What's The Difference Between a Rules-Based System and a Learning System?

One of the major differences between Click360 and other sales and marketing intelligence platforms is that Click360 is a Learning System while other platforms are Rules-Based Systems.This blog post examines what makes them different and why Learning Systems are superior for sales and marketing analytics.

TL;DR: Rules-Based Systems are expensive to build and maintain, typically have poor results, and quickly become out of date. Learning Systems, in contrast, are cheaper, have better, more measurable results, and automatically improve over time.

What is a Rules-Based System?

As the name implies, Rules-Based Systems use rules (in the form of IF x THEN y) to choose an action.

For example, "IF a user viewed a page about an outdoor firepit THEN show them retargeted ads about that outdoor firepit" or "IF a user views a video, add one point to their score".

What that means is somebody needs to make up a large number of rules ahead of time to try to handle all possible scenarios. Often times, the rules are just made up based on gut instincts. They are almost always gross simplifications of complex scenarios, strung together like a Rube Goldberg Machine. Sometimes, the simplified rules applied to complex scenarios lead to absurd outcomes (Dear e-commerce sites: just because I bought a firepit doesn't mean it's the first firepit in my firepit collection!).

That is the first major problem: Rules-Based Systems are too limited to solve most real-world problems, such as segmenting customers or responding to user behavior. (To say nothing of complex fields such as ethics.)

The second major problem is that Rules-Based Systems are expensive and time-consuming to maintain. Anybody who has tried to add a rule (or 1,000 rules) to an already-large system knows the struggle: it is extremely difficult to add new rules without introducing contradictary rules.

Before even making a change, you need to know and understand all the other rules and how they interact... which often look like this:

And if you happen to make a mistake (personally, I've never made a mistake but I know somebody who did), it could be many months before you even know it... all the while your system is sending false-positives or false-negatives. 

(From the company's point of view, just pray that the employee who set up that Byzantine set of rules never leaves!)

Finally: even if a company 1) could make up the perfect set of rules for a complex scenario like customer behavior, and 2) had a small-army of employees that never leave and never make mistakes that can be dedicated to maintaining those rules, the perfect Rules-Based System would still be obsolete the moment customer behavior changes. That is because intricate rules for one specific scenario don't usually work well for other scenarios.

Let me repeat that: when (not if) customer behavior changes, all the time and effort you put into your Rules-Based System is all for nothing.

In short: Rule-Based Systems are explicitly-defined models that do not change with new information. Perhaps they are better than nothing. Perhaps. But they are not what you should build your sales and marketing efforts on if you have any alternative.

The good thing is, Learning Systems are the perfect alternative.

What is a Learning System?

In contrast to Rules-Based Systems, Learning Systems observe data and continuously learn from it.

That is, while Rules-Based Systems quickly become out of date, Learning Systems automatically improve over time. And Learning Systems improve without the massive expense of having people maintain complex rules.

This may all seem like black magic, but fundamentally learning systems just find patterns and treat similar things similarly. Think of how Netflix or Hulu keep track and learn from what you watch, and then compare it to what people like you watch to make recommendations and keep you binge-watching. There is some complexity in execution, but it's a much simpler concept. And a concept more in line with how humans also learn.

Finally, Learning Systems' results are necessarily more measurable. It's the only way they can improve over time; because if they cannot measure results, they could not learn what actions are better and what actions are worse.

Conclusion: Learning Systems are the Future of MarTech

In short: Learning Systems significantly outperform Rules-Based Systems in complex scenarios such as sales and marketing intelligence, automatically improve over time using measurable results, and are much less expensive to maintain and improve.

Click360 is ahead of other platforms. We use real-time Learning Systems for segmenting customers based on behavior, predicting customer intent and prioritizing leads, and personalizing each customer's journey.

If you're ready to ditch the old and see how Deep Learning is fundamentally changing sales and marketing platforms, let's talk.