“Take control over your marketing data.” If you are a marketer today, you’ve heard that advice many times over. Your peers, the consultants you hire, market research firms, vendors, your boss — just about everyone thinks it’s common wisdom. What is also equally common are the failed marketing data warehousing projects littered throughout the industry.
As incredulous as it sounds, daily consolidated reporting on all the media spend across Facebook, Google, TV, Catalog and other channels remain an unsolved challenge.
Many modern brands have technology natives running marketing. Building massive databases, data pipelines, and advanced data processing systems are very much second nature to them. They are masters at managing their core operational data, like customer order history, surfacing it to make it simple to order the same thing again or personalizing the customer experience to recommend similar items.
All that data is central to their core experience and managing it well is mandatory for the brand’s operations. On the other hand, a data stack that pertains to where and how they advertise their products is not central to a brand’s operations and hence does not get managed with the same level of diligence.
However, each marketing channel is a unique snowflake and nuances of its supply chain all end up shaping how the data stack looks — and more importantly how it should be quality controlled.
In short — marketing data is a different beast, and a beast it certainly is.
In the marketing analytics world, “source of truth” (SOT) is a loaded phrase. It connotes that the data being consumed is of the highest quality and can be treated as the truth like accounting quality data that CFOs rely on. Typically, systems that are closest to the data are the “source of truth.” For example, Google Analytics would be the SOT for web analytics reporting. Shopify might be SOT for ecommerce orders reporting. Facebook reporting would be SOT for Facebook spend, reach and performance metrics. Catalog mail merge reporting would be SOT for catalog circulation and so on.
Now imagine bringing all these data sets together and being the cross-channel SOT for a brand.
Now, imagine the CEO, CFO, CMO, marketing and analytics stakeholders are all logging in, to use this consolidated data asset for all their daily operational purposes.
Let’s let that sink in. If you’ve been around the space for a bit, you should feel a lump in your throat about now.
Data quality is the ravine between “data in a database” and a data-driven marketing organization. Given how fragmented each data source is, figuring out how to properly collect, transform, persist and quality control each snowflake data source is a mammoth endeavor.
Here are some things you’ll run into along the way:
APIs, including Google Ads, Facebook, and other massive systems, are not perfect. The exact same API data requests might have worked for many months, and would start failing “randomly.” If the proper instrumentation is not in place, it’ll just be holes in the data in your database and you’ll never know until you QA it. There are plenty of other use cases, including API throttling, API breaking changes, changing specs and other legitimate behaviors that can lead to “breakage” in your data. All these challenges multiply with data sources that are not API-driven.
To state the obvious, all data are not equal. Orders and revenue reporting data from Shopify have different dynamics from Facebook spend and clicks reporting or Facebook reach reporting or hit logs from the site. Each data source is meant to serve a specific purpose and hence needs to be collected, warehoused and quality controlled in very specific ways.
Just landing data into a database table without applying critical thinking around how the data is going to be consumed usually renders the data useless. Also, defining proper quality control rules that are very specific to each source requires an intimate understanding of the data and what it can be reasonably compared with.
There are typically hundreds of campaigns across channels, thousands of ad sets, keywords and line items, and hundreds of thousands of creative iterations across channels. It’s usually a mountain of data, and failing to properly classify the data into meaningful groups aligned to decision-making also renders all this data useless.
There are plenty more angles like different attribution windows, different definitions of conversion and revenue within Facebook vs Google Ads and other issues that affect data quality.
Bringing all this data together requires a well-planned cross-channel marketing data framework that clearly specifies where each of these data sets fit in and how to bring them together to meaningfully inform marketing investment and operations decision-making.
So, where does this leave marketers? Is this a lost cause? Is the juice worth the squeeze?
Forward-thinking early adopters are seeing transformational benefits from consolidated marketing data. It feeds a test-learn-grow marketing practice, streamlines cross-channel planning, feeds channel accountability and feeds daily media mix reporting. Brands are realizing all the benefits of the data-driven marketing promise.
Unused marketing data is a giant untapped asset within B2C brands. We don’t just let the sunshine and wind blow, we put it to work. Let’s rally together and put that marketing data to work.
Madan Bharadwaj is co-founder and CTO of Measured.
The post How marketers can take control of their data stack appeared first on ClickZ.Reblogged 1 year ago from www.clickz.com