The Signals Framework

Signals are the foundation of Raleon's approach to customer segmentation. Unlike traditional rules-based systems, Signals are sophisticated behavioral models that automatically analyze and adapt to your customers' actions. This guide explains how Signals work and how to use them effectively in your marketing.

To create a segment in Raleon, you simply select one or more signals. The signals automatically filter your customer list to only the customers that fit the objective of the signal. That's all you have to do! Choose signals, click Go. You'll never have to waste time on complex and impossible to maintain rules again. With Raleon, it's drag and drop.

Signals vs. Rules

Traditional rules are like trying to identify basketball fans by checking who's at the stadium today, while Signals understand who loves basketball by analyzing their full history of game attendance, merchandise purchases, and how they engage with your basketball content year-round.

What Makes Signals Different

The power of Signals lies in their ability to automatically adapt to your business. As your customers interact with your brand - whether through email opens, purchases, or website visits - Signals continuously update their understanding. In other words, the more interactions the signals see, the smarter they get.

The additional benefit of this is you don't need to manually adjust rules when your business changes. If your average customer starts buying more frequently, or your email engagement patterns shift, Signals automatically adapt to this new normal.

At their core, Signals come in two varieties:

Model-based Signals

These use proven statistical approaches to track straightforward patterns like purchase frequency or email engagement. These are incredibly reliable and work well even with limited data.

AI-based Signals

These signals use machine learning to uncover complex patterns in customer behavior that might not be obvious at first glance.

The Signal Learning Process

You're probably thinking - that's all great, but how do these signals actually learn? What goes into it? Well, it first all comes down to data.

Signals get smarter with every interaction. Each email campaign you send, every purchase made, and all customer interactions contribute to Signal accuracy. This is a key advantage over traditional segmentation: instead of using fixed rules that become outdated, Signals evolve with your business.

For example, if you start sending more email campaigns, Signals gather more data about how different customer groups engage with your content. This helps them better predict which customers are likely to engage with future campaigns. The same applies to purchase patterns, browsing behavior, and other customer actions.

Signals as a result are updating their brand-tailored score on a daily basis. These scores are what then go into segments, and allow you to do even more fine-tuned tailoring if desired.

Understanding Signal Scores

Every Signal generates a score from 0-1,000 for each customer, representing how strongly they exhibit a particular behavior. While it varies from brand to brand, generally the following holds true:

• A high score (>800) indicates a strong match with the behavior the Signal tracks
• Medium scores (500 - 800) suggest moderate alignment
• Low scores (<500) show minimal alignment with the behavior

For instance, a "Likely to Purchase" Signal score of 900 indicates a customer strongly matches the patterns of someone about to make a purchase. You might prioritize these customers for promotional campaigns or special offers.

Using Signal Scores

What's important to remember is that these scores are predictive in nature -- they are probabilistic.

This is important is because it's easy to fall into the trap of thinking you just always want to target the highest scorers. Think of Signal scores like choosing where to cast your net in a pond. The score tells you where the fish (your target customers of a given type) are concentrated.

A score of 900+ is like casting in the deepest part of the pond where the biggest fish consistently gather. You'll catch fewer fish overall, but almost every catch will be exactly what you're looking for.

A score of 700-900 is like casting in a wider area that includes both the deep spots and the nearby waters. You'll catch more fish overall, including those prime catches plus others that are still worth pursuing. Your net brings in more variety, but still maintains good quality.

Below 600 is like casting in the shallow edges. You might catch some fish, but you're mostly pulling in weeds and stirring up mud. The effort isn't worth the return and is likely hurting your deliverability and future conversion chances. This is not a very fun way to fish.

What Signal Score Should You Use?

The right Signal Score threshold will vary from signal to signal, and from brand to brand.

Signals can be, but are not always, normally distributed. In other words, some signal scores will have many customers above 800 and few below. In other cases the median may fall right at 500. And in others still the median score may be well below 500. Each customer is scored based on their own probabilities; we do not force a regular distribution.

This can feel overwhelming at first, so by default our AI automatically picks the optimal score for you.

Most brands and agencies let our AI do the heavy lifting with small interventions over time as required. It's only if you want to really start tuning the Signals yourself that these changes really start to matter. For most brands, Raleon's default signal threshold will be optimal.

Of course, you can always reach out to us as well! We'd love to help!

Signal Library

For a list of all the available Signals in Raleon, you can view the Signal Library in the Raleon platform itself. As we make frequent updates to the available Signals, the library will have the most up-to-date list, as well as what the signal is best used for.