The success stories of the new digital economy are businesses who understand that bad data leads to poor guest experiences, both in B2B and B2C. The bastion of search, Google, which was assumed to be untouchable, is now threatened by Amazon which is using its consumer purchase data to power its ad platform.
Good Product Data = Good Customer Experience
Product data is no different, even though most businesses treat it as if it is. The ability to tie product data to customer insights, behaviors, and purchasing rhythms is a developing realm that most businesses have yet to explore. Understanding that a customer bought a particular item is important, but understanding why they bought it is Business Intelligence for the next digital revolution.
You cannot be part of that next digital revolution if you do not have high quality product data.
The rise of digital commerce has created a data landscape that self-enforces good data practices.
What is Product Data Quality?
There are many tenants to data quality — accuracy, completeness, timeliness, and consistency are just a few of the primary dimensions.
Standardisation of data
Product data requires these dimensions, but its collection is more difficult because it comes from unverified sources outside normal business channels. Whether you are a retailer attempting to acquire data for products from suppliers and distributors, or a supplier trying to corral that data for competing retailers with different data expectations, finding consistency and accuracy in your data is one of the biggest challenges in product data today.
Economics of Collections
Another important aspect with product data quality is the economics of collection. There are two elements to the economic equation: system requirements and feasibility. Without an understanding of these two factors your product data collection expenses could outweigh the benefits of that product data.
First off, systems play a huge role in product data collection. Simply put, the perfect product taxonomy is useless if the technology to collect the data does not support it, and the perfect technology for data collection is worthless without a good product taxonomy. If your investment in these areas is not balanced your data acquisition costs will include more remediation and lower data quality.
Secondly, there is always a decision to be made on how much data to collect. As product taxonomists we tend to want to capture every facet of a product, assuming that every attribute is valuable. This is not the case. The color of the plastic on the inside of an iPhone is not valuable product data even if it’s technically possible to collect. However, the cost to the supplier to source this data and the loss in speed-to-market for the retailer in asking this question make it unfeasible to ask. It may be accurate, consistent and complete, but it is not economically viable.
Why does Product Data Matter?
On it’s own a single piece of data might not have a huge impact. On a larger scale though, product data can affect many important aspects of your business. It can influence:
- Faceting — error or inconsistency can make a product disappear from a faceting or search experience.
- Findability — the product may end up classified incorrectly and wind up in a product listing page that makes no sense.
- Presentation — a red lamp shade will be difficult to sell if the title says “Blue Lamp Shade”.
However, there is a bigger issue in product data today. Most suppliers and retailers have worked out at their own level how to make product data mostly accurate. What they have not done is ensured that the data is presented consistently across channels. It is great to have accurate product data on your own ecommerce platform, but what if the data on a marketplace site contradicts your own website?
Controlling your product message is important because it is your product and brand reputation you are messaging. At the heart of this message is product data, so the two are intertwined. If you want to have an engaging product message you have to have consistent product data across all channels that product may appear.
In another blog I recently demonstrated how a single lamp on 4 websites had 4 different experiences. This lamp exhibited simple mistakes at the product data level that made the message on any single channel appear odd, and together those oddities added up to an inaccurate, inconsistent product message.
For example, the dimensions of this lamp were different on the 4 different web sites. Determining which web site was correct would be impossible for a customer, as all 4 had completely different measurements.
At the same time, one website had an attribute of the lamp labelled “Gender”. The value for this particular lamp was “Unisex”. Obviously that site had forced the product data setup resource to fill in a value, but the gender of a lamp when it is unisex is irrelevant. The economics of capturing that gender attribute didn’t make sense, as only 2 lamps in their assortment had values other than unisex.
There were structural issues with how the data was provided to each company as well. The titles were never the same, with one website calling it a “goosneck” lamp instead of “Gooseneck”. A simple data mistake regarding an included light bulb made the lamp unsellable in California due to restrictions on incandescent light bulbs even though the lamp didn’t ship with a bulb.
Because the data is inconsistent across these 4 channels this single lamp suffers from a huge issue. If anyone cross-shops this lamp they will find wildly contradicting messaging regarding dimensions, and potentially cannot even buy the lamp in California if they attempt to purchase it from one website.
A lack of control in product messaging might mean that a single data point could seem reasonable on its own, but in context become obviously flawed.
These flaws contribute to lower conversion rates, lower findability from both internal search and SEO, and higher liquidation costs to remove unsold inventory when the product does not sell to expectations.
Fixing Product Data Quality Issues with PIM tools
The way to fix these product data and product messaging issues is to use a commercial Product Information Management (PIM) tool. A PIM tool allows you to control your product message by controlling the acquisition, normalization, and dissemination of your product data. As the central hub in your product data landscape, a PIM tool is the control mechanism for your product marketing experience on your business web site, as well as every other channel you sell that product in.
There are 3 keys to understanding how to use your PIM. They all start before you collect or remediate a single piece of product data. If you are looking to install a PIM, about to go through an implementation, or scoping a data remediation project here are the 3 things you need to do before you start that project.
#1 - Know what quality product data looks like.
Without an understanding of what you end product message is you can never achieve the goal of fixing your product data quality issues. Quality product data is consistent, accurate, timely, and complete… but it is also centralized and distributable in a consistent, accurate, complete, and timely fashion. If your plan is to solve your data quality issue without understanding these factors it is not a plan that will succeed.
#2 - Set up product data governance before you start your project.
If your plan is to fix your data and then set up governance after the project your data will not retain its quality very long. You will waste your investment remediating your data because your users will quickly deviate from your intended data standards as soon as they set up the next new product. If your governance is not available to set standards for case, spelling, grammar, and mutual exclusivity those tenants of product data will quickly be broken, causing more remediation and higher data acquisition costs.
#3 - Understand where your data flows.
In having performed many taxonomy rebuilds and PIM installs over my career I can tell you that few people in a company fully understand what systems consume their product data. Every company understands their data flows to their ecommerce platform, but what about your micro-sites? There are also your channel partners that you syndicate your product messaging through, and your printed spec sheets and catalogs. Does your Customer Service team use product data to answer support questions to customers? Where does the data for your shelf labels and point-of-sale register tapes come from? Understanding ALL your data output channels more important than understanding your ecommerce needs, because there is a high potential that your product data is seem more in places that are not your ecommerce site.
Controlling your product messaging is not just a marketing slogan I use with clients. There are companies speaking about Product Experience Management or Product Presentation Management that are saying the exact same thing, except they use fancy acronyms like PEC and PXM.
Most retailers and many suppliers see product data as an acquisition cost. It is to be managed as you would any other cost; minimize the expense to provide the highest margin for that product. However, product data is an asset, just as any other data used for business insights is an asset. If you continue to treat product data as a cost rather than an asset your ability to monetize that asset will be limited by your willingness to incur the expense.
Daniel O’Connor is product data quality & taxonomy expert and runs the website ControlYourProductMessage.com.
Learn more about how to get standardised product data for your catalog on semantics3.com.