Loft Troll: How we built a Kindle Direct Publishing (KDP) tech stack

Amerge began working with the US publisher back in November 2020, managing their full Amazon Ads business. In simple terms we have built a full KDP tech stack in partnership with Loft Troll.

Working with Loft Troll has been a true partnership – the client has deep dive knowledge of both the KDP platform and Amazon’s Advertising formats, and we partnered to turn their concepts into fully automated solutions. The first challenge was to allow us to understand our impact of advertising by forecasting the % of customers who were likely to purchase all (or read in the case of page views royalties) or some of the books in a series after the first purchase. This gave us ‘Real ACOS’ which we then integrated into Amerge View to give a dynamic view of performance.

In addition, a large majority of KDP sales come from KDP page views which is not directly reflected from conventional advertising metrics i.e. ACOS. As the KDP platform does not have a API integration we built a custom cloud integration to allow this to be represented in Amerge View allowing us to seamlessly report on the totality of performance for their campaigns. Campaign Performance: The client traditionally reviewed best seller lists on Amazon and integrated new and trending titles into campaigns manually on a weekly basis.

Our challenge was to fully automate this process. Best seller lists are updated on an hourly basis on Amazon – we built a fully custom solution to programmatically review these best seller lists on an hourly basis and integrate new titles into ‘shell campaigns’ through the Amazon Advertising API. This gave the client a fully dynamic always on competitor strategy. At the time of writing, we implement 75 best seller lists with over 7,500 titles. Finally for DSP the client was focused on only targeting new readers. Due to the set-up of KDP Accounts new to brand does not report correctly (over reports sales). By building a multi-layered negative targeting strategy excluding customers who purchased in the previous 356 days simulating NTB.