Artefact Value By Data

Amazon Ads – Golden rules to prepare Q4

The end-of-year period is ever more strategic and competitive for brands selling their products on Amazon. Both Black Friday and the holiday season are catalysts for sales, increasing buying opportunities for end consumers, but also galvanising competition between brands: on price, product quality and visibility through advertising both on and off Amazon. It’s essential today to take advantage of all the tools and solutions offered by Amazon. This article highlights our 10 golden rules for harnessing the full power of the American company's advertising ecosystem.

The road ahead: data-driven marketing is critical for the evolving car industry

To the casual observer, the motor industry might seem in good health, with electric and hybrid vehicles increasingly appearing on our streets and driverless cars becoming a reality, not to mention people preferring the safety of their car over public transport following the coronavirus pandemic. But this surface view belies an industry in flux, facing change from several directions.

Ad sequencing: how to generate storytelling on YouTube

Nowadays, advertising saturation on all channels causes the message we want to transmit to be too repetitive and does not inspire interest for the user. In this sense, the option of generating coherent Storytelling for the user can be a complicated task.If we have a brief where video actions and storytelling are proposed, I think that the YouTube Ad sequence could be an option to value.Using this type of strategy, we have seen increases in frequency up to 2.4x and Brand Awareness up to 4x through YouTube Brand lift measurements.

7 pandemic-driven business lessons learned for the long-term

For over a year and a half, organizations have undertaken a continuous cycle of innovation and adaptation to meet the changing environment as the pandemic continues to play out. Ghadi Hobeika, CEO of Artefact in the US, provides an overview of seven learnings from the pandemic for the long-term.

ML-Modelle in großem Maßstab mit Mlflow auf Kubernetes bereitstellen - Teil 3

Dieser Artikel ist der dritte Teil einer Serie, in der wir den Prozess der Protokollierung von Modellen mit Mlflow, deren Bereitstellung auf der Kubernetes-Engine und schließlich deren Skalierung entsprechend den Anforderungen unserer Anwendung durchgehen. Obwohl dieser Artikel unabhängig davon verwendet werden kann, um eine beliebige API-Antwort zu testen, empfehlen wir die Lektüre unserer beiden vorherigen Artikel (Teil1 und Teil2) über die Bereitstellung einer Tracking-Instanz und die Bereitstellung eines Modells als API mit Mlflow. Im Folgenden werden wir uns mit dem Problem der Skalierbarkeit befassen und einige Experimente durchführen, um das Verhalten des k8s-Clusters zu verstehen und Empfehlungen zu geben, wie Sie mit hohen Lasten umgehen können.

Nach oben