewing as part of their online purchase journey, regardless of device.
The pain point that the team is addressing is to help consumers remember what products they’ve evaluated throughout their decision journey. The solution the team designed is a browser extension that can be logged into for additional functionality, which collects data from the product pages viewed and allows the user to curate and organize products into shopping boards.
(Editor’s note: Google offers similar functionality through its Shortlists and browser extension, which are part of Google Shopping.)
What you need to know about shopping assistants and shopping bots
- They use structured data, like Schema.org markup, as well as identified patterns across verticals and industries to gather product information. Follow best SEO practices and implement structured data markup to help search engines understand more context for the content on your website — as well as to help new technologies like shopping assistants and bots.There are challenges with using structured data, a big one being that many companies have either incorrectly implemented it or used a customized schema protocol. Each time the shopping assistant comes across a new variation of structured data, it has to be trained on how to recognize and categorize the information. They have identified a significant number of websites that have incorrectly implemented schema protocols specifically surrounding pricing, causing incorrect values to be returned.
- It’s a big challenge matching products across websites to provide pricing and purchase information. Most product pages don’t contain a unique product identifier like an MPN or GTIN, so the team started looking at creative ways to match data.One method is to use the link to product images and product names to identify and match similar products. The challenge with this method is twofold: First, not all merchants and websites use updated product images when new products are released; second, some merchants and websites use their own unique product photos, so their content might not match to others.
- Shopping assistants can use shopping campaign feeds from the Merchant Center to provide additional information on product pricing and where a consumer can make a purchase. This ties into my Search Engine Land article from May about how data feeds can drive new search experiences. We’re just touching the tip of the iceberg for how data feeds can be used.Any traffic generated from the price comparison feature is free for advertisers and will not appear in Bing Ads shopping campaigns. Referral traffic is reported through analytics tools as a browser extension.Here is an example of price comparison on a book I’ve been looking at getting for my son:
Is there a future for shopping assistants?
When people face a pain point, there is always room for new tools to come in and alleviate the pain. While the Microsoft personal shopping assistant is still basic, it shows potential for helping individuals in the early stages of the consumer decision journey and in tracking research across devices.
It’s also still in development and has a few bugs, but the developers have been very responsive and quick to fix bugs and issues. I reported an issue with how products from Anthropologie.com were not reporting the price in the tool, and by the end of the day, they made a quick update, and the issue was fixed.
What is the future? The developers have big dreams of incorporating additional AI and machine learning components to help consumers discover new products similar to ones recently viewed or related to the curated content in their favorite boards. Many businesses are trying to find ways to incorporate the bot framework to allow consumers to interact with tools in more natural, conversational ways.
No matter how you look at it, more developers trying and testing solutions to ease consumer pains is a good thing for both users and advertisers. Part of the fun is waiting to see how developers and brands adapt to the evolving technology to create better shopping experiences.
Some opinions expressed in this article may be those of a guest author and not necessarily Marketing Land. Staff authors are listed here.