Congratulations on taking the first step to faster conversion ROI!

Dynamic Testing combines the power of automation and machine learning with your own marketing instincts. It monitors website traffic in real-time, then uses machine-learning algorithms to decide which split is the top performer. Best of all, it optimizes based on any goal: conversion, form submission, and more.

In this session we will:

  • Explore the science behind Dynamic Testing
  • Review use cases that capitalize on the new technology
  • Show you how to setup your first Dynamic Test

Web Seminar Q&A: Mastering AI Driven Personalization

Special thank you to our engaged audience for asking excellent questions. We’ve captured all the questions we ran out of time to address below.

You can watch the full seminar here.

Questions for Zachary Martz, Club Monaco

Q: Based on what you learned in your experiments, where else will you use The Engine?

We have further plans to place the correct offers and promotions in front of customers when and/or if needed. Additionally, we are hoping to recommend more of our editorial content to customers organically on-site that is relevant.

Q: In what ways does using The Engine change how you do your job?

The Engine has allowed us to make more strategic decisions directed towards KPIs rather than decisions based on setup bandwidth and the potential need for constant monitoring.

Q: Can you talk more about how you were able to reduce dependance on price promotions in your email sign up experience? Where else do you anticipate being able to reduce dependance on discounts?

We are finding that providing the correct content at the correct time is far more valuable than showing arbitrary discount incentives. Customers loyal to the brand are looking for our guidance on products and style, rather than incentivizing them alone on the site to hunt for discounts.

Q: In addition to improving conversion metrics, what are some other benefits of this approach for your team? Does it create less work for you or more?

Before we would closely monitor an experiment for anything immediately detrimental to the business and take action accordingly. With the AI technology in place, the system automatically takes an experience out of favor when it is deemed a poor performer. Additionally, if the habits of customers changed and the previously poor performer is now excelling, the AI would bring it back into the mix. The time consuming analysis and constant retest factor is no longer a part of the process. AI decisions have also provided interesting insights into the customer groups that are good performers in an experience.

Q: In addition to the data you are currently using, what else might you want The Engine to access?

Currently we are only using online data to optimize the AI but we are hoping to have brick-and-mortar or marketing channel specific data ingested to make better decisions.

Q: How much of effort/time goes into getting the data in place before you starting using AI?

The short answer, little time/effort to get the data. We have spent a decent amount of time making sure the correct KPIs and base data is being accounted for consistently/daily in Monetate. When setting up an experience we make sure a goal is identified and then the appropriate data is being collected to show performance. For the most part, standard metrics are used but, through internal review or best practices from Monetate, we make quick work of adding case-by-case metrics/events.

Questions for Adam Litle, Monetate

Q: Do you include transactional history in your models?

Yes, the system can support virtually any data element including transaction history.  One of the exciting things about using artificial intelligence is the ability to find connections we might not otherwise have been able to understand.

Q: Can you talk about the difference between optimization, segmentation and personalization?

The path to personalization is a journey that begins by optimizing our digital experiences. Traditional testing methods provide the foundation for our early learning. These methods (including A/B and Multi-variant testing) are very useful for lifting the overall experience for a homogeneous environment. However, we quickly learn that while we improve the experience for some visitors, we make it less attractive for others.

As a result, we begin to segment our audience into smaller groups. Applying rules which augment the experience for a group of visitors that fall into pre-defined segments.

True 1:1 personalization removes the constraints of segmentation by allowing us to serve the most effective creative for any given individual.

There are still applications for testing and segmentation, which is why we also include these capabilities in the platform. We don’t advocate for eliminating these approaches altogether, but rather using them for specific purposes. For example:

  • A/B Testing
    • Where learning is more important than driving ROI
    • When you are looking to test in one place to inform another
    • When you have a long cycle, like send one catalog a year and want to A/B test something associated with the catalog
  • Dynamic testing
    • Where you need to take action fast to drive results and can’t take the time to explore/learn
    • When there is a very low session volume
  • Segmentation
    • When you have very clean segments, like a loyalty club audience, that you wish to target with specific content.

1:1 experiences should be used for the rest. We estimate the usage of these different approaches to be weighted heavily to 1:1 with testing and segmentation representing a minority of the experiences being served to customers.

Q: How much of effort/time goes into getting the data in place before one starts using AI?

Our customers are surprised how quickly The Engine can add value, using data that is easily available. We generally recommend starting with a small set of data inputs to build a foundation, then grow complexity over time as resources allow and insights signal new areas of opportunity.

Q: How is AI different from Big Data Analytics?

Big data analytics is typically deployed to find patterns among individuals in order to formulate an understanding of visitor behavior after the behavior has occurred. In the best case scenario, that understanding is used to create visitor segments and those segments are targeted in a way that drives a particular outcome (e.g. purchase conversion, increase AOV, etc.). Whether that outcome is achieved or not is a matter of re-aggregating data, analyzing it, and the cycle continues. It is also important to note that there is latency between each of these steps.

The type of AI that was illustrated in the webinar and that Club Monaco uses so successfully executes a different cycle that eliminates the need for segmentation and the latency between steps. Visitors are analyzed, the best action is determined and delivered, and the outcome is measured in real-time. The models update in the matter of seconds so that any changes in visitor behavior can be detected and accounted for.

Q: Where can I get a look at Monetate Intelligent Personalization Engine?

We’re very glad you asked. You can schedule a demo here.

Monetate Logo No Tagline Blue.pngCopyright ©2017 Monetate Inc. All rights reserved.