Making of Sellercompass (postmortem)

I started developing Sellercompass in 2012. It was an app for the Shopify platform, targeting Google shopping (which is becoming the de-facto standard for product feeds). My plans were to extend the app to many more e-commerce sites (Amazon and the like) and to create a stand-alone service.

With Sellercompass, merchants were able to import their products and create a product feed that they can use to sell on other platforms (i.e Google Shopping).

The biggest challenge when creating product feeds is to keep up with the correct format of the target platform and handling a possibly large number of products. I aimed to make Sellercompass fast and as user-friendly as possible. One difficulty lies in assigning the right product category, especially for shops that have different product types. To ease the tiresome workload, Sellercompass has some sort of “Category Wizard” that tries to automatically assign a category. You can read about this here: Automatically matching a category in Google’s taxonomy.


Sellercompass was developed in Python using Flask. The frontend was built with Bootstrap and Jinja templates. I developed this using a plugin structure in order to make adding new sites and formats easier. On the persistence layer, I used Mongo and Redis.


Developing for an existing platform has many advantages. Marketing becomes much easier, you don’t have to deal with payments and so on. Due to this, Sellercompass made money from the beginning. But of course, there are also disadvantages, most of all being dependent on a third party. In the worst case scenario, the functionality gets offered by the platform itself and thus rendering the app obsolete. This happened on Shopify.

I sunset the app end of 2016 as it did not generate enough revenue to justify the time commitment.

Subscribe to Human Intelligence Engineering
Explorations of human ingenuity in a world of technology.