A process for predicting that someone is about to leave your website, so you can then change the user experience to stop them leaving.
The predictive capabilities of Exit Intent platforms differ.
The most basic track mouse movement and they look to see whether the user is moving the mouse to the top of the page. There’s a belief that if they are they’re about to leave the site, but the accuracy of this is questionable.
There are more complex, and more accurate ways of predicting user behaviour that consider more than mouse pointer movement. Some platforms take into account what the user has done on previous pages, whilst others consider mouse movement within the page itself, and not just if they’re heading towards the top of the page.
As you’re reading this on the Nudgr site, and we’re a platform that helps you predict Exit Intent in online forms and checkouts, it’s worth mentioning how we work. Nudgr uses a large (and growing) number of behavioural and environmental metrics and we use machine learning to predict Exit Intent.
The use of machine learning allows our algorithm to pick the right metrics (we call them features) for that particular website. As we’re focused on the checkout process we measure every key press, every movement between fields and timing data, along with environmental metrics.
Every website is different, not just in the way that it looks and feels but also in the demographic that uses it, and we believe that the best way to predict Exit Intent is to have an algorithm that can shift and change to get the best performance.
There are some basic rules to a great (high converting) popup:
It's also worth noting that the text you use in your popup is important. Not just for converying your message and being "on-brand", but also to avoid confirmshaming.
(I created a gallery of exit intent popups that I found, browse it and get inspiration for your own)
No, but they’re changing. The idea of interrupting the user experience is fundamentally flawed, because if you get it wrong and interrupt someone when they’re not intending to exit then you’ve made their experience worse.
If you monetise that person, by giving them a discount or money off, then you’re reducing your margin, and if you show a popup to someone who was going to convert then you’re reducing profit for no reason.
The future of Exit Intent lies in improvements in two areas: Improving the Accuracy and Improving the Type of Message. I’ll cover both of these in the next two sections.
No. Exit Intent has started out by being focused on a popup that takes up the entire page and makes the user consider the message. This is a basic, but understandable, way to use the technology and it has delivered results.
The problem with large popups is they interrupt the user experience as they take over the browser window.
If an Exit Intent platform predicts intent milliseconds prior to the user leaving then the message has to be large because there’s so little time to grab their attention.
If you can predict Exit Intent seconds, or tens of seconds, before a user is going to leave it allows you to create more subtle messages, ones that don’t interrupt the user experience but help remind them about why they’re making a purchase, or filling in a form.
Some sites already use subtle messages around scarcity, or social proof, or they give the user confidence about the transaction. The limited data around these popups is positive, and the sites continue to to use them suggesting that they are both profitable and sensitive to the user.
Live Chat gives you the ability to talk to another human being, to offer feedback or to fix a challenge that you’re having with transacting or giving information via a form. Many live chat platforms allow you to launch a chat to a user after X seconds on a page which is a poor way of engaging people, but by combining Exit Intent prediction and live chat it allows website visitors to engage with another human being based on them experiencing issues.
Nudgr is a platform that gives you incredible flexibility on how you engage users that we predict are about to exit (and our algorithm can predict Exit Intent well in advance). You can plug in parts of your stack (live your Live Chat) or you can use subtle messages that appear in parts of the page, or that appears from the side or the base of the page. You can even use Nudgr to remove elements on the page to focus the user on a task.
A good starting point is to stop using mouse movement alone. There are countless reasons why someone would move the mouse pointer to the top of the page, here’s some of them:
It’s worth noting that some Exit Intent platforms predict intent based on you moving the mouse away from the window, on any side and not just on the top. If you have multiple monitors, and you move the mouse from one to another, this will trigger the Exit Intent message.
Moving from one basic measurement (mouse movement) to tens or hundreds of different signals gives you a much wider dataset to mine, giving greater accuracy. The use of Machine Learning to then prioritise the signals will increase accuracy further.
(Read more on creating a great exit intent popup and use our guidelines to produce a high converting overlay)
We asked 913 consumers how they felt about exit intent popups, and why they felt that way.
The basics of Machine Learning and how it will help you improve your website conversion rate
What makes bad exit intent popups so bad? Learn from these examples of poor popups.