If you have not heard or seen it yet, in the Fall 2016 Release, we were super excited to launch our ContentAI (Predictive Content) for Email. This included a revamp of the ContentAI App (previously known as Predictive Content).
ContentAI is an Application Marketo offers that helps you auto-discover your content, engage your web visitors and email recipients with the most relevant, recommended content powered by machine learning and predictive analytics. See more in Docs here:Predictive Content - Marketo Docs - Product Docs
Here are regular FAQs on ContentAI we hear from customers and those interested in the App:
Where do I learn more about Predictive Content?
- Marketo.com Website Product Page: ContentAI Recommendations - Marketo
- Customer Success Story using ContentAI: Case Study / Webinar Video
- Webinar: Behold the magic of Predictive Content
- Marketing Nation Summit 2016: Predictive Content: Amplify your content through machine learning
- Request a demo from your Account Manager
How do I find the ContentAI App? Is it the same as Web Personalization?
ContentAI is a paid for stand alone App offered by Marketo. It is a separate product to Web Personalization. If you have purchased the ContentAI App, you will see a ContentAI icon on your My Marketo page. If you are interested in hearing more about the App, speak to your Account Manager.
Are there any restrictions on where the content resides? Does that content have to reside in Marketo or can it be in any CMS?
How does ContentAI discover my website content?
Marketo ContentAI "listens" to your website activity, and when any web visitor clicks and views a specific content piece (HTML, PDF, Embedded Video, PPT) on your domain, we capture and discover the content title and content URL and start tracking the content's views and engagements. This discovered content piece is then displayed under the All Content page. For set up details of Content Discovery see: Getting Started with Predictive Content.
Does the ContentAI reporting also show which content has the highest engagement?
Yes, the All Content page shows all web content (html pages, videos, pdf, etc...) on your domain that web visitors viewed. These results can be displayed from highest views to lowest.
On the ContentAI page, the reporting shows how many web visitors or email recipients clicked on theContentAI piece and can also be sorted to see which was the most engaged piece of ContentAI per channel (web or email) or overall.
How many contacts or content do I need to start using ContentAI?
ContentAI relates to web content. Content Discovery will find in most cases, we have seen hundreds of content pieces on your domain. The marketer can then go ahead and approve and enable these discovered content pieces for Predictive Content. You can also manually add content pieces to the collection. It is recommend you have 10-15 pieces of content enabled for Predictive (for any channel) to start and keep adding more to this.
Can you create your own Conversion Criteria?
We track and display the views on content and also clicks on ContentAI. In addition, the metrics we show are conversions (direct and assisted) per content piece, which shows attribution to the content piece in bringing in new People.
- Conversions: The number of visitors who clicked on recommended content and completed a form in the same visit
- Assisted Conversion: Visitors who clicked on recommended content in a past visit and completed a form later
Currently, you cannot create your own conversion criteria.
Can you clarify on how the algorithms predict the content?
The ContentAI algorithms use machine learning and predictive analytics to decide and recommend which pieces of content are relevant for each visitor or email recipient individually. We use a mixture of algorithms that are constantly optimized and improved for best performance and results analyzing clicks and conversions. The algorithms are based on trending, popular content viewed on your website, new pieces of content added to your site, and also takes into account URL paths and user behavior clicks and visits.
Will ContentAI show content the visitor or email has already received?
There are certain checks and validations in place in order to provide the best experience for the end user receiving ContentAI. These include:
- The Recommendation Bar will not recommend a piece of content the web visitor is currently browsing
- The Rich Media Recommendation will not show repeat predictive content on the same page and if the user clicks on a piece, we will not recommend the same piece of content they clicked on during that visitor's session.
- In Predictive Content for Email, a email recipient will not see the same Predictive Content more than once in follow up emails sent with Predictive Content in them.
Does Predictive Content now take into consideration any type of segmentation? So predictive content for segment A versus segment B?
No, it is not related to segments, and this would be more of a rule-based style Web Personalization campaign. Instead, Predictive Content takes into account many variables to calculate before producing the Predictive outcome result for each visitor. Predictive Content can also be categorized and shown per category in the Rich Media and Email. See more here: Set Up Categories in Predictive Content
Is the Predictive Content picked at random, or is it based on customer criteria?
The Predictive Content is prepared and approved by you. You also select the look and feel of the outcome and where it will appear on your site or in your email body. However, in the end it's the Algorithm that decides which of the approved Predictive Content pieces is best to show the current web visitor or email recipient.
Displaying Predictive Content on Web or Email
What are the Character Limits (Char limits) for predictive content on Web and Email?
- Content Name: 255 Chars
- Bar Title: 110 Chars
- Rich Media Title: 50 Chars
- Rich Media Description: 215 Chars
- Email Title: 50 Chars
- Button Label: 17 Chars
The char limits are in place in order for the text to fit into the defined styling and look for the bar, rich media and email layout.
Can you see which pieces will be promoted in the email before sending the email?
As the marketer, you can prepare and approve the piece of content for each source (email, web). In the end, the Predictive Algorithm decides what the end user will receive. For Predictive in Email, you can preview the layout and category of the Predictive Content, however the Predictive image is not seen until the recipient opens the email, so what you see in the preview is just an example, and will not necessarily be the image the recipient sees. See the process for setting it up and previewing Predictive Content in Email here.
Can we use only Marketo landing pages, or anything under our company webpage domain?
Yes, you can define all html pages, including landing pages, as content pieces to use in predictive. You can also manually add new pieces of content to the application.
Do you recommend Predictive Content for Marketo landing pages? Particularly Thank you pages?
Predictive Content for web in the form of the Recommendation Bar or Rich Media is suited for all types of pages. We have customers using rich media on their home page (3 pieces of Predictive Content in horizontal format) below the fold of the home page. We have the Bar running on Blogs or Resources pages. A Thank You page or Thank You email is a great use case to add Predictive Content as these pages or email, usually do not have any call to action, and adding Predictive Content can help create more clicks and engagement by the viewer. You can categorize your predictive content for Rich Media and Email so it would be recommended to show on that Thank You page Predictive Content related to the thank you submission (eg. white paper, case study, webinar, blog post).
Can you alter the code of the predictive content to be flush with a font style/coloring?
Can you use categories in the "web" or "rec bar" versions like you can in emails?
You can use Categories in Rich Media for Web and in Predictive for Emails. Currently, the Recommendation Bar does not support categories. Categories allow you to group your content and present predictive outcomes on web (rich media) or email. For example, you can work only with blogs, or with content in a particular language.