This post is written by Guy Yalif, CEO & Co-Founder of Intellimize.
The number of companies offering products that use Artificial Intelligence and Machine Learning (AI/ML) to help marketers be more productive and effective is growing. Marketers are trying to evaluate different technologies to find the right tool to address their specific needs; however they face challenges evaluating the claims and capabilities of each tool to separate hype from real value. I’ll be speaking at CXL Live in a couple of weeks on this topic and will share additional thoughts in a future post. Here are five questions you should ask when choosing an AI/ML-based testing and personalization solution to help grow your business:
It is important to understand how a particular ML/AI technology improves your key business metrics such as revenue, user signups, purchases, and conversions. Some services optimize directly for the key metrics, while others provide data or automation that fill a specific task in your overall process.
Lead scoring systems tell you which prospects are similar to your existing customers or which prospects are likely to convert. This can be valuable information to prioritize your sales message and time and should be evaluated based on accuracy. Content recommendation services present individual pieces of content that are most likely to engage the user – often optimizing for CTR. These services should increase time spent.
Other services are focused on directly optimizing for conversions. They use ML technology to take action to modify the user experience on your website. These services track your conversion rate with and without the service, using a control group to ensure the service is adding real value to your business.
In a previous blog post, we covered the potential lost conversions that come from choosing a “single best page” for site visitors. Make sure that the solution you’re evaluating has the ability to automatically personalize, which will always outperform systems that attempt to find the “single best page” for your audience.
Personalization unlocks more conversion potential in every site visitor rather than looking for the single best experience most likely to appeal to the average visitor. If your goal is more conversions, look for systems that can help you scale one-to-one personalization across your entire audience.
Some solutions require a lot of data to be collected up-front before the system can leverage that data to make predictions and take action. For conversion rate optimization, most marketers do not have a pile of historical data available to train an algorithm. As a result, you should look for systems that can start learning in real time from each visitor to your website.
Other solutions will learn by analyzing a specific data set and then require periodic manual tuning and updates. If the manual tuning and updating is not done regularly, your optimization can get stale as visitor behavior changes over time or as you change your marketing efforts. You can drive more conversions by using a system that continuously learns from new data and uses that learning to automatically adapt to any changes in visitor behavior.
Pro tip: Some solutions can only create a single model that works across all visitors – or one that works only for a small subset of visitors. Look for a ML system that can make the right decision for each individual user.
Prospects are dynamic and non-homogenous. Creating, managing, and tuning personalization rules for each segment of your prospects can take a lot of time and effort. Audiences and user behaviors change on their own and as a result of you evolving marketing efforts. Additionally, it is very difficult to come up with the right personalization rules that maximize conversion rates right off the bat. For example, can you think up the optimal experience to show each combination of user attributes like age, location, and industry?
Look for machine learning systems that don’t require manually created rules and can adapt and respond in real-time to changes in user behavior. You should also ask if the system can accept rules so that you can impose hard lines when you want them, for example when you want to offer a discount only to high value prospects.
If you can’t see what decisions your AI/ML tool is making for you, can you be sure it’s making the right decisions? Can you access the data—your data—reflecting the choices the system has made? Can you integrate the AI/ML system with your analytics provider of choice so that you can do longitudinal analysis and understand the ways you’re increasing conversions?
Look for systems that transparently reports every decision they made. Ideally, you will see impression-by-impression details on which variations were shown to each individual visitor. This allows you to ensure that the system is working as you expect and can help you identify opportunities across your customer base.
Adding tools to your marketing stack should improve your key metrics and make sure they accurately quantify the improvement in their reporting. Asking the right questions is essential to finding the right tool to help grow your business, and the questions above shouldn’t leave your prospective technology provider stumped. We hope these questions help you find the right AI/ML system to give you leverage and improve your conversion rates!