About a year ago on this site, I penned a post titled "Analytics lessons from Amazon." In it, I described the analytics lessons that employees and even entire companies can learn from the retail giant.
But there's so much more that Jeff Bezos et. al can teach us. Today, I'll focus on the data quality lessons we can glean from the largest Internet retailer.
Recognize that all data is important
By way of background, Amazon understands that all data is important whether it is:
- Big or small.
- Structured or unstructured.
- Internally or externally generated.
As someone who has followed the company since its inception, trust me: this core belief cascades throughout the organization.
Refuse to become complacent
Amazon stocks a ridiculous number of items. One of my favorite stats: its clothing selection is now bigger than 250 Walmart supercenters combined. With that much inventory, from time to time there's going to be a data issue or two. (I wrote a post about one such issue back in July of 2014.)
Still, Amazon is constantly its improving the data around offerings, even if that means acquiring companies such as IMDb in 1998.
With so many products and different lines of business, a top-down approach to data quality would never fly at Amazon. To this end, the company provides its customers, partners and vendors with tools to manage a great deal of their own data. Examples include:
- Customer self-service.
- Author self-service.
- Vendor self-service.
- Amazon Advantage, a self-service consignment program.
I'd wager my house that the company also allows employees to change address and tax information via employee self-service, but I can't be entirely sure.
Note here that there are limits. For instance, vendors can't cut themselves checks. Still, why make a central Amazon department responsible for managing each of these myriad relationships?
Force employees to routinely make decisions based on data, not simply gut feel
The New York Times' 2015 Amazon article sparked a national debate about the company and its cutthroat culture. I won't open that debate here, but the piece certainly stressed the degree to which employees at the company regularly use data to make decisions. No, not every blue-collar employee at its fulfillment centers consults dashboards and KPIs before making decisions. Make no mistake, though: At Amazon, employees either bring data or go home.
Be vigilant against data pirates, even if it means going to court
Nearly 70 percent of consumers these days consult product reviews before making purchases. It's a big deal, yet Amazon has struggled to prevent nefarious types from gaming its reviews. Forget authors who reach out to friends for book reviews. I'm talking about companies that profit from writing bogus write-ups of different widgets.
Unable to stop this via data and technology, the company was left with no other choice: Amazon is suing more than one thousand sellers of fake product reviews.
Simon Says: Data is sacred at Amazon
To be sure, data quality at Amazon is sacrosanct. If you doubt this after reading this post, give Brad Stone's excellent book The Everything Store a read some time.
What say you?Download a paper: Understanding Big Data Quality for Maximum Information Usability