I believe best-in-class product data is a big competitive advantage for any E-Commerce business for a number of reasons I’ll outline below based on talking to 10+ such companies every week for the last 6 months.
Online, there is no sales staff to explain a product to potential customers, build trust and help them to find the right product fast. This means I need to convince a customer to search, evaluate and buy a product from me based only on the online available information.
Today’s customers like you and me are in complete control of the buying process and are likely to go and stick with the “go-to-place” that is most convenient and trusted for them to find and evaluate products.
“So what is the big deal here?” many people ask now as bringing products online should not be a big thing in 2017.
Unfortunately, the reality is quite different. There are some product data standards out there, but they fall short in terms of precision, coverage, and semantics. Essentially, there is no unified global product data standard, but a huge variety of different file types (i.a. CSV, XML, JSON, PRICAT, EDIFCACT), and every organization names their own product attributes and maintain their own product data model (if there even is one).
Common approaches I see to manage product master data preparation today are:
1) Templates: Force suppliers to fill out custom templates to gain data in the desired structure and format
2) Manpower: Pay employees to manually integrate & transform incoming data to make it “online-ready”
3) Traditional Systems: Implement rule-based systems to support manual data preparation
All of the above have their advantages, but in most cases none of them enable the on-boarding of new products fast (e.g. in <2 days overall upon data receipt), in high-quality (e.g. usage of 90%+ of the existing information) at low operational costs (let’s say <1$ per SKU) for these reasons:
1) Templates: Unless you have Amazon-like market power as a business, suppliers are reluctant to invest efforts to fill out custom templates only to list their products in 1 more shop.
2) Manpower: It does not scale. Expanding from 100’000 to 1’000’000 products online by expanding your content team from 10 to 50+ people is a bad idea for your operational costs
3) Traditional Systems: Complexity and maintenance efforts increase over time, do not adapt to changing data structures and do not handle “fuzzy” logic (handling typos, similar products etc.)
So how can the challenges above be mastered in order to have new products online fast and of high quality? I think we have a great opportunity to embrace artificial intelligence (AI) in this area as it enables consistent, adaptive, widely automated and scalable product master data preparation.
Glad to receive your comments on how you think about today’s product data preparation challenges and how they can be overcome.