Problems with internal site searches for small and large e-commerce. Which are easier to resolve?
Whether the e-commerce website has a small or large number of products, it will have a different range of problems. Some of them can be solved by an advanced site search completely, while others have only a partial solution and even giants like Amazon and eBay still struggle with them.
Problems of Large-scale E-commerce
1) Too many products to find the right one
When your e-commerce website sells thousands of products in one category, it might be difficult for customers to find their desired item. Even if they use filters to limit the search results, sometimes it’s not enough to reduce the staggering page counts that online shoppers find themselves facing.
Having a large assortment of products might also lead to customer’s choice overload. Even if your e-store has the products your shopper needs, he or she might become paralyzed by the sheer volume of options available.
There is no definite solution to this problem, but cutting the number of products and enhancing faceted filtering on your website can help.
2) Too many attributes
With a broad number of products comes another problem - redundant attributes. It’s difficult to manage so many attributes, decide which ones to display and find out if there are duplicate values or not. Some customers are overwhelmed with the options to choose from. Others, on the contrary, become too specific about what they need, tick too many filters, and get no results. Both types of customers end up with the same frustrating feeling and leave the website.
Also, some product catalogs might contain misspelled variations of a common term, abbreviation, or jargon name of a product. This leads to an even larger number of misaligned attributes.
Transforming and clarifying data is still a big challenge for the industry. With some manual work, we can reduce the number of attributes to make them more consistent, but it’s not a universal solution.
3) Data harmonization
Usually, one e-commerce store has several suppliers, which leads to the differences in catalog data. The names of the products and brands might vary, and there may also be problems with differences in measurements or sizing systems, etc. Through machine learning capabilities, we manage to improve the quality of data. Data harmonization helps to turn two or more product definitions into a single one. For example, if a store receives three different clothing catalogs in EU, UK, and US sizes, we can unify them to US-only sizes. Even so, data delivery and transformation stay overly complicated. The more products a store has, the more it is prone to mistakes in data, over-categorization, and unnecessary filtering.
4) Keywords overlap for different product categories
Sometimes people use the same words to look for completely different things. When a user types “ski gear,” there are many products that can be suitable: helmets, goggles, ski boots or pants, etc. It’s difficult for the search engine to identify the intention of the shopper. When he or she searches for a “golf club,” where should the engine redirect the user – to the “Sport accessory” or “Hotels and Resorts” category? Even as machine learning advances, this is a case that is hard to resolve.
Problems of Small-scale E-commerce
1) Too few attributes to create filters
Owners of small e-commerce websites often manage data themselves or upload catalogs that tend to miss some nice-to-have details. If there are not enough product specifications it’s hard to define what filters will help users narrow down the search results and find the products they need.
But this problem can be resolved with attribute extraction. Pulling information out from product data and configuring attributes will help to significantly improve customers’ experiences on the website.
2) Need for a great search experience with a small budget
Custom development takes a lot of time, testing, and money. Building an in-house site search is an expensive venture, and there is no guarantee that it will work as well as the e-merchant needs it to. The good news is there are plenty of site search and filtering apps for small and middle businesses on the market. Ready-made solutions are better options for e-tailers. They offer technology that has already been tested and enhanced based on the experience of dozens of other e-commerce websites.
3) Not enough traffic for regular machine learning algorithms
Small e-commerce websites can’t boast the same level of website visitors as e-commerce giants have. Less traffic means less data on customers’ behavior. The search engine can’t utilize the “power of the crowd” and learn how visitors of this particular online store like to shop.
Third-party search providers come to the rescue here, too. They have data collected on how users behave while shopping for different things and can use the knowledge specific to your industry to configure the search engine.
To Sum Up
As you see, having a small but well-organized e-commerce website is sometimes better than selling thousands of products. Most small stores' problems are handled with reasonable costs. On the contrary, large e-commerce websites are troubled with things that can’t be resolved with the current state of technology development.