To get a hold of every customer profile has increasingly become challenging, since customers get in touch with various brands on different online platforms.
Customers who have their contact information contained in the data base are Known users and the Unknown users, who neither have customer profiles nor have authenticated in any channel. They can’t benefit from unique identifies, who help companies collect meaningful profile attributes and create unified customer profiles for them.
Every unique visitor has an ID. These IDs indicate the first-party data source along with the timestamp of identification.
- • Select Source Fields, use the profile data that is stored in your web and choose the fields that represent IDs with demographic data
- • Add rules to merge duplicate records
- • Configure match rules and conditions
This stage offers you 2 options to work with. One in which you include all records and create unified profiles for known users including their corresponding unknow users IDs. One where you include all records that create separate profile records for unknown users.
With the new feature Data ingestion, companies have customer data stored from their website, as event data, directly to their in-house or third-party services. As long as your website has an authentication flow integrated with an authentication service, the web data will contain both known and unknown users.
Profile personalization is based on user behavior, demographics, previous transactions and it converges unknown users with known users. Data unification is a feature that enables personalized content to help customers get to the most relevant products or services that they are interested in at that point.
It’s called unification because on this instance we can bind Customer insights’ data with user profiles.
What does the prediction feature give you in advance?
• It identifies high-value customers and process it insight
• Personalized campaigns can be run through strategical customer segments creation that can target sales, marketing and support efforts
• Guide product development
• Take your customer outreach higher with sales-optimization or marketing strategies
• Use loyalty or reward programs for high-value customers
The most important prerequisites for this new feature include:
• Transaction history should be for at least 1 year, across multiple dates and on the same customer ID
• Web activities, loyalty activities or customer profile information
• The contributor permission is a necessity
• A unique identifier is needed to match transactions to an individual customer
• Less than 20% missing values in required fields
Define model preferences
To accurately define the customer lifetime value, a comparable period of historical data is required. The respective model only predicts active customers who had, at least once, a transaction.
The model calculates purchase intervals and sets the intervals manually. It defines the percentile of high-value customers.
The model calculation uses the 80/20 rules. Customers who have contributed to the cumulative revenue for your business, makes up to 80%, and are considered of high-value.
Add required data
Add data for customer transaction history. Proceed with SalesOrder or SalesOrderLine and if the activity hasn’t been set up, select here to create it.
For Activities, if semantic mapping occurred you can choose the specific attributes or entity. In case there hasn’t been any semantic mapping done, you can select edit and map your data. You can review your attributes and select next.
Review and run the model configuration
Before you create the prediction, you can review and make a summary of the configuration. It provides you with changes before you complete the process of prediction, since the finalization of data may take several hours depending on the amount of data used.
View prediction results
Customer Insights assess how the AI model performed in predicting the high-value customers as compared to a baseline model.
There are 3 primary sections of performance for data.
- A – The model predicted 5% more high-value customers as compared to a baseline model
- B – The model predicted between 0-5% more high-value customers as compared to a baseline model
- C – The model predicted fewer high-value customers as compared to the baseline model
The data consists of email and meeting activity. They are currently supported in specific regions, for instance Europe, North America and the UK.
As with every feature, there are a few prerequisites to take into consideration before applying the new features of this section.
• An active Office 365 cloud license
• Customer profiles based on business accounts
• A Microsoft Dataverse
• Administrator permissions
• Consent from your Office 365 tenant administrator
How to configure the enrichment?
Go to Data, select Enrichment and Discover tab, proceed with enrich my data on the Account engagement tile.
Only data from the listed email addresses is going to be aggregated. The total number of email addresses must be between 2 and 2500, any number higher than this will not be included in the process of relevant communication.
Remember to receive the consent of the Office 365 tenant administrator and proceed with the standard operation of enrichment. Select a profile, provide a name, save and return to the enrichment page.
First-time attempt of enrichment
It’s important to obtain the access of the Office 365 tenant administrator before downloading the data from Office 365, which will take some time depending on the size of your of your Office data. Running an enrichment will allow Microsoft to process the data within the Office 365 to create aggregated insights, which will later be added to your Customer Insights environment.
When the enrichment process is completed, you can review the results. The number of enriched profiles will be showcased over time.
With the calculating system, you’re provided with an engagement score, ranging from 0 to 100, that will display measures of the account engagement across emails.
Customer card
Enrichment is available by selecting a customer profile and then customer card. Through this method you’ll get access to the engagement score, the total number of emails and meetings aggregated over the last year.
In order to be able to unify contact profiles you require foreign keys to connect them.
However, as with every feature, we have limitations.
• Contacts without a link to an account are dropped
• Deduplication of an account leads to a winner record that is identified based on the merge preferences. Thus, a loser record, or any other contact that is bound to a loser record, will be dropped.
• An account can have multiple contacts, but is linked to a single account
• The maximum number of attributes available to a contact is 17 mapped in the semantic mapping.
The proper lining up of the unification process is as followed:
Source field
It helps you select the entities within your datasets that you want to unify, for instance name, address, phone number or mail.
Remove Duplicate records
Lets you set up rules for eliminating duplicate records within an entity
Matching conditions
This step-in unification defines the match order and rules for cross-entity matching. This step requires at least two entities.
Unified customer fields
Choose and exclude attributes to merge within your unified profile entity
Semantic fields
It maps your unified contact fields to semantic types.
Set the relationship between contacts and accounts
It connects your contact data to its corresponding account data.
• For the foreign key
Choose the attribute that connects your contact entity to the account.
• For account entity
Choose the account entity associated with the contact.
Review
You will see the summary of the changes, unified profiles and the results.
