Data Architecture

June 16, 2014 | Posted in eCommerce | By

Last week, while attending the Internet Retailer Conference in Chicago, someone asked me what the greatest challenge is with ecommerce today.   Not wanting to give a less than thoughtful answer, but also not wanting to get into the vortex of all that is right and wrong in the world of ecommerce, I simply gave a four letter word as my answer.

Data.

Inevitably, with clients that I’ve worked with in the past, or my own sites that I’ve led, the root of all issues in ecommerce always seems to come back to that four letter word.  Poor data can be the reason why conversion rates are bad, why bounce rates are bad, or for a less than ideal user experience. Incorrect data can be a reason why otherwise correct code doesn’t work properly, why faceting breaks, or why images don’t display properly. In a nutshell, it can make or break an entire ecommerce business.

I’ve had clients at large, global brands asking why their ecommerce sales are not what they expect.  And then after doing a thorough assessment of their site, you find that they have a mismatched taxonomy, no defining attributes, inaccurate descriptive attributes, and generally terrible product data.  This is not an issue limited to small companies.  In fact, in my experience, it is actually more prevalent at the larger, more established organizations where the existing data long precedes ecommerce, or even the internet itself.  More and more businesses are implementing a D2C model (Direct to Consumer), and these organizations are typically manufacturers by nature.  Manufacturing data is typically created by engineers and not merchandisers, and therefore the data is not thought about in the same manner that a merchandiser or marketing professional would consider it.  A common naming convention in the industry may not make any sense whatsoever to an ordinary customer.

So with that being said, here are a few things to consider when organizing and planning your data architecture.

  1. Group your products in a way that makes sense.  If you have 10 variations of essentially the same product, don’t make them all individual products.  They should be a single product, with 10 varying child SKUs under the product.  The only data that should differ among those products would be whatever makes that SKU unique from the rest of the products.  For example, if you have a bunch of shirts that are identical other than the fact that they are available in red, black, and blue, and each comes in size S, M, L, and XL, then you should set this up as a single product.  It would be 12 unique SKUs under this product (4 for each size in each of the 3 different colors), but the only thing that would differ between those SKUs would be Size and Color.  The name, description, and all descriptive attributes would all be identical, and the defining attributes would be the method in which you select the unique SKU to Add to Cart, by selecting Size and Color.  Too many sites have these broken up into too many product pages, which is a waste of time and energy that goes against ecommerce best practices.
  2. Make sure the data that you do have is accurate and valid.  There is nothing that will kill a conversion more than giving a customer a reason to doubt that the product they are looking at is the product that they need to buy.  It is imperative to make sure that every piece of information that you have is available on the product detail page.  Nobody likes surprises later on in the checkout flow, so if a product it out of stock, has an additional shipping charge, or requires special tools for installation, then you should let the customer know BEFORE they add the item to the cart.  Otherwise, you risk alienating them as a potential customer and cause problems for yourself in having to make the situation right in the future.

Read More...