BLOG

Why AI is essential for continuous digital experience monitoring

Digital experience monitoring used to focus on simple metrics like availability and backend performance. But those alone won’t guarantee your customers always have a positive experience. Here, we look at how AI allows you to take digital experience monitoring to a whole new level.

Introduction

Modern digital marketing is about delivering content across a variety of platforms and using this to generate leads. Your customer-facing website probably sits at the heart of your strategy. Chances are, you are sharing blogs and news, giving access to your online support, providing product updates. Often, you will also be using gated content to generate leads. All this is probably tied in with your CRM system in the backend. Most likely Salesforce, Hubspot, or something similar. 

There’s no doubting the vital role played by marketing, but all too often, they lack technical resources. In most cases, the marketing team is completely responsible for creating and running the website. They may have occasional access to a front-end designer or a web developer, but more often than not, they will rely on contractors or internal team members for this. So, how can you make sure your website is delivering? How can you avoid problems that will directly impact sales and revenue? Well, for years, the answer has been digital experience monitoring.

What is digital experience monitoring?

Digital experience monitoring (DEM) refers to systems that monitor the experience your customers receive on your web app or site. Traditionally, these systems ran in your backend and allowed your DevOps and Admin teams to identify potential problems quicker. These classical DEM systems tend to rely on relatively passive monitoring. For instance:

  • Monitoring site availability and accessibility
  • Measuring the performance of your backend database
  • Checking for support tickets indicating system issues
  • Relying on 3rd party issue reporting tools, like DownDetector

All of these approaches are great at identifying large-scale issues. For instance, spotting when a major piece of infrastructure has failed. Or identifying that your servers are under too much load and need to be scaled up. The problem is, these issues aren’t the only ones that ruin the digital experience. 

Issues that are invisible to DEM

As we said, DEM is great for identifying obvious issues. But what about all those other issues that can ruin the user experience?

Broken links and 404 errors. We’ve all had bad experiences when you visit a website and half the links are broken. Or they lead to a dreaded 404 error. The problem is, these issues are ever more likely as websites increasingly use dynamic content.

Missing or incorrect content. There’s something rather sad about seeing a web page with a missing image. Hopefully, there’s a bit of alt text there so you can see what the image was meant to be. But more often, you just get that sad broken image icon.

Why AI is essential for continuous digital experience monitoring

Equally bad, is when there is missing 3rd party or embedded content. 

Failures in localization or internationalization. Websites are increasingly targeting an international user base. That, in turn, makes localization and internationalization critical. Errors here can be subtle but very confusing. For instance, to a German, one thousand and one is written 1.001 rather than 1,001. In most of Europe, dates are written in DD/MM/YY format. 

Problems with forms and interactive content. Most websites embed one or more interactive elements. For instance, your marketing team may be using gated content to capture lead details. You might have a newsletter for people to sign up for. In general, these are driven by CRM systems like Salesforce or Hubspot. The trouble is, it’s all too easy to mess things up. For instance, someone signs up for one white paper and gets sent something completely different. 

Why testing may give the solution

The issues above are pretty standard for all web-based applications. If you were talking about your core company product, your engineering team would be tasked with checking for exactly these issues. But of course, all too often, marketing teams struggle to get access to technical resources to help them deliver their content. If you did have access to engineering resources, your QA team would look at whether your site content is working, whether there are missing images, etc. They would create detailed test plans and use automated testing to speed up their work. The problem is, even if you had this luxury, this testing would be completed before release. In other words, it happens in a much more controlled environment. 

But imagine if you could tools like these to check your marketing website? That would boost DEM from doing purely passive monitoring to active checking. You might call this continuous digital experience monitoring. 

What’s the blocker?

Using testing in production might seem like a no-brainer. A quick google search will turn up countless articles about Shift Right testing. The problem is, they don’t really mean active ongoing testing of your production. More typically, they refer to approaches like canary testing, A/B testing, dark launching, and system monitoring. In other words, the main aim is to test the backend system rather than the frontend experience. 

So, what makes it hard to test your frontend? Well, for a start we are talking about automated testing here. There’s no sense in trying to use manual testing for automated monitoring! Automated testing generally means script-based testing using frameworks like Selenium. This is when your problems start.

  • Poor Reliability. Scripted tests are notoriously fragile. Even simple UI changes can break them. That means the test team is constantly faced with ongoing test maintenance. The last thing they want is for you to create more opportunities for tests to break! 
  • Limited scope. Selenium and related frameworks are also really bad at testing dynamic or embedded content. That makes them ill-suited for the sort of monitoring needed here.
  • Incomplete testing. Test scripts are pretty limited. Typically, they only focus on broad functionality. They certainly never try to check the entire content of any page. Writing scripts to do so would take an age. And then they’d be inordinately slow to run.
  • Slow test creation. Test scripts are effectively mini software projects. You have to write them, test them, and then debug them. This process has to be repeated for each browser and device. As a result, most test teams struggle to automate more than half their tests. 
  • Poor scalability. Test automation tools need to run on servers. This usually means servers in your own office. These servers are probably already fully committed for testing new features for release. 

Put simply, all these mean that scripted testing is a poor fit for modern DEM.

Why is AI key to delivering continuous DEM?

Here at Functionize, we are great believers in applying AI to solve problems. Our mission has always been to transform testing for everyone. All our tools embed different forms of AI to ensure they enable smarter testing. But why is that significant? After all, many companies claim to use AI. Well, AI really is transformative for testing. Let’s see why. 

Reliability

Test scripts break because they rely on static identifiers called selectors to determine which elements you are trying to test. Machine learning allows you to identify the elements dynamically and reliably. A test script would break if you move the buy now button on your webshop. Or if you rename it “add to basket”. But our approach means the system understands your site like a human would. So, it isn’t fooled by moving elements, changing their names, altering their style, etc.

Complete testing

AI allows you to test the entire UI, not just selected elements. For instance, you can compare screenshots between tests and see when something unexpectedly changes. That allows you to spot the difference between an advertising banner that always changes and some actual error like the wrong image appearing, or a missing image. Functionize tests are also able to properly test embedded and dynamic content. That makes them far better for modern DEM.

Simple test creation

Architect is our AI-powered test recorder. You can create a test simply by recording your actions as you interact with the site. Behind the scenes, Architect is building an advanced ML model of your site, learning how it works. It does this by recording a vast volume of data, which is added to each time a test runs. The upshot is anyone can create robust tests. 

Cloud-scale testing

The final clincher for using Functionize as a basis for modern DEM Is the Functionize Test Cloud. This allows you to launch tests from anywhere in the world, and to run tests at an unlimited scale. The great thing is, each test runs in its own virtual machine with an IP address suitable for the location it’s in. In turn, that allows you to test for localization and internationalization issues. It also allows you to check for performance issues with your backend using things like our visual completion metric.

Conclusions

Functionize offers you a great way to modernize your digital experience monitoring. Better still, you don’t need to be an expert developer or DevOps guru. Functionize will empower anyone to become an expert. You’ll be able to create and run advanced AI-powered tests against your production and spot errors before they cause problems. And this isn’t just theoretical—we help one of the biggest CRM producers in the world to ensure their customer website is fully operational. This has helped them catch problems like 404 errors, missing or duplicated content, and page misdirects before their customers noticed. If you want to try Functionize for yourself, you can sign up for our free trial. Or book yourself a  live demo with our team.