Data Taxonomy

Learn more about data taxonomy in the 6sense Academy training videos.

6sense Data Taxonomy settings help you to classify your data with the assistance of AI-driven taxonomy models. 6sense Predictive models use these classifications to trace buyers’ intent. Classification is also known as “data mapping”.

For classification, 6sense AI-driven taxonomy models analyze and MAP the data across the system to different activities or product categories. You then verify and update the mappings per your business needs.

6sense maps data from the following sources:

Website
Customer Relationship Management (CRM)
Marketing Automation Platform (MAP)

You can control and view data used by your Predictive models. Mapped data builds and trains the Predictive models to generate Predictive scores and to make data available in 6sense. The more accurate the mappings are, the more granular and accurate the signals will be.

The mapped data is used in the downstream systems to build, train, and run Predictive models. The accuracy of the taxonomy process is important. Most data in the system has a purpose associated with it. Activity and Product taxonomy views uncover the purpose and the “why, how, and what” of the stored data.

Prerequisites

The 6sense Predictive add-on.
Your user account must have a Primary Administrator, Administrator, or Operation User role.

If your organization sells a single product, 6sense builds product taxonomy models automatically, and additional taxonomy setup and verification are not required.
If your organization sells multiple products, you need the additional setup and mapping verification.

Set up data taxonomy

Each organization has unique products. To accurately map the data to the products, 6sense builds product taxonomy models for each of its customers who have licensed the 6sense Predictive add-on. Since the model differs for each customer, we need the customers to help us build and train their product taxonomy model.

To access the data taxonomy interface, in ABM go to Settings > Data Settings > Data Taxonomy.

Data Taxonomy page

Most data in the system has a purpose associated with it. Activity and Product taxonomy tabs uncover the purpose and the “why, how, and what” of the stored data.

There are three tabs:

Overview: A summary of activity and product taxonomy models.
Activity: Represents the Activity taxonomy model. The AI-driven taxonomy models analyze tons of data across the website, CRM, and MAP systems and predicts the activities associated with those. For CRM and MAP data, these predict additional signals to identify the data precisely.
Product: Represents the Product taxonomy model. The product taxonomy models are used to map product categories to the data. However, in many scenarios, data does not relate to a product is unmapped by the model. For example, typically the careers page on a website will not have a product associated with it and will be left unmapped.

Overview tab

Activity Summary statuses

The taxonomy model maps the data to various activities or products based on the probability score. However, users can manually override each of the mappings derived by the taxonomy system.

Mapped: These are data that have activities or products mapped to them. The probability score of each mapping is higher than the defined threshold.
Unmapped: The data can have unmapped status if the taxonomy model mapped with a probability score less than the defined threshold.
Do Not Map: This status provides leverage to the users by providing them an option to mark the data as “Do Not Map” if they have reviewed the data, want to keep it unmapped, and do not wish the data to keep showing under Unmapped or Mapped status. On assigning “Do Not Map”, users can move ahead and verify all the other data by filtering out this option.
For example, if a record is mapped with “Activity1” but the user wants to remove the default mapping, then “Do Not Map” status should be selected.

Activity tab

Apart from different activity types associated with the CRM and MAP data, the data can have more attributes to identify the context. Accurately mapped data becomes the input to the Predictive models.

6sense Taxonomy model analyses the CRM and MAP data and maps the action and channel associated with each record.

For example, the “Unsubscribe Email” activity type from MAP would be mapped to “Action – ‘Unsubscribed'” and “Channel – ‘Email'”.

Activity > Website URL tab

The dataset doesn’t directly impact the models. Taxonomy mappings in the Activity tab assigned to the dataset do.

For example, if a webpage is mapped to Blogs under Activity then that webpage will impact the model, while if a webpage is mapped to Careers then that webpage won’t.

The following table shows the impact of different mapped activities on the models.

Mapped Activity	Used in Intent Models?	Used in Reach Models?
blogs	YES	NO
campaign	YES	NO
careers	NO	NO
case_studies	YES	NO
chat	YES	NO
communities	YES	NO
company_research	YES	NO
conferences_tradeshows	YES	NO
contact_us	YES	NO
data_sheet	YES	NO
demo	YES	NO
home_page	NO	NO
login_page	NO	NO
pricing	YES	NO
product_landing_page	YES	NO
product_research	YES	NO
product_trial	YES	NO
search	YES	NO
solutions	YES	NO
support	NO	NO
training	YES	NO
unsubscribe	YES	NO
video	YES	NO
webinars	YES	NO
white_paper	YES	NO

Activity > Marketing Automation Platform (MAP) Tab

The following table shows the impact of different mapped actions on the models.

Mapped Action	Mapped Channel	Used in Intent Models?	Used in Reach Models?
attention	marketer	NO	NO
attention	sfdc	NO	NO
click	ad	YES	NO
click	email	YES	NO
click	form	YES	NO
click	web_page	YES	NO
delete	lead	NO	NO
error	email	NO	NO
error	web_page	NO	NO
filled	form	YES	NO
other	other	NO	NO
send	alert	NO	NO
send	email	NO	YES
unsubscribe	email	YES (Used in contact intent models)	NO
view	email	YES (Used in contact intent models)	NO
view	form	NO	NO
view	video	YES	NO
view	web_page	YES	NO

Activity > Customer Relationship Manager (CRM) tab

The following table shows the impact of different mapped actions and channels on the models.

Mapped Action	Mapped Channel	Used in Intent Models?	Used in Reach Models?
attended	live	YES	NO
attended	on_demand	YES	NO
attended	other	YES	NO
clicked	email	YES	NO
clicked	link	YES	NO
downloaded	email	YES	NO
downloaded	link	YES	NO
engaged	live	YES	NO
filled	registration	YES	NO
filled	search	NO	NO
filled	web_page	YES	NO
invited	contact	NO	YES
invited	email	NO	YES
not_attended	live	NO	NO
positive	contact	YES	NO
positive	email	YES	NO
responded	contact	YES	NO
responded	email	YES	NO
responded	telemarketing	YES	NO
sent	email	NO	YES
sent	list	NO	YES
sent	live	NO	YES
sent	search	NO	YES
sent	web_page	YES	NO
sent	telemarketing	NO	YES
view	email	NO	NO
view	web_page	YES	NO
other	email	NO	NO
other	live	NO	NO
other	search	NO	NO
other	telemarketing	NO	NO
other	other	NO	NO

Activity > Customer Relationship Management CRM > Campaign tab

Mapped Type supports the “Attended Webinar” use-case within segment creation and SI timeline. Mapped Type provides additional granularity to the dataset and helps identify the campaign in which the activity was “Attended Webinar”. The field can only be manually updated, the taxonomy models don’t create mappings for mapped type.

Permitted Combinations for Mapped Action/Channel/Type

Mapped Action	Mapped Channel	Mapped Type
Attended	live	webinar
Attended	live	field event
Attended	live	tradeshow_conference
Attended	live	other_in_person_event
Attended	live	–
Attended	on_demand	webinar
Attended	online	webinar
Attended	on_demand	–
Attended	online	–
Engaged	live	webinar
Engaged	live	field event
Engaged	live	tradeshow_conference
Engaged	live	other_in_person_event
Engaged	live	–
Engaged	on_demand	webinar
Engaged	online	webinar

Activity > Customer Relationship Manager (CRM) > Event Tab

The following table shows the impact of different mapped event actions and channels on the models.

Mapped Action	Mapped Channel	Used in Intent Models?	Used in Reach Models?
attended	other	NO	NO
attended	meeting	NO	NO
attended	virtual	NO	NO
attended	call	NO	NO
attended	other	NO	NO
blocked	other	NO	NO
blocked	virtual	NO	NO
blocked	call	NO	NO
blocked	meeting	NO	NO
invited	other	NO	YES
invited	call	NO	YES
invited	meeting	NO	YES
invited	virtual	NO	YES
booked	call	NO	NO
booked	other	NO	NO
booked	virtual	NO	NO
booked	other	NO	NO
booked	meeting	NO	NO

Activity > Customer Relationship Manager (CRM) > Task Tab

The following table shows the impact of different mapped task actions and channels on the models.

Mapped Action	Mapped Channel	Used in Intent Models?	Used in Reach Models?
answered	call	NO	YES
left voicemail	call	NO	YES
no voicemail	call	NO	YES
other	call	NO	YES
click	email	NO	NO
open	email	NO	NO
other	email	NO	YES
reply	email	NO	NO
sent	email	NO	YES
other	other	NO	NO
linkedin connection	social media	NO	NO

Rules-based mapping

If your data isn’t standard and requires overrides, rules-based mapping provides you with a high level of control over overrides. It lets you add rules using the data source fields and define the new mapping to apply to the resultant data set.

Since there are multiple levels of overrides available, there is a defined precedence hierarchy. In terms of precedence: Manual Override > Rule-Based Override > AI-driven mapping, which means that if the user is applying rules on a dataset, but a sub-set has manual overrides, then the manual override stays on that sub-set. You can override the desired mapping by manually overriding that subset.

The filter section has a filter for mapping/override mode. By selecting Manual Override, you can see the manually overridden records and then update accordingly.