The ecommerce space is becoming more competitive each year. A key success factor that increases the user conversion rate in your shop and puts your business ahead of your competition is the quality of your product master data. However, achieving high quality product data can be a daunting task as there is a lot of messy product data out there.
Professional data management
Professional product data management does not only include a structured product onboarding process but also a consistent product data model that defines how you describe products to your customers. The maintenance of product data sometimes requires very domain-specific knowledge, which is why a joint approach between IT and the according category or product managers is highly recommended.
Is it really worth the effort?
In reality, product data quality varies greatly between suppliers. One supplier might give you the product data in the quality you need, while another one will only provide you with limited product data where even key information is missing. This can lead to lots of manual work over long periods of time, involving countless iterations and feedback loops. When set up professionally, product data management has a huge positive effect on your business. You can:
- onboard new suppliers faster and sell more products
- increase your conversion rate by offering improved search filters
- reduce your operating costs for data management and free up team members’ capacity
Where to get the data from
It’s important to get the best possible product data available to power your shop. Available options are getting the data from the manufacturers, suppliers, distributors, content providers, or obtaining product data from other web pages. The best sources usually are the suppliers and manufacturers themselves. Use the information they provide before you purchase expensive product data from content providers.
If you let your suppliers fill in product data templates you define, quite often, you receive low-quality data, forcing you to improve and enrich the data yourself which costs you time and money. Why not let suppliers send what they have so you receive all available product information, instead of losing information in incompletely and wrongly filled-in templates? Even though this means you have more data preparation effort to be done on your side. In the end, best-in-class product data is your competitive advantage, so rather see this as an investment than a cost driver.
8 steps to better and richer product data
At Onedot, we believe that preparing your product master data is ideally conducted in these 8 steps (data examples here):
1. File Ingestion
In the first step, product data from third-party sources need to be loaded into your product data management system so you can start preparing the data. We recommend transforming all product data feeds into a flat data table, so you can work across feeds and not on a single feed at once. Make sure you track the source of the product data in your system as this will allow you to prepare the data in a source-specific way, which is often needed. Once this is done, gathering and merging all product data feeds into a single coherent structure can start.
2. Attribute extraction
Lots of product data is unstructured and product attributes are hidden in product titles or product descriptions. Having all relevant attributes available and in a structured format is an important step so you can: Lots of product data is unstructured and product attributes are hidden in product titles or product descriptions. Having all relevant attributes available and in a structured format is an important step so you can:
- enhance your faceted search in your shop
- show all available product variants to a product
- publish richer product feeds to search engines, online marketplaces and comparison platforms
Make sure that you identify the relevant attributes in the text, extract their values, and save them as dedicated attributes for further use.
Some of your shop visitors probably use the search bar to find the products they need. Many visitors are using the category tree to navigate to the product type of their choice. Therefore, a consistent and complete categorisation of all your products is essential. If not all products are categorised, or some products are not categorised correctly, a shop visitor can’t find these products and therefore can’t buy them.
It is up to you how granular you want to set up the categories. Keep in mind that the more granular the categories are, the more specifically a user can find products. The most granular level in categorisation is normally the product type level. We recommend going down to that level only for categories where you have lots of products. To say it in other words, we don’t think it makes sense if you have a category with let’s say less than 20 products. Start thinking about splitting a category into several categories if all your products don’t fit on 2 to 3 shop pages anymore. We also do not recommend using product characteristics like colour or size to define a new category, as this limits further use of that category and causes lots of pseudo-categories to be created.
4. Schema mapping
Each product has a set of potential attributes which is usually different for each category. It’s therefore a prerequisite to identify the category of each product before you can map the supplier product attributes to the attributes specified by your data schema.
Not all supplier product attributes will have a corresponding attribute in your data schema. Cross check if the supplier product attribute is worth integrating in your structure or the attribute can be ignored. More information is not better per se; what matters is providing relevant information. If the supplier product attribute is a valuable additional piece of information to your products, then add the attribute to your data schema.
5. Data integration
Once the schema mapping is defined, you can integrate the individual products from your supplier data feeds. All your products and attribute values need to be mapped to your specific data types. Specific formatting rules may apply. In case you have defined lists of values for specific attributes, make sure that product attribute values are correctly mapped to the corresponding value.
As a result, you will end up with a unified product data feed that can be loaded into a product information management (PIM) system or a shop system. We encourage you to go a few steps further in order to provide your customers the best possible online experience.
6. Attribute normalisation
Once a user navigates to a specific product category, it is sometimes still a challenge to identify the right product among all the others. Faceted search can help shop visitors to see how the products differ and what options they have. Faceted search should not include all available product attributes, it should only include relevant ones. Which attributes are relevant often requires business know-how of a category manager and measurements from web analytics.
7. Golden record generation
Should you receive the same products from different vendors, golden records become relevant for you. Either you show one product and list the different vendors next to it, or you list all the products by all the vendors separately. It is considered best practice to show one product and then the different vendors a product can be bought from, which requires you to merge the different product records into one while using the maximum available information from the different sources. Merging product records often gives you the chance to identify product data inconsistencies which you can address as well.
8. Product variants identification
Many products come in different variants: different colours, sizes, materials, etc. As an example, if you have a t-shirt in 8 different colours and in 5 different sizes, then you have up to 40 variations of the same product. Instead of listing all those t-shirts separately with a dedicated product page link each, you can use a single product page with configurable attributes to choose product variants from. By doing so, you significantly reduce the number of product pages the search engines need to index. Normally, the search engines don’t index all your product pages, they index only a selection of them. The more pages are indexed, the more search traffic you will get. How many pages are indexed depends on your specific page rank and other factors.
High quality product data is an opportunity to increase user conversion and gain a competitive advantage in the market. We recommend following a structured product onboarding process similar to the one described above. A structured product data preparation process helps you to turn messy supplier data into high-quality product data information, ultimately keeping your customers happy and your business soaring.
Guest blog by Bernhard Bicher, Founder and CEO of Onedot.
Onedot enables commerce companies to onboard products up to 10x faster, with 25% more information and at lower costs, by automating the entire product data preparation process.
To learn more about Onedot, visit www.onedot.com.