SortWise: Product Attribute Management with the Help of AI

Peter Brenko

Publisehd
18. sept 2023

Why did you develop the SortWise tool, and who did you want to help?

We began developing the tool to solve a long-standing and pressing issue for our existing clients. With years of experience in setting up websites and online stores for our clients and partners, we found that most companies face the same problem—organizing and importing attributes (specifications) on products that they import into their online platform/PIM from suppliers or manufacturers. In our case, we've observed this issue on the Shopamine platform, but the problem is consistent across other platforms as well.

Can you explain the problem? What is the actual issue with managing product attributes (specifications), and how have companies been coping with this issue so far?

Companies struggle with the time component and errors that occur during the manual process of organizing their product catalogs. The data received from suppliers needs to be transformed into a machine-readable format. This means that product data is stored in the platform's database in a way that each value is in its own column or in the correct format.

Why is this necessary? So that this data can be used for product filtering. This is beneficial for both the retailer and the customer. The retailer can more easily search, filter, and manage products in their digital catalog through the platform's administrative interface. Customers can more easily browse and find products in the online store.

Problems arise because, in most cases, companies manually manage these catalogs. A company has an employee or a team that regularly reviews and manages products. Each specification has to be manually entered for each item, or they are imported from pre-prepared CSV files.

Another option is to create a custom script. This involves an engineer first carefully reviewing the existing file content (CSV tables, XML files, etc.), and then programming a script to execute over the data. The script attempts to extract product attributes from the available data, clean them of unwanted characters, standardize naming, etc. However, as each supplier may have attributes in different formats, a unique script must be prepared for each supplier. Moreover, the supplier controls the data format, meaning the script stops working as soon as the document's format changes. One has to find the error, correct the script, test its functioning, etc.

For certain industries, there are providers that offer pre-organized product attributes, but according to our clients' experience, these services are quite expensive and often do not have a comprehensive range of all products or all brands that a company sells.

Can you share an example?

As an example of attribute usage for the customer, let's consider a laptop. The attributes (specifications) of the laptop include features such as screen size, type and size of hard drive, color, brand, memory, processor, etc. If the customer already knows they need a laptop with an Intel Core i7 processor, there's no need to display all other laptops that feature an Intel Core i5 processor.

So, if each laptop has clearly defined attributes, the online store or B2B portal can display these in a filter available to the user. The user then selects the desired attribute, and the online store automatically filters the product list to show only the laptops containing the searched-for attribute.

Quick and efficient searching is the foundation of a good user experience. And a good user experience boosts sales.

Where and how does the problem occur?

The issue arises when a business sells 1,000 or more products. There are at least three reasons for these challenges.

Different Data Formats: Businesses often receive data in a variety of formats, such as PDF, Excel, CSV, XML, and others. Many times, the attributes and other details within these documents are not adequately structured for import.

Multiple Suppliers: When a retailer has multiple suppliers for the same products, each one may describe attributes differently and even use different naming conventions for the same attributes (e.g., screen size, screen diagonal, monitor dimensions—same meaning but different terminology).

In addition to these issues, a third problem is tied to the variable number of specifications provided by different suppliers. For example, one supplier may send 5 specifications for a product, while another sends 15. This necessitates the aggregation of data from different sources.

Before publishing a product, all this information needs to be carefully and accurately organized, standardized, and merged. This requires a significant amount of focused work, time, and consequently, money. Once everything is properly arranged, the attributes can be loaded or imported onto the products using matching identification codes (SKU/EAN). Only then can appropriate filters be prepared, making it easier for both administrators/employees to manage the catalog and for customers to make purchases.

In which industries would companies benefit the most from using the Shopamine SortWise AI tool?

The industry doesn't play a role here; the crucial factor is the number of products. As soon as a retailer has to manage more than a few hundred or thousand products, it becomes a time-consuming task. This problem appears across all industries.

Practical Example: Textile Industry (Fashion). The problem arises when different suppliers have varying interpretations of product colors, sometimes due to branding. For instance, shades of red can have names like cherry, rubin, or merlot; shades of blue could be navy, sky, or sapphire; and shades of green could be emerald, lime, or menthol. Online shoppers are not interested in these different names for colors; they want to search for all, for example, green or blue items.

What is the Shopamine SortWise tool, and how does it solve the mentioned problems?

SortWise leverages AI algorithms to solve the highlighted issues. The tool can rename various shades of colors into their basic colors based on pre-defined primary colors. For example, it can interpret different names for shades of green like emerald, olive, and lime, and categorize them all as green. So, if a potential customer in an online store is searching for a green shirt, they can easily find all the variations and shades of green with one click using a filter.

In the case of computers, the tool can consolidate various forms of supplier-provided data files, differently named specifications, and different numbers of specifications, organizing them into a uniform format. It also automatically updates the products in the catalog.

As mentioned earlier, without such a tool, retailers spend a significant amount of time organizing the various attributes of each product. Often, this requires manually adjusting each product individually or creating custom software scripts. The latter approach provides some level of automation but is time-consuming and laborious. Additionally, scripts are not immune to changes in data over which the retailer has no control but the supplier does.

The Shopamine SortWise tool, integrated with Chat GPT, can read, compile, and properly organize all the acquired data. Initial tests show that the tool operates with 98% accuracy. This offers significant added value to companies, allowing them to reduce the time and cost of managing their digital catalog by up to 90%. Consequently, they can redirect the time and energy of their employees towards more beneficial work.

Is the tool included in the price of all packages, and can all Shopamine users use it?

The tool is not included in the price of all packages and will only be available in higher-tier packages. As we are tied to the Chat GPT business model, there will be an additional cost for the retailer, which will be based on the number of products processed.

Is this investment financially sensible?

Absolutely. According to our metrics, a retailer can save up to 90% of the time and consequently costs associated with that time by using the tool. The tool allows companies to improve the customer experience by equipping their catalog with attributes and enabling easy product filtering.

We assume there are many areas where AI will make it easier for companies to do business, especially in online sales. How will Shopamine use AI in the future to ease operations for customers?

The Shopamine platform is continuously being upgraded and improved. Our vision in this area is focused on AI tools for translating product descriptions and other information into different languages for simplified and automated sales and expansion into foreign markets. We also plan to introduce an AI tool for product categorization, which will be very beneficial for companies that have to seasonally manage large shipments of products from various suppliers.

Rešitev SortWise smo razvili v sodelovanju z implementacijskim partenrjem Zerodays