ePrivacy and GPDR Cookie Consent by Cookie Consent Skip to main content

Step By Step Implementation Guide

Data Sources Implementation 

Get credentials

  • Request for access or find credentials to data source in 1Password vault dedicated for the client
  • Confirm that the credentials are valid ASAP after received
  • Confirm access to data sources match with the defined scope of work (that we are given accesses / credentials to all the data sources in scope and none of them are out of scope) 
  • Request for sample of data (Whether the access is available or not) - this should give the analyst a good overview of what to expect.

Initial data exploration

  • Prepare data sample description following the Data Exploration checklist; paying attention to:  
    1. List of fields
    2. How sparse the data is
    3. Data values sample (SELECT DISTINCT something)
    4. Distribution
    5. Volume
  • Prepare customer events description which includes the following. Also refer to How to Format Events 
    1. Types of events (eg: purchase, shipment, delivery, cancellation)
    2. Example of payload
  • Prepare a list of unique IDs from all events
    • Compare this against the identity graph defined with the client. For each data source's events, we should see the same types of unique IDs
    • Check if there is any kind of data validation necessary
  • Check the format of PS identifiers across all data sources:
    • "None", "0", "NULL" in text format can result in stitching huge customer entities
    • Is the format of the identifier the same across all data sources? "12345" and "12345.0" are different values for PS algorithm
    • Are there placeholder values which should be removed? 

Data source implementation

Implement the workspace:

  1. Extract
  2. Clean (exclude events we don’t need, remove unnecessary fields to remove, clean the values)
  3. Validate the Identity Stitching IDs
  4. Confirm the valid version/branch of format_events and cdp_db_loader components
  5. Write in Documentation:
    • General description, business assumptions of the workspace 
    • Comments in configs

Please refer to the following documentations:

Review, Finalize

  • Code review - Assign a senior analyst to do code review
  • Final checks:
    1. Double-check through queries
    2. Benchmarks with source data (eg. validate total web visits against GA numbers or total email subscribers with source. This is a quick way to identify if there are significant gaps in our data ingestion. 

Identity Stitching

Define identity stitching configuration 
Run PS workspace

Check the quality of Identity Stitching after running. 

In each project, there is always the potential to have edge cases where we wrongly stitched some customer profiles or create super entities. Refer to this guide for cleaning wrongly stitched customer entities without re-running PS, which can be a time and resource-consuming process. 

If the situation calls for it you can re-define the PS configuration and re-run identity stitching from scratch.

Attributes

In the business requirement gathering process, the PM should have defined a list of required attributes from the client following this Attributes Library template. 

Based on this list, analysts should first verify all the attributes requested that the data or events for them are available and possible to calculate. Double-check on the formula we should be using as well. 

It is also important to verify if attributes for any Destinations are available from each source's events. Refer to individual loader documentation for the required input parameters.

It is best practice to set first & last name as the first two attributes in the customer profile highlights if available.

When implementing a new source, always check and review existing "mixed sources"/"all sources" attributes ( attributes that use events from multiple sources ) if they need to be updated. A common example would be email or phone number attributes.

Finally,  implement the attributes in Business Explorer: Attributes Calculation configuration

Data Destination

  • Before configuring data destinations, ensure that you have the relevant access/credentials to the destination in 1Password
  • OAuth repo (usually managed by Meiro IT)
    1. Prepare OAuth repo, if relevant
    2. Confirm that OAuth repo is ready and the loader component can be authorized
  • Implementation
    1. Set up the connection between MI and CDP (create API “users” for API connection, and test them)
    2. Check what are the requirements for the data in the destination - depends on the destination, certain IDs need to be set as exported attributes
    3. Prepare exported attributes
    4. Set export destination in MI and MBE
    5. Set up alerts for: MI Workspace and Meiro Events monitoring dashboard. This is crucial for making sure that we are always aware of failed workspaces and issues on implementation before the impact is noticed by the end user of the CDP. 
    6. Test segment export with users - send a non-empty segment to the customer, confirm that the end destination receives the segment export and behaves as expected.

Final CDP and BE Settings

As a best practice, Project Managers need to set up the Business Explorer instance for going live

No Comments
Back to top