This post is the first in a series on the eContext API and how our clients get value out of semantic classification. To learn more about our technology and its uses, feel free to get in touch.
Retail professionals, I have a question for you:
If I go to your site and search “gym shoes”, what will I get?
A) A short list of results that include the words “gym” and “shoes”
B) A long list of results that include the words “gym” or “shoes”
C) How the hell should I know? Go ask our IT guy.
D) A healthy selection of athletic footwear.
If you answered anything other than C, I think we might have a problem.
Thanks to search titans like Google and Bing, online consumers are growing accustomed to sophisticated query response. The top search engines know what I’m looking for better than I do; they correct my spelling mistakes, automatically look for synonyms, and offer relevant follow-up keywords. As a result, an expectation is formed, and services that can’t meet this expectation take a hit.
Consumers are losing their patience for the query gap: when a retail site has what I want, but I can’t find it because I didn’t use the exact syntax necessary. Thanks to search titans like Google and Bing, online shoppers are growing accustomed to sophisticated, intuitive experience. The query gap puts an undue burden on users, forcing them to guess at alternative keywords or, worse, try a different retail site. Whether you’re losing customers or simply impeding their path-to-purchase, the smart question is: how can you provide sophisticated retail search without a Google-sized budget?
Connecting Products to People
eContext’s semantic classification is a boon to retailers for two reasons.
First, it aligns disparate syntax with a robust system of vocabulary rules. That means, even if you call them “sneakers” and your customers call them “gym shoes”, eContext knows you’re both talking about the same thing.
Second, it structures concepts into a massive category hierarchy. Even if you don’t carry the exact product a shopper is looking for, you can utilize a coherent topic structure to offer relevant alternatives.
Of course, all of this is purely theoretical so far. Here’s a quick outline of how actual eContext clients access our classification to improve retail performance:
Step One: Obtain an API Subscription
If you are interested in getting your hands dirty and seeing what this whole classification thing is all about, we encourage you to obtain a trial subscription, which offers a limited number of API calls free of charge. Beyond the trial allotment, subscribers pay-per-use, with lower CPM rates for larger volumes of data. For a refresher on the different calls available via the eContext API, check out our previous post.
Step Two: Classify Your Inventory
Product names, descriptions, image tags… go ahead and run it all through the eContext API. The call you’ll make depends on the length and format of the submitted text, but in most cases, classify/social will give you the most actionable result set. Don’t worry, it won’t take too long: eContext has clients that routinely classify one or two million inputs per day.
We’ll give you back, for each individual element, a list of the topics we were able to identify (and remember, each topic comes packaged with a category path to describe its relationship to other topics). Tag each of your products with eContext’s classifications and suddenly, your inventory just became a lot more organized.
Step Three: Classify Incoming Searches
Of course, labeling your products is only half the battle. To bridge the query gap once and for all, you’ll want to connect your site’s search functionality to the eContext API and classify shoppers’ search queries in real-time. Use the Categories/search function to ingest a keyword and automatically identify the corresponding eContext topic.
Now, since your inventory’s already been classified, your system can match queries with products that have been tagged with that same topic (or a closely-related one, depending on how broadly or narrowly you want to respond to searches.)
Ancillary options available with semantic classification:
- Make additional content searchable via the same topic-based search experience: product reviews, FAQs, instructions and how-to guides, etc.
- Utilize product classifications for an additional layer of intelligence in evaluating KPIs
- Annotate unique visitor IDs with the topics they search and/or browse, developing a profile of interests to make better product recommendations and other personalized experiences
One final note on making online retail more intuitive: we talk about “your retail site” when, really, consumers are getting used to a diverse set of shopping options. With separate desktop- and mobile-optimized destinations, virtual agents, and the advent of chatbots (not to mention Facebook Messenger’s new in-app purchasing), the nuts and bolts of consumer navigation can vary quite a bit. But the principle of placing a low burden on the user (in this case by offering seamless query response) will always be a high priority.