As an amateur musician, I love a good music analogy. The beauty of singing in a choir is achieving perfect harmony: different voices singing different tones, resonating together in a manner that’s a joy to listen to. As with vocal harmony, data harmony works in a similar way. Harmonized data may be complex, varied and sourced separately, but it still plays well together.
Using Marketing Cloud Intelligence, you essentially put a pitchfork into your data – combining, enriching, and fine-tuning it so you can use it for analyses, visualizations, and automated insights (Einstein Marketing Insights). Being an advanced-level analytics platform, Marketing Cloud Intelligence offers a wide array of methods to achieve this; choosing the right method from the start will save you time and energy. In this post, I’ll cover which data harmonization method you should choose for specific business use cases.
Enriching Data
Data Classification
Group together and aggregate data based on a classification file.
What It Does
Granularity is a good thing in most cases but sometimes you get more value from taking a birds-eye view. Using data classification, you can aggregate data under classifications, e.g. by mapping SKUs to products, brands to categories, and countries to sales regions. In Marketing Cloud Intelligence, this is done by uploading a csw file containing classifications and corresponding the values mapped to them.
When to Use It
When you have access to a limited number of fields from a Data Stream and want to aggregate granular data to a level that is more appropriate for your business.
Where to Find It
Connect & Mix → Harmonization Center → Data Classification
Patterns and Harmonized Dimensions
Extract patterns from a shared naming convention and map the individual parts to dimensions in the data model.
What It Does
Patterns are a powerful tool to enrich data from channels like social media and paid search, where you cannot add information on fields like you can in Salesforce. Instead, you add the information to the campaign name itself as strings of text separated by a delimiter like a hyphen or underscore. Marketing Cloud Intelligence extracts the texts and maps them to harmonized dimensions for use in dashboards, analyses, and workflows.
When to Use It
When you are unable to capture data into separate fields, such as in the case of media campaigns, and you need to append that information to the campaign name itself.
Where to Find It
Connect & Mix → Harmonization Center → Patterns
Media Name Builder
Name your cross-channel campaigns following a shared naming convention directly from Marketing Cloud Intelligence.
What It Does
Use the inbuilt Media Name Builder tool to build media names for multi-channel campaigns based on a shared naming convention. Marketing Cloud Intelligence is then able to extract that information via patterns and harmonized dimensions.
When to Use It
Often, companies use Google Sheets or Excel to manage campaign names. These are fine in the short term but cause issues in terms of versioning and admin over time. With the Media Name Builder, you can follow a naming convention from a centralized location, without concern over outdated naming patterns or user access issues.
Where to Find It
Installation: Marketplace → (Search for: “Media Name Builder”) → Get it now
Usage: Marketplace → Apps → Installed Solutions → Media Name Builder
Fix & Maintain
Fix classified and harmonized dimensions with Fix & Maintain. It only fixes the output, not the source data.
What it Does
Fix & Maintain ensures that your classifications and naming patterns work as intended. It does this in two ways. First, it displays the overall success rate of data governance. This means the percentage of data rows that successfully map to a classification or pattern. Naturally, this percentage only includes data rows within entities that are supposed to follow a classification or pattern. Having a high success rate means that the data is uniform and conforms well to the classification or pattern. Conversely, having inconsistent campaign naming or dimensions used for classifications will yield a lower success rate.
Second, Fix & Maintain allows you to export data that does not adhere to the classification or pattern. Exporting will generate an Excel file that you can edit on your desktop as needed. Any non-adhering values will be shown in cells as “Not Valid”, so you can easily filter them and edit them en masse. Once done, you can import the edited file through the same interface, which fixes the data on the Marketing Cloud Intelligence side.
When to Use It
Fix & Maintain isn’t perhaps a tool to use regularly. Instead, an admin should use it for bi-weekly or monthly checkups to ensure that data quality is constantly improving. I would especially keep an eye on it when doing alterations to the data models or the classifications and patterns themselves. Identifying possible issues early will save you time and mitigate the risk of providing false insights to Marketing Cloud Intelligence users.
Where to find it
Connect & Mix → Harmonization Center → Fix & Maintain
Connecting Data
Custom Classification (aka Cross-Channel Entities)
Map and share entities across different data streams.
What It Does
To understand custom classifications, you must first understand how entities work. Think of entities as you would data objects in Salesforce. An entity describes a type of data, such as “Ads”, “Media Buy” or “CRM Leads”. Each entity contains a set of measurements (e.g. impressions and clicks) and dimensions (e.g. dates, names, regions). Each data stream can map fields to one or more entities. For instance, an e-commerce data stream can map data to “Product”, “Channel” and “Geography” entities.
Let’s assume that you want to classify all your marketing analytics under an overarching attribute, like “Brand” or “Region”. Instead of altering the source data or setting up data classifications for each data stream separately, we can use custom classifications. It allows you to filter and view data across all channels based on a shared attribute. For instance, you’d be able to look at organic social, web analytics, and CRM data for the EMEA region with a single page-level filter.
When to Use It
Custom classifications are powerful but come with strict limitations. Firstly, a data source needs to be mapped to a custom classification and have the classifying information on the main entity level; e.g. a social ad campaign must have the “Brand” or “Region” mapped in the data stream. Second, each mapped data source must have a many-to-one relationship with the custom classification. This means that multi-brand or cross-region campaigns would not be supported with custom classifications.
Where to find it
Connect & Mix → Data Streams → Map to Model → (Add a Custom Classification from the data model and give it a descriptive name)
Vlookup
Map together individual data fields with or without a shared key.
What It Does
Any Excel users will be familiar with the VLookup function. In Marketing Cloud Intelligence, Vlookup functions similarly, albeit with a couple of important caveats. Vlookup is a data stream formula that you use when mapping fields between a data stream and your data model (i.e. mapping incoming data to fields within Marketing Cloud Intelligence). The Vlookup formula consists of the following parts:
- ValueToSearch: What value are you looking for? This can be a data stream (csv) field, data model field (dat), or text string.
- searchDimension: Which dimension are you searching the value from? You can add one or several fields, e.g. csv [‘Field_1’]+ csv [‘Field_2’].
- returnDimension: Which value do you want to get? This must be a single dimension.
Note: The searchDimension and returnDimension must belong to the same entity. Both must be dimensions; measurements cannot be searched or returned via Vlookup.
When to Use It
One typical use case for Vlookup is when you want to classify data based on a data stream (vs. data classifications or custom classifications). In such a case, you classify data dynamically based on a data stream instead of having a static classification file uploaded into Marketing Cloud Intelligence. If the classification is changed on the separate data stream, that would be reflected on the data stream that uses the Vlookup formula.
You can use Vlookup within the current data stream as well (vs. at the workspace level). This is useful if you want to look for a relationship between two different dimensions that don’t share the same unique id or key.
An example would be that you can have Vlookup look for a match based on name, SKU, brand or other shared attribute. This is a handy tool for looking up duplicate rows that may have a different key.
Overall, Vlookup is a valuable tool for mapping individual values between different data streams. It should not be used as a replacement for data fusions or parent-child relationships, since those affect the data model on a more fundamental level.
Where to Find It
Connect & Mix → Data Streams → Map to Model → (Add a new custom attribute) → Add formula
Data Fusion
Merge two data sets using a shared key via an outer join.
What It Does
If you’ve used SQL in the past, you should know your way around an outer join. Data fusion is basically just a fancy name for an outer join in Marketing Cloud Intelligence. As is the case with an SQL outer join, data fusion combines two different datasets with a shared key, resulting in a collection of fused dimensions. The original datasets are now affected by the same page-level filters, making constructing dashboards much easier and more end-user-friendly.
When to Use It
Data fusion comes in handy when you want to relate two datasets that wouldn’t normally be harmonized due to differing data types or hierarchies. Let’s say that you want to fuse CRM campaigns and opportunities together so that they can be filtered by the same page-level filters. In this situation, data fusion is the way to go.
You would simply use source campaign id from the opportunity as a basis for the fusion. With the fused dimensions, you would be able to build pivot tables and widgets displaying opportunity and campaign data on the same level, on the same rows. This is a great way to enrich datasets and build more holistic analyses within Marketing Cloud Intelligence.
Another typical use case for data fusion is when you want to apply multiple classifications to the same data. Remember when we talked about custom classifications that support only one-to-one relationships with a particular data source? Well, data fusion is your way around this limitation. For this, you first need to add multiple classification columns on your classification file or upload multiple classification files. Then, you create data fusions based on these, so that you can use either classification as a basis for filtering data.
Note: Since data fusions are transitive, they automatically apply data classifications downstream between the fused datasets. This means you only have to apply data classifications for one of the datasets, and it will be reflected on the other one as well.
Where to Find It
Connect & Mix → Dimensions → Data Fusion
Parent-Child Relationship
Form a hierarchy between data streams based on a shared key.
What it Does
Those familiar with Salesforce CRM object relationships will be happy to know that the Marketing Cloud Intelligence parent-child relationship functions essentially the same way. With it, you can share dimensional values between two datasets that have a shared key and use parent-child relationships to enrich and data granularity to your data, while maintaining a clear hierarchical structure of the data model.
When to Use It
You could have a multi-channel campaign as a parent with individual channel campaigns as children. The parent-child relationship operates through a one-to-many relationship, meaning that a parent can have several child datasets but a child can have only one parent. This works well with linear data but not if you need to map child datasets to several different parents. However, there is a fantastic workaround called the “empty parent hack” by Decision Foundry that you can use to overcome this limitation.
Where to Find It
Connect & Mix → Data Streams → Advanced → Parent Child
Summary
I hope this article provided some insight into how your business can effectively utilize Marketing Cloud Intelligence to optimize your data and ultimately drive better outcomes!
By understanding the various data harmonization methods available and selecting the right one for specific use cases, businesses can save time and resources while unlocking the full potential of their data.