Behavioral Targeting – Rules and Personas

In my previous post, I described how to get started with Behavioral Targeting in Sitecore Online Marketing Suite, I used a Paint Store as an example.

For the store I created a mock-up, that looks like this:
Mock up of PaintStore

In this post I will create a simple prototype of the Paint Store, where I will use different rules and Personas, to show content according to the visitors onsite behavior (Behavioral Targeting).

WIth Paint Store, the goal is to sell paint – and my task is to show relevant paint according to the visitors clickstream.

To do this effect fully, it is crucial to have access to real-time session analytics data.

In my prototype I will not use all content profiles as explained in my previous post (because this is a…. prototype), in fact I will only use Content type and Category.

I will have 4 Categories; Blue paint, Red paint, Green paint and Yellow paint and 5 Content types; Products, News, Articles, Offers and Company info.

I create content, so I have two content of every content type in each category, this means:
- 2 blue products
- 2 yellow products
- 2 blue articles
- 2 yellow articles
- Etc..

I create two layouts in my prototype, one is a list with 4 elements and 4 spots (see screen dump below), this layout will be used, when a visitor clicks in the top navigation bar and on the frontpage.

Mock up of PaintStore, showing variables

The second layout is when a visitor clicks on a specific content item (see screen dump below), this layout will show the details of the selected item.

Mock up of PaintStore showing item variables

In this prototype – a visitor will hit the frontpage and we know NOTHING about the intends of this visitor (which rarely is truth, in real life you can look at parameters, such as external keywords, campaign triggers, previous onsite behavior etc., but again this is a prototype ;-)). So we would present a list for this visitor, which has products from each of our 4 Categories (Blue, Red, Green and Yellow) – we would also use the 4 spot areas to show different content types (News, Articles, Offers, Company info) in different categories.

.. And then we will “listen” and use the visitors clickstream to show relevant content, if a user clicks on products – he’s interested in products and we will show products in the spot areas, if he’s interested in yellow – we will show yellow content in the spots area – it’s quite simple, all we have to do is listen and have rules to present content.

However there a many ways to setup these rules and many parameters to consider, this is why this area is so exciting! In the prototype I will “listen” to:
- Category content scores (on a scale from 0 to 5)
- Content type clicks
- Where do the visitor click (which spots)

If the visitor is interested in Yellow products and most clicks come from the left bottom spot, I can then show the most relevant content in this spot, because I know for a fact, that the visitor pays attention to this spot.

The Rules
I have 4 spot areas and I use these 4 simple rules, to show relevant content:

Rule 1: “1 out of 4 spot areas”
If score is less than 5 in any category, then
- Show content from 4 different categories in the 4 spot areas

Rule 2: “2 out of 4 spot areas”
If score is greater than or equal to 5 and less than 10, in one category, then
- Show 1 spot with highest Category and Content type
- Show 1 spot with highest Category and second highest Content type (because we only have two of each Content type in each Category)
- Show 1 spot with second highest Category and highest Content type
- Show 1 spot with third highest Category and third highest Content type

Rule 3: “3 out of 4 spot areas”
If score is greater than or equal to 10 and less than 20, in one category, then
- Show 1 spot with highest Category and Content type
- Show 1 spot with highest Category and second highest Content type
- Show 1 spot with highest Category and third highest Content type
- Show 1 spot with second highest Category and highest Content type

Rule 4: “4 out of 4 spot areas”
If score is greater than or equal to 20, in one category, then
- Show 1 spot with highest Category and Content type
- Show 1 spot with highest Category and second highest Content type
- Show 1 spot with highest Category and third highest Content type
- Show 1 spot with highest Category and fourth highest Content type

The cool thing is, once content profiles and rules are created – it’s easy to change the values (test and optimize) to getter better online performance.

So now I have profiled my content and created the rules above and have Behavioral Targeting on my 4 spot areas, which should give better conversion rates, than a “static” website, which doesn’t triggers on behavior.

But there are so much more you can do, to get even better performance..

Personas
I have created four Personas, where I will match the onsite visitor behavior and check if it matches a known Persona – and if it does, use this Persona to show relevant content. The four Personas created are (da da):
- Blue paint Persona
- Red paint Persona
- Green paint Persona
- Yellow paint Persona

Each Persona is profiled with the different categories, this means that the Yellow paint Persona, has the values:
- Blue 4
- Red 4
- Green 4
- Yellow 20

So in order to get a match to this Persona, the visitor needs to have the scores above in the Category profile.
I will calculate and match the scores, using weighting between the scores, like this:

If Blue(OnSite) score are greater than or equal to Blue(Persona)value AND Red(OnSite) score are greater than or equal to Red(Persona)value AND
If Green(OnSite) score are greater than or equal to Green(Persona)value AND Yellow(OnSite) score are greater than or equal to Yellow(Persona)value AND
(Weighting in Blue, Red, Green and Yellow (Persona)value EQUALS Weighting in Blue, Red, Green and Yellow (OnSite) score) THEN
Set (Session)Persona to (matched)Persona

If a visitor match one of the Personas in the prototype, the video background changes and shows the color of the Persona (like a yellow video background).

In real life, a persona could have more parameters to match, like behavior – and you would know which call-to-actions to show onsite.

Behavior
Different Target Groups and Behaviors for Web visitors

It could also be interesting to look at behavior of the visitor, is he looking, confirming or buying – so I could present prober content and call-to-actions, you wouldn’t (shouldn’t) use BUY BUY BUY all over the page, if the visitor is in the looking phase! Here it would be better to focus on sign-up to newsletter, or add to wish-list etc. so we could focus on an ongoing dialogue for moving the visitor to the buying phase AND THEN present with buy call-to-actions.

In order to calculate on behavior in the prototype (show content according to visitor phase is not implemented), I use four variables (inspired by research by Wendy W. Moe):
- SUMCAT, which is the number of product category pages shown against total number of pages shown
- SUMPROD, which is the number of product pages shown against total number of pages shown
- SUMINFO, which is the number of information pages shown (this covers News, Articles, Offers, Company info in the prototype) against total number of pages shown
- DIFFCAT, which is the number of unique product and product category pages shown against total number of pages shown

I have configured three rules, to classify in which phase the visitor are, like this.

The visitor is LOOKING, if:
- SUMINFO are greater than or equal to 70%
- DIFFCAT are greater than 10%
- More than 3 pages are shown

The visitor is CONFIRMING, if:
- SUMPROD are greater than or equal to 50%
- DIFFCAT are less than 40%
- More than 3 pages are shown

The visitor is BUYING, if:
- SUMPROD are greater than or equal to 60%
- DIFFCAT are less than 20%
- More than 3 pages are shown

Again it’s easy to change the values, and test if this means better performance.

For more on clickstream analysis and Behavior predictions, visit the website of Wendy W. Moe, which has done a lot of research in this area.

The prototype
Below is a screen dump of the prototype.

PaintStore prototype

You can test it here.

Try to click on a category (like yellow), until you’re classified as an Yellow paint Persona, then change behavior, by choosing another categori (like click on products in the navigation and select a red product).

At the bottom of the prototype, theres a statusbar – showing the variables used for the rules:
PaintStore Statusbar

I hope the prototype shows, how much you can do with rather simple rules and content profiling – and sorry for the lack of good content in the prototype (dont’t you know that “Blue saves time”)..

If you add on parameters, like:
- Mouse movement
- Time pr. page
- Behavior on last 4 clicks
- Spot that doesn’t get clicked on
- etc.

And use these parameters in your rules and your Behavioral Targeting, would be even better.

It’s more easier than ever to use Behavioral Targeting on your website, the technology isn’t a problem any more (and it’s affordable), it’s just a question of “adding” prober resources.
By using Behavioral Targeting, you will optimize the way to listen to a visitor and reflect upon this with relevant content in prober context (looking, confirming or buying). In my prototype I have used a eCommerce site (without the basket and call-to-actions), but Behavioral Targeting can be used on all sort of sites.

The important question to ask yourself is “Why do we have a website?”, this should trigger thoughts about measurable goals for your website, like:
- you have an eCommerce site, your goal is to sell products (an alt. goal could be a sign-up for a newsletter)
- you have an newspaper site, your goal is returning visitors, time on site, register to newsletter/RSS
- you have an B2B site, your goal is collecting leads (call-me up forms, contact form etc)

Using Behavioral Targeting you can bring your goals closer to the visitor in the “shape” that fits the visitors Behavior.

Please comment if you have anything in mind, would love to hear your opinions and suggestions for additional parameters to use.

Bookmark and Share

Category: Behavioral Targeting, Online Marketing | Tags: , One comment »

One Response to “Behavioral Targeting – Rules and Personas”

  1. Erik

    Thanks for this very clear article Lars. Very useful indeed!


Leave a Reply



Back to top