Skip to main content
All CollectionsABOUT AI TOOLS AND ADEXT DASHBOARD
Everything you need to know about Amplify
Everything you need to know about Amplify
Adext AI Support avatar
Written by Adext AI Support
Updated over 3 years ago

Amplify is Adext newest AI Tool. With this tool, you’ll have the option of boosting your winning audiences, widening the possibilities. Once these audiences have stood out by their performance, you’ll be able to choose among 3 different AI Machine Learning models that offer more options for your campaigns, and the KPI you want the AI to focus on, this AI tool will help you to improve the results and take your ads further.

Amplify Machine Learning models are:

1. DISCOVERY

This model needs a lot of data to work and in order to boost the campaign or ad with the best results, it does simulations to gather more data. It needs to see the distribution of probability that each ad has (exploration) and select the one with the highest probability of being the best one, based on the data obtained. It needs more time than other Adext’s models to find which ads are winning and which ones are losing.

It's best to use the Discovery model when you: Want Adext to run your campaigns in an exploration mode, are willing to give Adext AI more time to gather enough data, or, want to run a longer campaign.

2. JACKPOT MODEL

This model will focus on which ad has the best performance and focus on that one, exploiting it as much as possible. This model is recommended for shorter campaigns when you have little time to run them since you can iterate quicker than with other Adext’s models.

It's best to use the Jackpot model when you: Want Adext to find and boost the top-performing audiences quickly, want to run a shorter campaign (e.g. Black Friday), or, need to get conversions faster and not focus only on CPA.

3. CHALLENGE

This model needs more time to learn since it’s based on learning by reinforcement. It gathers its learning from “experience”, focusing on losing and winning rewards or punishments. For example, it has 10 possibilities to select the audiences, but if it selects one that didn’t have a positive result, its reward is lower than if it selects one that had a good performance, this way its start to choose the best-performing ones only.

It's best to use the Challenge model when you: Have mid to longer duration campaigns, want to boost the best-performing audiences and cut the losses on other audiences, limiting the budget, or, want to focus on what is working, hence, delivering your ads only to certain audiences.

Did this answer your question?