Help Center

Explore our knowledge base
article

Discover our new targeting method: Smart Targeting

Optimise your targeting approach thanks to Smart Targeting

Thanks to iAdvize, you can choose to target specific visitors. Some of them can have a chat with you whereas others are not disturbed as they are browsing on your website.

Targeting can be done with Custom Targeting: you choose situations in which notifications are displayed or not. It is however not easy to anticipate which situations are the most appropriate to display notifications and create all targeting rules which are necessary to correctly deal with each specific case.


That is when Smart Targeting comes up: our intelligent method automatically targets visitors at the right moments.

⚠️The public beta version bêta* will be available mid-march 2020

What is it?

This feature is available in Engagement > Campaigns. Create a campaign for your website and use our new method "Smart".

Visitors are targeted whenever the system detects that they are very likely to start a chat conversation if a notification is displayed. It therefore avoids visitors who could be disturbed by such notifications and the ones who are going to make a purchase anyway. It really focuses on visitors for who a conversation makes a difference between a purchase and a bounce.

This new method can be chosen at step 3 of a campaign creation :


So that you can easily control the targeted audience, you can choose between two options:
  • Volume management and targeting quality you wish to enable. The more people your targeting strategy adresses, the more visitors you will target whilst maintaining your performance according to your objective.
  • You can also restrict the targeting by URL and therefore create a Smart Targeting campaign according to some specific categories:

How does it work?

Smart Targeting is based on a Machine Learning system which is specifically built up and trained for your website. It measures how likely a visitor is willing to chat at a given moment by using what is known about this visitor at this moment and the impact of displaying/not displaying a notification. If the difference between displaying and not displaying is high, the visitor is more likely to be targeted.

The impact is calculated each time the JavaScript tag is loaded by a visitor. By comparing this impact to the ones which have been observed during the system construction, the system determines whether it is high or low. The level of targeting which is chosen in the platform uses this ranking to determine whether a display is necessary or not at this moment.

All data known about the visitor are collected thanks to his/her browsing information. It can be the device which is being used, the time spent on the website, how fast does he/she browses and above all, the page URLs he/she visited.

This system is regularly updated: data which have been collected over the past week are used to trained the current system. Visitors' browsing data compiled during the week-end are therefore collected for the following week-end. A visitor does not always make a purchase the same day he/she starts browsing, but the system takes into account the purchases which are made the three following days, which is enough to differentiate profiles patterns.

Limitations & evolutions

Smart Targeting contributes to optimise the iAdvize solution impact on conversation rate on its own, without any human intervention - and therefore without any bias, loss or error. You do not need to spend much time to set up your targeting strategy anymore.

The machine has its own limitations and has to discover statistical regularities on its own in oder to make decisions which could be obvious for a human user, with the hep of its specific knowledge. If you precisely know which visitors have to be targeted on your website, then Custom Targeting may be more performant. Only a comparative usage and test can tell you for sure.

It is our very first version of this feature and it focuses on one indicator optimisation: the conversion rate. If required, we will consider an evolution to optimise new indicators: average basket value, turnover or click rate so that we can better choose the appropriate moment to display a notification.


If you have specific needs, please let our Product Team know on the dedicated portal.