5 Decisive Data Feed Management Practices

2022.04.29

Home > Articles > 5 Decisive Data Feed Management Practices

Your Data Feed or Product Feeds may have issues, but there are some best practices for what you should DO and how to avoid pitfalls you DON’T want to make. We understand that the information your feeds provide is prone to human error, but there is also an expectation of the receivers for it to be dynamic and expedient. With new tools like ProductsUp, ZAMARTZ can help manage this anarchy and improve your quality and speed to market.

Q: Simply, what is a feed? 

A: It’s easiest to describe a feed as a list of records, where each record has a unique identification and values that are formatted against their friends in a consistent manner. You can think of this like ROWs and COLUMNs in a spreadsheet.

Even as late as the early 2010s, data feeds / product feeds were still mostly used for paid search marketing and product display ads. These areas are still major applications, but there has been an explosion of additional syndication channels that are utilizing this data. These additional channels range from affiliate links sharking, unified shopping marketplaces, social media shops, and even customized triggered email campaigns. Each of these locations requires a myriad of different required fields and methods of delivery.

Product Feed Processing Example

Here is an example of an apparel product and what this process looks like.

1. We will start with a sample data file that is imported as a .XML and is set to a column structure.

Product Group IDProduct Variation IDNameVariation Detail SizeVariation Detail ColorAdditional Detail Material
Logo-Tee-ABCLogo-Tee-ABCOrganic Logo TeaOrganic Cotton
Logo-Tee-ABC1001234567891SmallAzure
Logo-Tee-ABC1001234567892MEDAZURE
Logo-Tee-ABC1001234567893SMLRose
Logo-Tee-ABC1001234567894MediumROSE

2. As we can see, there are inconsistencies and missing data. We will try and fill-in the blanks or extend the content so we have as much data as we can for potential exports. This may mean formatting cases, bringing in missing data to a child from a parent, or even creating mapping lists to use existing data to create new data.

Note = Removes first comma /^\,/

Note = Removes last comma /,$/

Product Group IDProduct Variation IDOptimized Product NameVariation detail SizeVariation detail ColorAdditional Detail MaterialParent or Child
Logo-Tee-ABCLogo-Tee-ABCOrganic Logo T-shirt,Small,Medium,Azure,RoseOrganic CottonParent
Logo-Tee-ABC1001234567891Organic Logo T-shirt size Small, in Color Azure SmallAzureOrganic CottonChild
Logo-Tee-ABC1001234567892Organic Logo T-shirt size Medium, in Color Azure MediumAzureOrganic CottonChild
Logo-Tee-ABC1001234567893Organic Logo T-shirt size Small, in Color RoseSmallRoseOrganic CottonChild
Logo-Tee-ABC1001234567894Organic Logo Tea size Medium, in Color RoseMediumRoseOrganic CottonChild

3. Finally, we will take this sanitized and expanded data and optimize it to be sent to Google in a .csv to their FTP location. We will remove irrelevant data and map in some generic terms that are better received into Google’s platform.

item_group_ididtitlesizecolormaterialgtin
Logo-Tee-ABC1001234567891Organic Logo T-shirt size Small, in Color Azure SmallBlueCotton1001234567891
Logo-Tee-ABC1001234567892Organic Logo T-shirt size Medium, in Color Azure MediumBlueCotton1001234567892
Logo-Tee-ABC1001234567893Organic Logo T-shirt size Small, in Color RoseSmallRedCotton1001234567893
Logo-Tee-ABC1001234567894Organic Logo Tea size Medium, in Color RoseMediumRedCotton1001234567894

This may sound daunting, but there are tools and experts like ZAMARTZ here to support brands and businesses with each and every step While there is much to discuss, and we could write a book on this subject, we will go through some of the basic fundamentals  below.

DO’S AND DON’TS OF FEED MANAGEMENT

DoDon’t
IMPORTS
Import all the base data you can, within reason. It may not appear relevant at the start, but having something later in the process you weren’t anticipating will come in handy.

Over-engineer your import and spend hours developing the correct content or coding for errors.
ADDITIONAL DATA
Add additional data sources to your feed that connect, extend, or generally improve the content of a record.

Think that what you see in the initial import is what you will end up with. Expanding your data is part of the process.
MONITOR
Add notifications about the main factors that will prevent your feeds from being successful.Know when your most important imports and exports fail or have not updated from day to day. 

Rely on the feed provider or receiver to be flawless.  They will have errors and it is better for you to catch them before they do. 
IMAGES
Know your image processing services. They can help you provide the correct images to external platforms without needing to do development in the original feed.

Crop your images or request a low resolution in your base feed import. Having the highest quality image will allow you more flexibility. Note: Some images are not served through a service.
ANALYZE
Review each of your data columns so that they meet the export requirements, have consistent formatting, and required fields are populated.

Use consecutive white spaces, html, or ALL CAPS. Make sure all product links and product images work.

What’s in it for you?

From our experience and from feed management providers, we have found that having your data feed corrected, extended, and optimized has many benefits. In normal situations where you have not participated in programs or never optimized, we often see these key improvements:

  • Disapproved Products down to 0%
  • Increased Traffic Reach up to 30%
  • Increased Impressions up to 40%
  • Increase in tROAS up to 50%
  • Reduced CPC up to 70%

Appendix

Common Data Specification Formats

While many syndication partners have their own unique uses for your data feeds, there are some common formats that most will accept as a standard. 

Common Image Processing

Most data feeds use images, but your import may have just the basics. Don’t get discouraged or feel limited, there are some providers where you can use the basic platform and get alternate views or crops.

ServiceDocumentationCrop FormulaTest URL
Scene7 / Adobe Dynamic Imagehttps://experienceleague.adobe.com/docs/dynamic-media-developer-resources/image-serving-api/image-serving-api/http-protocol-reference/command-reference/c-command-reference.html?lang=enhttps://s7ondemand5.scene7.com/is/image/{{accountName}}/{{imageName}}?{{rules}}
Preset = ?$RETINA_X_174$Manual = ?wid=1000&hei=1500&align=0,1&fit=crop
https://s7d1.scene7.com/is/image/S7learn/backpack?wid=1000&hei=1500&align=0,1&fit=crop 

Salesforce Commerce Cloud (SFCC)https://documentation.b2c.commercecloud.salesforce.com/DOC3/index.jsp?topic=%2Fcom.demandware.dochelp%2Fcontent%2Fb2c_commerce%2Ftopics%2Fimage_management%2Fb2c_image_transformation_service.htmlhttp://<image server host name>/dw/image/v2/…/on/demandware.static/…/<static image path>/{{imageName}}?{{rules}}
Manual = ?sw=1000&sh=1500&sm=cut
(this is a bit of a customization to do this in SFCC)
Widen DAMhttps://community.widen.com/collective/s/article/How-do-I-customize-share-links-and-embed-codes-for-imageshttps://eudaimonia.widen.net/content/ig5uijtr0y/{{format}}/{{filename}}?{{rules}}https://eudaimonia.widen.net/content/ig5uijtr0y/jpeg/Bliss%20environment_1.jpg?w=1000&h=1500&keep=s&crop=yes